--- Day changed Sat Mar 29 2008 | ||
fenn | appears to be a blank image? http://heybryan.org/mediawiki/upload/graphviz/Graphviz---digraph G.png | 00:00 |
---|---|---|
kanzure | really? I see it. | 00:00 |
kanzure | the URL loads for me | 00:01 |
kanzure | http://heybryan.org/mediawiki/upload/graphviz/Graphviz---digraph%20G.png | 00:01 |
fenn | weird. works in konqueror but not firefox | 00:02 |
kanzure | Kickass. I made +4 and +3 on Slashdot today. | 00:03 |
kanzure | http://slashdot.org/comments.pl?sid=503546&cid=22901672 | 00:03 |
fenn | works on firefox 1.5 but not 3.mumble alpha .. oh well | 00:05 |
kanzure | jackpot | 00:41 |
kanzure | http://www.google.com/search?client=opera&rls=en&q=computational+materials+science&sourceid=opera&ie=utf-8&oe=utf-8 | 00:41 |
kanzure | fenn: Computational Materials Science: From AB Initio to Monte Carlo Methods | 00:47 |
kanzure | http://www.amazon.com/Computational-Materials-Science-Springer-Solid-State/dp/3540639616 | 00:48 |
kanzure | What do you think? | 00:48 |
fenn | getting from theory to practice and back again is the hard part | 00:48 |
kanzure | Google Books has the first chapter. | 00:48 |
kanzure | Monte Carlo methods are **good** | 00:48 |
kanzure | because if we can use those then we're set | 00:48 |
kanzure | instead of running massive GAs | 00:48 |
fenn | that's like random sampling to find a global optimum? | 00:49 |
kanzure | yes | 00:50 |
kanzure | do you know the inverse DNA/RNA folding problem? | 00:50 |
kanzure | well, mostly inverse protein problem | 00:50 |
fenn | er.. which one is inverse? | 00:50 |
kanzure | given a structure that you want, find the amino acid (and thus nucleotide) sequence that will generate it | 00:50 |
fenn | ok | 00:50 |
kanzure | same thing here | 00:50 |
kanzure | except with other materials | 00:50 |
kanzure | and not necessarily "folding" | 00:50 |
kanzure | obviously it's a more about the functionality | 00:51 |
kanzure | which I don't know how to express exactly | 00:51 |
kanzure | the functionality is not at the molecular level necessarily | 00:51 |
kanzure | but rather a macroscale phenomenon | 00:51 |
kanzure | so that you can get those dependency loops working | 00:51 |
fenn | and.. this is different how? (noting that the protein problem is hard) | 00:51 |
kanzure | do you see the difference between protein structure and material fabricational-dependency? | 00:52 |
fenn | yes, they are so wildly different i'm having a hard time seeing the similarities | 00:52 |
fenn | basically oyu're saying, 'i need a material with such and such properties, find me the atomic structure' | 00:53 |
kanzure | I wonder if I am suggesting that the code might need to be (physically) functional just like in DNA | 00:53 |
kanzure | sort of, yes | 00:53 |
kanzure | I suppose what I am expecting to find is some sort of classification of the materials that we need in the dependency-loop | 00:53 |
fenn | but the way i see it, we dont have that kind of bulk nanoscale production technology yet | 00:53 |
kanzure | i.e., "for black box #5 you need an amorphous gel with an electrical conductivity of 0.5%" | 00:53 |
kanzure | but somehow this all requires human creativity | 00:54 |
kanzure | argh | 00:54 |
fenn | computers are good at optimization | 00:54 |
fenn | you need to provide a starting point | 00:55 |
fenn | otherwise the search space is too large | 00:55 |
kanzure | ah, that's right | 00:56 |
kanzure | so how about this | 00:56 |
kanzure | let's identify the materials that material scientists have characterized well enough that simulations with molecular dynamics can be done | 00:56 |
kanzure | I bet it happens to be clay/sand just like we've been using | 00:56 |
kanzure | as well as silicon and germanium | 00:56 |
kanzure | lots of money behind those elements | 00:56 |
fenn | yep | 00:56 |
kanzure | I need to go find some software packages. | 00:56 |
fenn | prepare to be underwhelmed | 00:57 |
kanzure | http://www.ices.utexas.edu/ccm/itamit/ Institute for the Theory of Advanced Materials in Information Technology. Heh. Maybe I'll go have lunch with a professor and ask for some STMs. | 00:59 |
kanzure | http://www.mcc.uiuc.edu/cgi-bin/software/software.pl various software packages | 01:00 |
fenn | you know it's going to be clunky when they refer to it as 'a computer code' | 01:01 |
kanzure | no kidding | 01:01 |
kanzure | well | 01:01 |
kanzure | you know where that comes from, right? | 01:01 |
fenn | no | 01:01 |
* fenn guesses los alamos | 01:01 | |
kanzure | it's from back in the 1960s when NASA and the military were doing different military codes (as in, paperwork) for protocols in computational simulations | 01:01 |
kanzure | yep | 01:01 |
kanzure | or something like that | 01:01 |
kanzure | so each one had a code, like GAG1201 | 01:02 |
kanzure | and so they called them codes. | 01:02 |
kanzure | I sure hope that this isn't where we get people saying "GIVE ME TEH CODES!!11one" | 01:02 |
kanzure | http://abcnews.go.com/Health/Longevity/Story?id=4520397&page=1 "Live to be 150 .. can you do it? Tuesday at 10 ET on ABC News." <-- I know the guy being interviewed. Know as in, through the internet. | 01:06 |
fenn | aubrey de grey? | 01:06 |
kanzure | no, | 01:07 |
kanzure | Tripper McCarthy | 01:07 |
kanzure | but I have talked with AdG before | 01:07 |
fenn | i was listening to arthur clarke's last interview, on ieee spectrum | 01:09 |
kanzure | any good? | 01:09 |
fenn | sounded really terrible, like he was dying on tape | 01:09 |
kanzure | we need a new Big Three | 01:09 |
fenn | and the guy was asking all these questions that i'm sure he'd answered a million times | 01:09 |
kanzure | what a terrible way to go out | 01:09 |
kanzure | I am expecting papers like "Class-23 type materials can always be used to produce 45-a-j-k polymers". | 01:16 |
epitron | hmmm | 01:19 |
epitron | kanzure: do you think you could make a self-replicating fiber optic network cable that tunnelled under the ground? | 01:19 |
kanzure | where would it get the materials? | 01:20 |
epitron | dirt | 01:20 |
epitron | :D | 01:20 |
kanzure | possibly, what about the energy/heat needed to make the fiber optics? | 01:20 |
epitron | (i dunno what it would need, but i figure optical fibers are semi-easy to make using heat and sand) | 01:20 |
epitron | hmmm | 01:20 |
kanzure | optical fibers are pulled as crystal-glass last I checked | 01:20 |
kanzure | so that's some serious heat for such a small thing :) | 01:20 |
epitron | how about if it grew leaves | 01:20 |
epitron | and collected solar energy | 01:21 |
kanzure | well, since solar cells can barely heat houses for the winter, I dunno :) | 01:21 |
epitron | or maybe it could run on fossil fuels... as it grew, it could extend a thin tube | 01:21 |
fenn | the whole point of fiber optics is that you can go long distances without signal degradation, because the glass is so amazingly pure | 01:21 |
epitron | and one end of the cable would have a gascan attached to it :) | 01:21 |
kanzure | epitron: What about replicating wireless nodes? | 01:21 |
epitron | and it would grow until it ran out of gas | 01:21 |
fenn | that purity comes from chemistry, in particular chlorosilane decomposition | 01:21 |
epitron | wireless nodes are a piece of cake -- there are factories everywhere making those | 01:21 |
epitron | the problem is the long distance connections | 01:22 |
kanzure | replication is exponential, you can mesh the planet in 33 days | 01:22 |
epitron | there are vast tracts of space with no habitation, and hence no wireless nodes | 01:22 |
fenn | you can send a shortwave broadcast around the planet | 01:22 |
epitron | hmm | 01:22 |
epitron | shortwave is unreliable | 01:22 |
epitron | planetwide meshing is a neat idea | 01:22 |
fenn | yes, so are fiber optic cables | 01:22 |
* fenn has quit - connection reset by backhoe | 01:22 | |
epitron | yeah, that could work | 01:22 |
epitron | pfft :) | 01:22 |
epitron | you only have that problem in cities | 01:23 |
epitron | the network cable plant could instinctively avoid highways | 01:23 |
epitron | but i like this self-replicating wifi mesh idea | 01:23 |
kanzure | floating/flying weather balloon wireless hubs | 01:24 |
fenn | even better if they are flying sensor nodes | 01:24 |
kanzure | hahah | 01:24 |
epitron | the thing that's neat about the underground method however is that it's secret | 01:24 |
kanzure | just what I said :) | 01:24 |
epitron | nobody knows it's there | 01:24 |
fenn | i typed it first.. my stupid wifi is slow | 01:24 |
kanzure | suuure ;) | 01:24 |
kanzure | where's my patent application form | 01:24 |
kanzure | quick! | 01:24 |
kanzure | nobody would know it's in the air either | 01:24 |
fenn | where's my automated patent application agent | 01:24 |
epitron | wouldn't there be millions of little balloons? :) | 01:25 |
kanzure | maybe in /dev/random | 01:25 |
fenn | quick! i must patent automatic patenting, and the patenting of meta patent strategies | 01:25 |
epitron | i patented patenting | 01:25 |
epitron | ur all screwed | 01:25 |
fenn | all ur patents r belong to US | 01:25 |
* epitron nods | 01:26 | |
kanzure | we need a woman in here | 01:26 |
epitron | or an official subsidiary | 01:26 |
epitron | hahah | 01:26 |
kanzure | that too | 01:26 |
epitron | true dat | 01:26 |
* fenn nominates epitron | 01:26 | |
* epitron purses her lips | 01:26 | |
fenn | how bout sarahemm | 01:26 |
kanzure | the tranny? | 01:26 |
fenn | she's transhumanist i think :P | 01:26 |
kanzure | she's ambiguous, that's what he is | 01:26 |
epitron | same diff | 01:26 |
epitron | i didn't find her that sharp either | 01:27 |
kanzure | but if she has the time, sure | 01:27 |
epitron | anyhow... | 01:27 |
epitron | hold on i'll get a girl in here | 01:27 |
epitron | i'll see how long she lasts | 01:27 |
kanzure | speaking of getting people in here, how about Superkuh? | 01:27 |
kanzure | He doesn't talk much. | 01:27 |
fenn | this is why i was thinking sarahemm http://www.sarahemm.net/bookview.php | 01:27 |
kanzure | I remember she was doing portable computers | 01:28 |
kanzure | wearables, I mean | 01:28 |
fenn | is superkuh still obsessing about flu pandemics? | 01:29 |
kanzure | is that what he was last doing? | 01:29 |
fenn | last i talked to him | 01:30 |
kanzure | how long? | 01:30 |
fenn | uh, a year? | 01:30 |
-!- mech0r [n=DION102@cpe-74-73-121-30.nyc.res.rr.com] has joined #hplusroadmap | 01:30 | |
kanzure | hm, the last I talked with him, we were talking some heavy-duty neuroscience | 01:30 |
mech0r | ARE YOU SERIOUS | 01:30 |
epitron | there we go | 01:30 |
epitron | meet mech0r | 01:30 |
epitron | she is a robot | 01:30 |
mech0r | ARE YOU SERIOUS? | 01:30 |
epitron | o_O | 01:31 |
fenn | _-_ | 01:31 |
mech0r | O_o | 01:31 |
kanzure | R. U. Sirius has contacted me and would like me to write a few articles for his newsletter. | 01:31 |
kanzure | http://en.wikipedia.org/wiki/R._U._Sirius | 01:31 |
* mech0r parts | 01:31 | |
fenn | is he related to the australian guy? | 01:31 |
kanzure | "The" australian guy? | 01:31 |
fenn | yahoo | 01:31 |
fenn | yahoo sirius | 01:31 |
kanzure | not sure | 01:31 |
epitron | The Revolution Party | 01:31 |
epitron | kickass | 01:31 |
kanzure | it's closely related to the Pirate Party, in concept | 01:31 |
kanzure | except not grassroots-tech | 01:32 |
kanzure | just grassroots-extropic | 01:32 |
epitron | R.U. Sirius sounds like he has a fun life | 01:33 |
mech0r | please sum that article up in 1 word | 01:34 |
fenn | irrelevant | 01:35 |
mech0r | ok awesome i'm good to go | 01:35 |
mech0r | so ahem* | 01:36 |
mech0r | pardon me | 01:36 |
mech0r | where is everybody from | 01:36 |
kanzure | Austin, Texas. | 01:36 |
kanzure | mech0r: you? | 01:36 |
fenn | bloomington IN | 01:36 |
mech0r | <--- NYC | 01:36 |
kanzure | So, what did epitron tell you to get you in here? | 01:37 |
fenn | kanzure: you know it's funny the similarities between bloomington and austin | 01:37 |
mech0r | he said i'd get pics of bois | 01:37 |
kanzure | fenn: Indiana has more university smarts last time I checked. | 01:37 |
epitron | |:) | 01:37 |
epitron | | :) | 01:37 |
epitron | |:) | 01:37 |
epitron | | :) | 01:37 |
kanzure | mech0r: Sorry, no boys around here. Only men. | 01:37 |
epitron | |:) | 01:37 |
fenn | nah just philosophy crap | 01:37 |
mech0r | i don't believe it | 01:37 |
epitron | do you demand PROOF? | 01:37 |
epitron | send her pics boys! | 01:37 |
mech0r | i do | 01:37 |
kanzure | I'm sure I have a link somewhere. | 01:38 |
mech0r | also he promised me a bag of sleptons. | 01:39 |
fenn | i mean, they separated engineering into its own school, thus killing the vibrancy of it (purdue) and turning the other school (bloomington) into a big wank-fest | 01:39 |
kanzure | yikes | 01:39 |
* epitron hands mech0r a bag of decayed sleptons and runs away | 01:39 | |
kanzure | Purdue and Wisconsin have always impressed me | 01:39 |
mech0r | WHAT | 01:39 |
kanzure | but I hear that Purdue just *sounds* nice | 01:39 |
fenn | its really depressing to be there | 01:39 |
kanzure | and that they've gone downhill in recent | 01:39 |
* mech0r shoves the decayed sleptons up epi's butt | 01:39 | |
kanzure | fenn: I did a bookmark count to determine where I should apply to. By far, MIT won. But Wisconsin was way the hell up there. They do some serious engineering and serious physics -- no bullshitting with them. | 01:40 |
mech0r | Madison? | 01:40 |
kanzure | yes | 01:40 |
epitron | i get a girl to join | 01:40 |
epitron | and you guys start talking about what universities you want to go to? | 01:41 |
mech0r | i know someone who went to madison | 01:41 |
epitron | what is wrong with you :) | 01:41 |
kanzure | epitron: I'm not into group cyberbanging. | 01:41 |
epitron | hahah | 01:41 |
fenn | hmm.. i'm thinking about MIT also, but dunno if i really want to go down the academic path | 01:41 |
mech0r | others went to carnegie n rpi | 01:41 |
kanzure | fenn: Just good people to know. | 01:41 |
mech0r | cuz the didn't feel like working hard at mit | 01:41 |
fenn | right | 01:41 |
kanzure | undergrad MIT is supposedly hell | 01:41 |
epitron | oh right, MIT is Waterloo South | 01:41 |
fenn | their motto is "IHTFP" i hate this fucking place | 01:41 |
mech0r | it's like why even bother | 01:41 |
kanzure | awesome | 01:41 |
epitron | mech0r: yeah, it's the slavery abyss :) | 01:42 |
epitron | you do not want to drift too far into the machine | 01:42 |
epitron | it burns you out | 01:42 |
mech0r | kant in my chan is in comp/elec engineering | 01:42 |
epitron | uses you up | 01:42 |
mech0r | and he didn't have all too much fun at it | 01:42 |
mech0r | ^_^ \/ | 01:42 |
kanzure | Hrm. | 01:42 |
epitron | we are not machines... yet. | 01:42 |
kanzure | but on the other hand you have lots of smart kids and an energetic environment | 01:42 |
mech0r | oh but i am.... | 01:42 |
kanzure | so that's the advantage | 01:42 |
mech0r | energetic? | 01:43 |
epitron | well, true | 01:43 |
epitron | there is an art to living | 01:43 |
mech0r | lullz beer and math | 01:43 |
epitron | if you can find a balance at a crazy ass school like that, then you'll be set for life :) | 01:43 |
kanzure | epitron: How about finding us a serious chick? | 01:43 |
fenn | kanzure: not gonna happen | 01:43 |
epitron | dude | 01:43 |
epitron | mech0r rocks | 01:43 |
kanzure | :( | 01:43 |
mech0r | serious about kicking ass | 01:43 |
mech0r | hi | 01:43 |
fenn | kanzure: every single time i find a girl who is technically minded, she turns out to be a transsexual | 01:43 |
-!- kanzure changed the topic of #hplusroadmap to: happy bunny day - day of the replicator || http://heybryan.org/ http://chris.ill-logic.com/ http://fennetic.net/ | 01:43 | |
* mech0r hides bulge | 01:43 | |
kanzure | ah crap | 01:44 |
epitron | yee | 01:44 |
epitron | don't put that in there | 01:44 |
epitron | people will find my shit! | 01:44 |
kanzure | fenn: That's ... unfortunate. | 01:44 |
epitron | plz remove webpage plz | 01:44 |
fenn | i dont think so really | 01:44 |
mech0r | you remove it | 01:44 |
kanzure | epitron: seriously? | 01:44 |
epitron | yes :) | 01:44 |
-!- kanzure changed the topic of #hplusroadmap to: happy bunny day - day of the replicator || http://heybryan.org/ http://fennetic.net/ | 01:44 | |
epitron | i give out my site to those who i want to see it | 01:44 |
epitron | so i can watch the traffic :) | 01:45 |
fenn | kanzure: there's so few technically minded people anyway.. | 01:45 |
epitron | if too many people have it it starts to get spammed | 01:45 |
epitron | then i can't read the logs | 01:45 |
kanzure | fenn: I figure there should be a way to induce technical competence | 01:45 |
mech0r | haha geez epi | 01:45 |
fenn | it has to come from them | 01:45 |
fenn | you cant force someone to be genuinely interested | 01:45 |
kanzure | but then why did I start off as a moron? | 01:45 |
fenn | uh, because you had nothing pushing you in the right direction | 01:46 |
epitron | also because his brain was a blank slate | 01:46 |
fenn | kanzure: you mean like, five years ago or whatever, you cared about power rangers, right? | 01:47 |
kanzure | fenn: more like eight years ago, when I first started watching television. My thing was pokemon, not power rangers. | 01:47 |
fenn | sorry | 01:47 |
fenn | pokemon is a sort of autistic thing in itself though | 01:47 |
kanzure | yes | 01:47 |
kanzure | I've traced back a lot of ideas back to that show, weirdly enough | 01:47 |
fenn | sure | 01:47 |
kanzure | "meme trainers" etc. | 01:48 |
kanzure | hunting down ideas | 01:48 |
kanzure | and then a very big ego | 01:48 |
fenn | when i was a kid, there was no pokemon, so i made up my own alien ecology | 01:48 |
kanzure | yeah? how extensive? | 01:48 |
fenn | i dont know, i think i made something and then forgot and then re-did it | 01:48 |
fenn | it was based in part on the computer game 'starflight' | 01:49 |
fenn | er, for the sega genesis | 01:49 |
kanzure | I try to run background searches from time to time across the pokemon forums, looking for kids that might be me, so that I can catch them early. | 01:49 |
kanzure | http://heybryan.org/docs/pokemon_forums.txt | 01:50 |
fenn | i think it was a rather shallow ecology, more like the african savannah than anything | 01:50 |
fenn | but it was a dense atmosphere planet so all kinds of weird flying crap | 01:51 |
fenn | bubble trees | 01:51 |
epitron | bubble tREES! | 01:51 |
fenn | none of this is digital though | 01:51 |
mech0r | so hmm | 01:51 |
kanzure | Well, if you want my internet history, there's random crap still around | 01:51 |
mech0r | how about them brain implants | 01:52 |
kanzure | http://web.archive.org/web/*/animeu.hey.nu/ | 01:52 |
kanzure | mech0r: yeah? | 01:52 |
kanzure | http://heybryan.org/mediawiki/index.php/Brain_implants | 01:52 |
kanzure | What about? | 01:52 |
fenn | brain implants are stupid | 01:52 |
epitron | oh yeah, mech0r, bryan has a crazy huge wiki | 01:52 |
mech0r | i've noticed | 01:52 |
fenn | by the time they get it implanted, there will be something better | 01:52 |
kanzure | One of the problems with brain implants that I focus on is that you can't predict what changes you can make | 01:52 |
kanzure | fenn: so the idea is to not implant a chip per-se | 01:52 |
kanzure | fenn: use viral gene therapy on a lab on a chip to modify your brain | 01:52 |
kanzure | chances are, viral gene therapy can't easily be updated anyway | 01:52 |
fenn | its like modems and satellite and cable and DSL | 01:53 |
kanzure | I mean, what sort of advances are we going to get there? | 01:53 |
fenn | in 5 years everyone will have 256Mbit DSL | 01:53 |
kanzure | I wish I would have started doing a wiki earlier. I only started this in February. | 01:53 |
fenn | then who knows, maybe optical mesh networks or something | 01:53 |
fenn | viral gene therapy should only be used to install a system that works reliably | 01:54 |
kanzure | correct | 01:54 |
fenn | like, wget apt.tgz | 01:54 |
kanzure | that's why we need self-replication for lots and lots of testing on neural slices | 01:54 |
kanzure | http://heybryan.org/mediawiki/index.php/Neurofarms | 01:54 |
fenn | more like tftp, not wget :) | 01:55 |
fenn | i never figured out a good way of doing a checksum on a genome | 01:55 |
kanzure | I don't know if that would be useful. | 01:56 |
fenn | you could kill anything that didnt match | 01:56 |
fenn | then you're guaranteed to be all cooperative cells | 01:56 |
fenn | or, instead of killing them, try infecting again ("transfecting" technically) | 01:57 |
kanzure | alright, good idea | 01:57 |
fenn | blah.. biology terminology is screwed | 01:58 |
kanzure | no kidding | 01:59 |
kanzure | bio-ontologies might work out in the end, but | 01:59 |
epitron | damn latin! | 01:59 |
fenn | most genetic engineering techniques are not very subtle, and you wouldnt want to put them in your body | 02:00 |
kanzure | but I haven't explored that scene sufficiently | 02:00 |
kanzure | they even have SBML - systems biology markup lang | 02:00 |
kanzure | right | 02:00 |
fenn | but in order to use more subtle gene expression, you have to ensure a high level of reliability in the vector | 02:00 |
kanzure | ideally we could engineer viruses on the spot with new nucleotide sequences | 02:01 |
kanzure | plus a kill mechanism, plus a maximum life expectancy | 02:01 |
fenn | the problem is that terms get entrenched before they know what is actually going on | 02:01 |
kanzure | and then let them infect the local population, then kill 'em | 02:01 |
kanzure | yes, that's true | 02:01 |
* kanzure changes off the jpop (this is supposed to be anime-only ... kawaii is letting me down) | 02:01 | |
fenn | see this is where the optical dna write-o-some comes in | 02:01 |
kanzure | fenn: Drew Endy took interest in your idea by the way. :) | 02:01 |
* fenn googlestalks | 02:02 | |
kanzure | http://heybryan.org/mediawiki/index.php/Drew_Endy should have something | 02:02 |
kanzure | ah here we go | 02:02 |
kanzure | http://heybryan.org/mediawiki/index.php/Synbio | 02:02 |
kanzure | that has a video interview with Drew | 02:03 |
fenn | alright who write wikipedia entries about assistant professors? | 02:03 |
kanzure | haha | 02:03 |
mech0r | did someone say anime. | 02:03 |
kanzure | Drew Endy did 23CCCC or whatever :) | 02:03 |
kanzure | mech0r: Yeah. | 02:03 |
kanzure | http://mp3.brascasses.eu/Hardcore%20dudur/Hardcore/Autres/Maximum%20The%20Hormone/Death%20note%20op2-%20What's%20up,%20people!-%20Maximum%20the%20hormone.mp3 | 02:04 |
mech0r | you watched death note? | 02:04 |
fenn | i have some manga phone-books with death note, but i didnt really get it | 02:05 |
kanzure | mech0r: yes. | 02:05 |
kanzure | fenn: his dx is not enough to kill the number of people he wants to | 02:05 |
kanzure | mathematically infeasible | 02:05 |
mech0r | i gave up watching it after they killed off l | 02:05 |
kanzure | I havent' gotten that far, you bitch. | 02:06 |
kanzure | hehe | 02:06 |
mech0r | o | 02:06 |
kanzure | joking, joking | 02:06 |
mech0r | k | 02:06 |
mech0r | was wonderin... | 02:06 |
kanzure | yeah? | 02:06 |
mech0r | it's been out for a while heh | 02:06 |
kanzure | Sorry, I've only been working on the whole overpopulation problem ;) | 02:06 |
kanzure | I'll try working harder | 02:06 |
kanzure | bwahah | 02:06 |
mech0r | yes | 02:06 |
mech0r | work harder | 02:06 |
mech0r | be more asian | 02:07 |
fenn | http://parts.mit.edu/wiki/index.php/Abstraction_hierarchy_and_PoPS | 02:07 |
kanzure | yep | 02:07 |
fenn | i'm still processing that, but it looks familiar | 02:07 |
kanzure | fenn: Ellington says that's all bullshit | 02:07 |
kanzure | and yes, it's familiar because I probably showed it to you already | 02:07 |
kanzure | remember my DNA compiler idea ? evolvable in vitro DNA logic? etc. ? | 02:08 |
fenn | no, i mean its like black-boxing | 02:08 |
kanzure | yes | 02:08 |
kanzure | but there's emergent effects =) | 02:08 |
kanzure | and they don't know how to account for it | 02:08 |
kanzure | join the mailing lists | 02:08 |
kanzure | faceface in #bioinformatics has a list of links to mailing lists | 02:08 |
fenn | uh, isnt the whole point of winfree's stuff that it's relatively deterministic and abstraction-friendly? | 02:08 |
kanzure | I think it's http://biodatabase.org/ and then just search for User:Kanzure | 02:08 |
fenn | the dna logic circuits | 02:08 |
kanzure | yes, but Winfree doesn't do biobricks | 02:08 |
kanzure | biobricks is more than logic, it's components and devices that make up new proteins and whatever | 02:09 |
kanzure | actually I'm not sure if it really is more | 02:09 |
fenn | so, it's bullshit because you cant control/predict the interactions between proteins? | 02:10 |
kanzure | it's also bullshit because there's no real way of getting data in and out | 02:10 |
kanzure | PoPS is mostly useless, you just have analog data reporting there | 02:10 |
fenn | well, that's the same problem with ellington's stuff | 02:10 |
kanzure | so obviously Ellington is more focused on the transcriptional toe-hold method for logic gates | 02:10 |
fenn | how do you get the data out? reverse transcriptase? | 02:11 |
kanzure | he hasn't told me | 02:11 |
kanzure | I think his point is that you can't | 02:11 |
fenn | harrumph | 02:11 |
epitron | drew endy is a great teacher | 02:11 |
kanzure | that's just the comp sci attitude | 02:12 |
kanzure | if you watched the vid on the wiki | 02:12 |
epitron | what's the compsci attitude, his teaching ability? | 02:12 |
kanzure | yeah, that's obvious abstraction and black-boxing | 02:12 |
epitron | (yeah, it was from the wiki) | 02:12 |
fenn | it's standard engineering | 02:12 |
kanzure | :) | 02:12 |
fenn | (literally) | 02:12 |
epitron | no, i just mean... he's good at top-down explainations and switching modes :) | 02:12 |
kanzure | he's had lots of practice I bet. | 02:12 |
epitron | what he's tlaking about is interesting too | 02:13 |
epitron | but not shicking | 02:13 |
epitron | -i+o | 02:13 |
kanzure | good, because that's what http://biohack.sf.net/ is about | 02:13 |
epitron | the 2nd video on your page has been removed btw | 02:13 |
fenn | my poor 433MHz laptop doesnt like multiple embedded videos | 02:14 |
kanzure | epitron: Google has it, I am sure of it. | 02:14 |
epitron | ok | 02:15 |
epitron | you want me to fix your wiki? :) | 02:15 |
epitron | oh, embedding disabled by request | 02:15 |
kanzure | ? | 02:15 |
kanzure | ah | 02:16 |
epitron | fix0red | 02:16 |
fenn | uh... asteroid mining to make tissue culture experiments? that's a little out there | 02:16 |
fenn | re: neurofarms | 02:16 |
epitron | OMG this guy has the best name | 02:17 |
epitron | NORRIS HUNG | 02:17 |
mech0r | that's my cousin | 02:17 |
epitron | lies! | 02:17 |
epitron | you're just trying to be cool | 02:17 |
kanzure | fenn: yes, it's out there, and I don't need it to be out there | 02:17 |
mech0r | i don't have to try | 02:17 |
* epitron snaps | 02:18 | |
kanzure | fenn: but how else are we going to figure out what changes we are going to make? | 02:18 |
* mech0r breaks epitron in half | 02:18 | |
* epitron crackles | 02:18 | |
kanzure | I am pretty sure NASA's ISS is already doing tissue experiments | 02:18 |
fenn | kanzure: computer modeling | 02:18 |
fenn | they're interested in how biology adapts to zero-g | 02:18 |
kanzure | fenn: yes, but they prove that it can be done as well | 02:18 |
fenn | kanzure: if you have infinite computer processing power, is that enough? | 02:19 |
kanzure | I am not really sure it's a computational problem though | 02:19 |
fenn | or, infinite computer power + one tissue culture dish | 02:19 |
kanzure | how can you predict the emergent physics? | 02:19 |
fenn | with fine grained resolution :) | 02:19 |
kanzure | and at the moment it takes a very large supercomputer to simulate a simple dish | 02:19 |
fenn | meh | 02:20 |
kanzure | I have no problems with simulations | 02:20 |
kanzure | I'm all for it. | 02:20 |
fenn | atm it takes hundreds of millions just to send up an asteroid probe | 02:20 |
kanzure | millions of what? | 02:20 |
fenn | dollars | 02:20 |
kanzure | yeah .... | 02:20 |
kanzure | that's because they are clueless | 02:20 |
kanzure | I mean, they do not have engineers on their team doing everything | 02:20 |
kanzure | they all believe it's hard to do or something | 02:20 |
kanzure | meanwhile we have guys like John Carmack doing it with a team of five | 02:21 |
fenn | still, supercomputer access is more egalitarian than space access | 02:21 |
kanzure | http://en.wikipedia.org/wiki/John_Carmack | 02:21 |
kanzure | indeed | 02:21 |
fenn | bah | 02:21 |
fenn | doing something.. | 02:21 |
epitron | john carmack is a team of five | 02:21 |
epitron | :) | 02:21 |
fenn | sure, most of the new space companies are small | 02:21 |
kanzure | I get the feeling that John doesn't like me much, though. | 02:21 |
epitron | haha | 02:22 |
fenn | he gets pestered by millions of people every day, dont worry about it | 02:22 |
epitron | what'd he say to you? | 02:22 |
kanzure | (and I don't specifically bug him) | 02:22 |
kanzure | epitron: uh, I don't have the email in front of me, but he basically said that I was bullshitting around | 02:22 |
kanzure | and he was right, but he didn't offer a good approach for me to adopt | 02:22 |
epitron | what was he referring to? | 02:22 |
kanzure | http://heybryan.org/mediawiki/index.php/Scramjets | 02:23 |
fenn | heh | 02:23 |
kanzure | oops | 02:23 |
epitron | 404 | 02:23 |
kanzure | http://heybryan.org/mediawiki/index.php/Scramjet | 02:23 |
kanzure | try that | 02:23 |
kanzure | (btw, a scramjet is not a good idea, I've subsequently learned) | 02:23 |
epitron | do you want an irc search command? :) | 02:23 |
fenn | well i gotta agree, scramjet is not the best place to start | 02:23 |
fenn | its like talking about how to build tokamak reactors | 02:24 |
kanzure | mech0r: What's your ... interest? | 02:24 |
kanzure | yeah | 02:24 |
epitron | anyhow, don't sweat it. it's just carmack. and he interfaces with so many people that he'll forget about you in a month ;) | 02:24 |
kanzure | but not only that, a scramjet generally can't go from ground to orbit | 02:24 |
fenn | heh it cant go from ground to air | 02:24 |
kanzure | I was thinking of scramjet + liquid engine on board | 02:24 |
fenn | you have to drop it from a b-52 just to test | 02:24 |
mech0r | i have many interests | 02:24 |
kanzure | fenn: right, but why not just use jet propulsion to get to that height in the first place | 02:25 |
kanzure | mech0r: indeed? | 02:25 |
fenn | because then your mass ratio sucks | 02:25 |
fenn | SSTO is a Hard Problem | 02:25 |
mech0r | ask epi | 02:25 |
kanzure | I googlestalked the guy who did the whole X-43A project, found his AOL id. | 02:25 |
mech0r | he's more awake | 02:25 |
kanzure | his woman posted to some geneaology forums back in 2006 | 02:25 |
fenn | *drool* | 02:26 |
fenn | kanzure: if you want an 'actionable' ground to orbit system, check out jordin kare's modular laser launch http://www.niac.usra.edu/files/studies/final_report/897Kare.pdf | 02:31 |
kanzure | I've heard about this before, but this paper looks nice | 02:33 |
kanzure | I'm still interested in chemical propellants. I must be old school. | 02:33 |
fenn | yeah it's been done to death, and it's gone as far as it can | 02:33 |
kanzure | not true | 02:33 |
kanzure | the fabrication costs | 02:33 |
kanzure | DIY LOX manufacturing | 02:33 |
kanzure | That'll make things sufficiently different, I think. | 02:33 |
fenn | LOX is cheap | 02:33 |
kanzure | no it's not | 02:34 |
kanzure | $500,000 for enough to get into LEO | 02:34 |
fenn | compared to the cost of a rocket, it's negligible | 02:34 |
kanzure | on an average payload | 02:34 |
kanzure | whatever that means | 02:34 |
kanzure | and what's the rocket, just mostly metal, some electronics, | 02:34 |
fenn | yeah | 02:34 |
kanzure | I bet it's mostly the man hours that are costing really | 02:34 |
kanzure | don't pull the costing BS on me | 02:34 |
kanzure | John was complaining about that earlier today ;) | 02:34 |
fenn | you're right, it is the man hours mostly | 02:34 |
fenn | there's a whole planet full of dirt and water and air | 02:34 |
kanzure | "but this will all cost billions of dollars!" and John just laughs and walks away from the funding-source ;) | 02:34 |
fenn | but you know, there really is a lot of engineering that goes into those rockets | 02:35 |
kanzure | simulations, testing, re-testing, etc., sure | 02:35 |
fenn | because that's what the market demands, a high reliability vehicle | 02:35 |
kanzure | also, they don't design new rockets | 02:36 |
kanzure | they are doing reuse of old designs | 02:36 |
kanzure | so some guy was smart enough to make the first one | 02:36 |
kanzure | and then they said "Hey, let's not risk it" | 02:36 |
kanzure | they mean "risk money" more than anything | 02:36 |
kanzure | and "risk man-hours" | 02:36 |
fenn | well, i dont know what to do about that | 02:36 |
fenn | did you ever look at thermoacoustics? | 02:37 |
kanzure | Do it ourselves. | 02:37 |
kanzure | not yet | 02:37 |
kanzure | plan to :) | 02:37 |
-!- Andares [n=andares@unaffiliated/jacco] has joined #hplusroadmap | 02:37 | |
kanzure | Hey Andares. | 02:37 |
kanzure | This is my little hangout ;) | 02:37 |
Andares | Hey guys/kanzure. | 02:37 |
Andares | Cool. | 02:37 |
Andares | What's hplus? | 02:37 |
fenn | basically you absorb the thermal energy with piezo's, and that cools the gas off until it liquifies | 02:37 |
kanzure | Andares: Transhumanist, sort of. | 02:37 |
Andares | Cool. | 02:37 |
kanzure | http://heybryan.org/mediawiki/index.php/Roadmap for hplusroadmap | 02:37 |
fenn | bs++ | 02:37 |
kanzure | fenn: bs++ where? | 02:37 |
kanzure | thermoacoustics? | 02:37 |
Andares | Oh, makes sense. Human+. | 02:37 |
fenn | all around us | 02:37 |
kanzure | huh? | 02:38 |
Andares | kanzure, so, do you think we can escape entropy? | 02:38 |
fenn | we are living in a sea of bs | 02:38 |
kanzure | Andares: I am _not_ a negentropist. | 02:38 |
Andares | What's negentropist? | 02:38 |
kanzure | http://www.orionsarm.com/polities/Negentropism.html "Entropy is the flaw of the universe" -- Negentropism. | 02:38 |
Andares | It doesn't bother you that by definition we must die? | 02:39 |
kanzure | "Ethical philosophy of the Negentropy Alliance; formulated by the AI-cluster known as the Judge. Based on the view is that entropy is the fundamental flaw of the universe, and ethical actions are those that serve to slow or stop the increase of entropy." | 02:39 |
kanzure | Andares: All things must die, even a being that lives millions of years. | 02:39 |
Andares | Perhaps. | 02:39 |
kanzure | ;) | 02:39 |
Andares | Though. | 02:39 |
Andares | Seems like we could find a way to alter the universe to keep entropy from taking over. | 02:40 |
kanzure | I am the opposite of a negentropist: entropy is not a flaw per-se, and I should maximize entropy (do as much as possible). | 02:40 |
fenn | uh, what if you upload yourself into a non-temporal continuum | 02:40 |
kanzure | ENTROPY IS NOT DISORDER | 02:40 |
kanzure | fenn: that's assuming mind-uploading | 02:40 |
Andares | It.. | 02:40 |
Andares | fenn, what would you upload to? | 02:40 |
fenn | kanzure: well, an AI could do it too | 02:40 |
kanzure | when the hell will we collectively stop capitalizing ai | 02:41 |
Andares | Anyway, to break the current subject: | 02:41 |
kanzure | we've been using the phrase for 50 years now | 02:41 |
fenn | or do you identify as this particular arrangement of atoms (which is continuously being flushed and renewed) | 02:41 |
Andares | I just rewatched Primer for the 4th time. Excellent movie. | 02:41 |
kanzure | aren't we tired of typing it? | 02:41 |
kanzure | Primer? | 02:41 |
fenn | DIY time travel | 02:41 |
Andares | And had a great discussion with my wanna-be-girlfriend's sister's boyfriend on DCM extraction of caffeine. :) | 02:41 |
kanzure | fenn: Me? I have shady ideas on identity ... don't think it's a useful concept really. | 02:42 |
Andares | kanzure, an excellent independent film. It's about time travel, but it's not cheesy. | 02:42 |
Andares | They can only go back to the time where they started the machine. | 02:42 |
Andares | And it stays close to believable physics. | 02:42 |
fenn | kanzure: got an alterntaive to "AI" for non-uploaded sentience? | 02:43 |
kanzure | fenn: possibly replacement scenarios and genome rewriting scenarios, but these might be called uploading | 02:43 |
Andares | Why not just call it "I?" It's not like implementation matters all that much. | 02:43 |
kanzure | My real alternative is actually programming | 02:44 |
fenn | kanzure: no, i mean, a smart computer program that appears for all intents and purposes to be alive | 02:44 |
kanzure | where I would make something new -- possibly a new child -- and try to tweak the kid to the extent that I would like to | 02:44 |
kanzure | fenn: I do not have an alternative, no. | 02:44 |
kanzure | Most of my thoughts on 'intelligence' is wet. | 02:44 |
fenn | ok, i'll continue calling it AI | 02:44 |
kanzure | oh | 02:44 |
Andares | Oh, speaking of AI. | 02:44 |
kanzure | you mean a term | 02:44 |
fenn | yeah | 02:45 |
Andares | My neural network is so far kind of a disappointment. | 02:45 |
Andares | It sometimes overflows, and takes FOREVER. | 02:45 |
kanzure | fenn: non-uploaded sentience; how about in silico intelligence? | 02:45 |
kanzure | erm | 02:45 |
kanzure | that's ambiguous | 02:45 |
Andares | It took it around 2 days to go through 20k epochs with ~1000 training samples. | 02:45 |
kanzure | Andares: http://heybryan.org/mediawiki/index.php/Compuational_biology | 02:45 |
fenn | it's always going to be ambiguous - what's your preferred level of terseness? | 02:45 |
fenn | me, i'm lazy, and i'm not victorian | 02:46 |
kanzure | http://minduploading.org/ \o/ | 02:46 |
Andares | Brain uploading. | 02:46 |
kanzure | I need to go recurse through the comp-matt-sci software packages and then hit bed. | 02:46 |
Andares | Doesn't seem feasible without totally cloning all the neuron weights. | 02:47 |
kanzure | Andares: Uh. | 02:47 |
kanzure | Andares: Seriously, go see computational biology. | 02:47 |
Andares | So many links. :( | 02:47 |
kanzure | There's much more to neural networks since the weighted ANN models in the 1960s | 02:47 |
kanzure | that's what I used to think computational neuroscience was | 02:47 |
kanzure | but there's been lots of advancements | 02:47 |
kanzure | some serious shit going on | 02:47 |
kanzure | ion channel modeling, for one | 02:47 |
kanzure | http://ionchannels.org/ | 02:47 |
Andares | Ion channels. | 02:47 |
* fenn yawns | 02:47 | |
Andares | reallyyy. | 02:47 |
Andares | That's neat. | 02:47 |
fenn | simulated worm excrement | 02:48 |
kanzure | and then neuropeptide modeling, diffusion models, electricity, piezoelectric modeling of neurons (Superkuh's ideas) | 02:48 |
Andares | But imho, a little to accurate. | 02:48 |
fenn | did you know pooping releases dopamines? | 02:48 |
kanzure | fenn: heh | 02:48 |
kanzure | fenn: is that why I get so many ideas when | 02:48 |
kanzure | well | 02:48 |
kanzure | nevermind. | 02:48 |
kanzure | My Throne. | 02:48 |
Andares | fenn, makes sense. | 02:48 |
fenn | Andares: i just made it up | 02:49 |
Andares | But it does make sense that some neurotransmitter is released. | 02:49 |
kanzure | fenn: it makes sense | 02:49 |
kanzure | heh' | 02:49 |
Andares | And DA is often used for reinforcement. | 02:49 |
fenn | sure, holding your breath does all kinds of wild stuff | 02:49 |
fenn | how are you going to model that? | 02:49 |
kanzure | that's rather large scale | 02:49 |
kanzure | I guess we can start with respiratory channel interactions in the lungs | 02:49 |
kanzure | and then we can move up to the brain oxygen intake via fMRI studies | 02:50 |
fenn | and before long you have a map the size of the universe | 02:50 |
Andares | Ooh let's talk about 5HT2A hallucinations! | 02:50 |
kanzure | let's not | 02:50 |
Andares | :( | 02:50 |
kanzure | I need to sleep. | 02:50 |
Andares | You and your nasty sleeping habit. | 02:50 |
Andares | I had a bit too much caffeine and thus won't sleep for some hours (currently 2:50AM) | 02:50 |
fenn | he's a prisoner, regimented schedule and all you know | 02:50 |
Andares | Ah, right. | 02:51 |
Andares | fenn, so do you much AI? | 02:51 |
kanzure | fenn: thanks for understanding, =) | 02:51 |
fenn | Andares: no | 02:51 |
kanzure | I mean, damn. | 02:51 |
fenn | buck up, it's almost over | 02:51 |
fenn | then you get to do it voluntarily | 02:51 |
kanzure | I should have ran away. | 02:51 |
fenn | yeah, you should have taken a train tour of china | 02:52 |
Andares | fenn, do what? | 02:52 |
fenn | Andares: intellectual slavery | 02:52 |
Andares | Oh. | 02:52 |
Andares | I am a slave. :D | 02:52 |
Andares | Hm, I wonder if I can manage to get the university to loan me some computer time. | 02:53 |
fenn | for your simulated blob of goo? | 02:53 |
Andares | Totallyu. | 02:53 |
Andares | *-u | 02:53 |
Andares | Well, I have kind of a cool idea for it. | 02:53 |
fenn | why is nobody doing hardware implementations? | 02:53 |
Andares | But it's probably been done. | 02:53 |
Andares | fenn, bloody hard to do. | 02:53 |
Andares | But a few people are doing them. | 02:53 |
fenn | parallel computations go so much faster on dedicated hardware | 02:53 |
Andares | Especially parallel hardware. :) | 02:54 |
fenn | right | 02:54 |
fenn | like a reconfigurable fpga | 02:54 |
Andares | I was wondering if you could use semiconducting polymers or something. | 02:54 |
Andares | Though | 02:54 |
Andares | do those exist? | 02:54 |
fenn | uh, not really | 02:54 |
Andares | They'd have to carry electrons somehow, and to do that they couldn't really be. | 02:54 |
Andares | k, pen and paper time. | 02:55 |
fenn | there are conductive polymers, and organic semiconductors, but in general they arent production-level | 02:55 |
Andares | Totally liquid computers would be cool. | 02:56 |
fenn | how liquid is liquid enough? | 02:56 |
Andares | Well, some kind of system where you can keep adding goo in. | 02:56 |
fenn | what's your minimum particle size | 02:56 |
Andares | *yawn* dunno. | 02:57 |
fenn | you can make a cluster of 0.5 meter particles already | 02:57 |
Andares | 0.5 meter? | 02:57 |
Andares | That's visible. :| | 02:57 |
fenn | ya | 02:57 |
Andares | And huge. | 02:57 |
fenn | pizza box form factor | 02:57 |
fenn | then there's microcontrollers and zigbee stuff | 02:58 |
fenn | then rfid tags, although they cant really do much processing | 02:58 |
Andares | True. | 02:58 |
Andares | Wonder if you could make logic-gate type things. | 02:59 |
Andares | Except as artificial neurons. | 02:59 |
fenn | point is, it's here already, you just arent looking | 02:59 |
Andares | And then wire those. | 02:59 |
fenn | that's what an FPGA is | 03:00 |
Andares | fenn, check Googel. | 03:00 |
Andares | 03:00 | |
fenn | say what? | 03:01 |
Andares | They just changed it. | 03:01 |
Andares | fenn, what if you had a hardware neural network. | 03:02 |
fenn | i wish they would say something meaningful | 03:02 |
Andares | Then a device on top that would somehow modify the weights. | 03:02 |
fenn | like, 'buy less crap' | 03:02 |
Andares | fenn, so FPGAs can be modified via software? | 03:04 |
fenn | or 'support your local alternative energy researcher' | 03:04 |
fenn | Andares: yes, but in practice they can only be configured all at once | 03:04 |
Andares | How does that work? | 03:04 |
fenn | you load a stream of configuration data after resetting the device | 03:05 |
fenn | beyond that, it's a big mystery | 03:05 |
Andares | fenn, has anyone ever implemented a neural net on an fpga? | 03:05 |
fenn | generally, there's a memory cell that controls what type of gate it is, and what bus it's connected to | 03:06 |
fenn | and what neighbors its connected to | 03:06 |
fenn | it's not as highly interconnected as a brain though | 03:06 |
Andares | How efficient is it compared to a hardwired chip? | 03:06 |
fenn | i dont really follow the neural net scene | 03:06 |
fenn | compared to an ASIC they are rather slow | 03:07 |
fenn | each cycle is like a nanosecond or so | 03:07 |
Andares | hehe, faster than the brain | 03:07 |
Andares | *. | 03:07 |
fenn | but it's parallel so that's bloody fast compared to a processor | 03:08 |
fenn | you can run processors in an FPGA layout, they go up to like 500MHz | 03:08 |
fenn | or something like that | 03:08 |
Andares | Argh. A single FPGA is $3k. | 03:09 |
Andares | (from Altera.) | 03:09 |
fenn | no, they range crfrom $10 to $250k | 03:09 |
fenn | in multiple qty the price goes down to like $3 | 03:10 |
Andares | Oh. | 03:10 |
Andares | Wait, $3k you mean. | 03:10 |
fenn | no, three dollars | 03:10 |
Andares | for an FPGA? | 03:10 |
Andares | Regular CPU cost more than that, what. | 03:11 |
fenn | well i didnt say it was a huge powerful fpga | 03:11 |
fenn | http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=122-1218-ND | 03:12 |
fenn | ^^ example | 03:12 |
fenn | 15000 gates 5ns delay time | 03:13 |
Andares | ah | 03:13 |
fenn | heres a slightly newer chip http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=122-1479-ND | 03:14 |
Andares | Ah. | 03:14 |
fenn | so, 100k gates isn't the same as 100k neurons | 03:14 |
fenn | but it's enough to do a lot of signal processing | 03:15 |
Andares | Not to mention that neurons are analog-ish. | 03:15 |
Andares | What kind of logic gates are on that thing? | 03:15 |
fenn | well, fpga's can be analog too | 03:15 |
Andares | just or/and/xor? | 03:15 |
fenn | i read some article where they used a GA to program an FPGA, and it did all kinds of analog stuff they didnt understand | 03:15 |
Andares | Neat. | 03:16 |
Andares | I bet I could model a virtual FPGA without too much work. | 03:16 |
Andares | But! | 03:16 |
fenn | eh, well, the difference between theory and practice is much bigger in practice | 03:16 |
Andares | My physical neural network requires time to repair its connections and store memory and stuff. | 03:17 |
Andares | Night fenn, nice to meet you. | 03:17 |
fenn | night | 03:18 |
-!- fenn [n=pz@adsl-75-62-112-183.dsl.bltnin.sbcglobal.net] has joined #hplusroadmap | 07:36 | |
-!- Topic for #hplusroadmap: happy bunny day - day of the replicator || http://heybryan.org/ http://fennetic.net/ | 07:36 | |
-!- Topic set by kanzure [] [Sat Mar 29 01:44:46 2008] | 07:36 | |
[Users #hplusroadmap] | 07:36 | |
[ Andares] [ epitron] [ fenn] [ kanzure] [ mech0r] | 07:36 | |
-!- Irssi: #hplusroadmap: Total of 5 nicks [0 ops, 0 halfops, 0 voices, 5 normal] | 07:36 | |
-!- Channel #hplusroadmap created Sat Mar 22 15:44:12 2008 | 07:36 | |
-!- Irssi: Join to #hplusroadmap was synced in 19 secs | 07:36 | |
-!- Andares [n=andares@unaffiliated/jacco] has quit [Connection timed out] | 10:29 | |
-!- andares [n=andares@unaffiliated/jacco] has joined #hplusroadmap | 10:55 | |
-!- fenn_ [n=pz@adsl-99-133-188-114.dsl.bltnin.sbcglobal.net] has joined #hplusroadmap | 11:05 | |
-!- Topic for #hplusroadmap: happy bunny day - day of the replicator || http://heybryan.org/ http://fennetic.net/ | 11:05 | |
-!- Topic set by kanzure [] [Sat Mar 29 01:44:46 2008] | 11:05 | |
[Users #hplusroadmap] | 11:05 | |
[ andares] [ epitron] [ fenn] [ fenn_] [ kanzure] [ mech0r] | 11:05 | |
-!- Irssi: #hplusroadmap: Total of 6 nicks [0 ops, 0 halfops, 0 voices, 6 normal] | 11:05 | |
-!- Channel #hplusroadmap created Sat Mar 22 15:44:12 2008 | 11:05 | |
-!- Irssi: Join to #hplusroadmap was synced in 36 secs | 11:05 | |
-!- fenn [n=pz@adsl-75-62-112-183.dsl.bltnin.sbcglobal.net] has quit [Read error: 110 (Connection timed out)] | 11:21 | |
-!- fenn [n=pz@adsl-99-133-189-140.dsl.bltnin.sbcglobal.net] has joined #hplusroadmap | 16:31 | |
-!- Topic for #hplusroadmap: happy bunny day - day of the replicator || http://heybryan.org/ http://fennetic.net/ | 16:31 | |
-!- Topic set by kanzure [] [Sat Mar 29 01:44:46 2008] | 16:31 | |
[Users #hplusroadmap] | 16:31 | |
[ andares] [ epitron] [ fenn] [ kanzure] [ mech0r] | 16:31 | |
-!- Irssi: #hplusroadmap: Total of 5 nicks [0 ops, 0 halfops, 0 voices, 5 normal] | 16:31 | |
-!- Channel #hplusroadmap created Sat Mar 22 15:44:12 2008 | 16:31 | |
epitron | kanzure: well, you can also run into the problem where your information can't be put into a single ontology... for example, if it belongs in two categories.. your only option is to create a new category that's the combination of those two categories. | 16:32 |
fenn | tags | 16:32 |
epitron | i mean, ontologies are just data compression | 16:32 |
-!- Irssi: Join to #hplusroadmap was synced in 33 secs | 16:32 | |
fenn | you're all it! | 16:32 |
epitron | there are some kinds of information that just can't be compressed easily :) | 16:33 |
epitron | like a computer program | 16:33 |
epitron | it's all freakin' cross linked up the wazoo | 16:33 |
epitron | the only way around is to be able to view it from different angles | 16:33 |
kanzure | Hey fenn. | 16:33 |
kanzure | mech0r: yes | 16:33 |
epitron | where each angle is a fitlered view that can have its own ontology | 16:33 |
kanzure | mech0r: http://heybryan.org/ has a link to my page | 16:33 |
epitron | i think that's where neils bohr's famous quote came from | 16:34 |
epitron | "The opposite of a correct statement is an incorrect statement. The opposite of a profound truth is another profound truth." | 16:34 |
kanzure | fenn: I've been recursing through molecular dynamics and comp-matt-sci-sims and found http://srdata.nist.gov/cccbdb/ a few minutes ago. Can't figure out where in my ontology to put this. Chem db or something. | 16:34 |
epitron | the two "profound truths" are just different filtered aspects of the same high-dimensional thing | 16:35 |
epitron | a thing that we cannot completely hold in our puny brains | 16:35 |
epitron | :) | 16:35 |
epitron | (if we could, we wouldn't need to compress the information into ontologies) | 16:35 |
fenn | epitron: what's more compressed, something written in assembly or something written in lisp? | 16:36 |
epitron | lisp, bigtime | 16:36 |
epitron | unless you create an assembly lisp metalanguage | 16:37 |
epitron | :) | 16:37 |
epitron | then i guess they're equal | 16:37 |
kanzure | yet lisp has more overhead | 16:37 |
kanzure | so I guess you mean that's compression when transferring | 16:37 |
fenn | assuming they are both 'perfect' | 16:37 |
epitron | and i suppose you could also create a more entropic language in lisp if you're a massochist | 16:37 |
kanzure | not in the expression of the program | 16:37 |
kanzure | epitron: or discordian | 16:37 |
epitron | :) | 16:37 |
epitron | i'm just talking about the compression of the information contents of the programs | 16:38 |
epitron | not the resulting machine code | 16:38 |
epitron | (information contents = source code) | 16:38 |
epitron | (and datafiles i suppose) | 16:38 |
kanzure | http://www.emsl.pnl.gov/forms/basisform.html I don't know what this is. | 16:38 |
epitron | oh man | 16:39 |
mech0r | i'm looking at the semiconductor manufacturing haha | 16:39 |
fenn | i'm guessing basis vectors for their parameter space | 16:39 |
epitron | i'm asking for a gaussian basis set for my birthday | 16:39 |
fenn | but how do you 'exchange' them | 16:40 |
kanzure | mech0r: yeah? | 16:40 |
kanzure | mech0r: http://heybryan.org/semiconductor.html | 16:40 |
kanzure | mech0r: I've collected some information on that subject at that ^ link. | 16:40 |
fenn | A basis set in chemistry is a set of functions used to create the molecular orbitals, | 16:40 |
kanzure | Oh? | 16:40 |
mech0r | yeah i'm looking at thagt | 16:40 |
fenn | http://en.wikipedia.org/wiki/Basis_set_(chemistry) | 16:41 |
epitron | the Silicon (Br)/otherhood | 16:41 |
epitron | haha | 16:41 |
fenn | sounds like epicycles :\ | 16:41 |
kanzure | Hm, I wonder what sort of functions these are | 16:41 |
epitron | \o/ | 16:41 |
kanzure | are these Shroeddinger? | 16:41 |
kanzure | I remember reading a bit about molecular orbital theory | 16:42 |
kanzure | but I also remember that VSEPR is the better stuff | 16:42 |
kanzure | plus density functional theory | 16:42 |
kanzure | Valence Shell Electron Pair Repulsion theory | 16:42 |
fenn | vsepr is like stuff they teach to babies | 16:42 |
fenn | with balloons and straws | 16:43 |
kanzure | ah, Hartfree-Fock == molecular orbital approximations | 16:43 |
kanzure | oh | 16:43 |
kanzure | http://en.wikipedia.org/wiki/Hartree-Fock | 16:43 |
kanzure | to-read: | 16:44 |
kanzure | http://en.wikipedia.org/wiki/Ab_initio_quantum_chemistry_methods | 16:44 |
fenn | as usual, wikipedia falls on its face when it comes to math | 16:44 |
kanzure | good stuff - http://en.wikipedia.org/wiki/Quantum_chemistry_computer_programs | 16:45 |
kanzure | yes | 16:45 |
fenn | math notation is like powerpoint | 16:45 |
kanzure | not many articles are good at explaining the connections between different ideas | 16:45 |
kanzure | hah | 16:45 |
kanzure | yes, it's just a placeholder really | 16:46 |
fenn | or like diagrams with no captions | 16:46 |
kanzure | and they should be explaining it more throoughly | 16:46 |
epitron | fenn: you ever seen scholarpedia? | 16:46 |
kanzure | me has | 16:46 |
epitron | i figured you hvae | 16:46 |
epitron | you've seen everything :) | 16:47 |
fenn | epitron: yeah it's bullshit | 16:47 |
epitron | hahah | 16:47 |
fenn | what's the point | 16:47 |
epitron | why is it bullshit? | 16:47 |
kanzure | it has some good stuff in a few places, but I've used it maybe once, ever | 16:47 |
fenn | they already have academic journals and reviews and compendiums | 16:47 |
fenn | the point of wikipedia is lowering the barrier to entry | 16:47 |
epitron | hmmm | 16:47 |
kanzure | fenn: I am impressed with Wikipedia on computational chemistry, actually | 16:47 |
epitron | but how many academic journals define the basics of a concept? | 16:47 |
kanzure | this looks comprehensive | 16:47 |
epitron | and wikipedia doesn't really encourage experts to update it | 16:48 |
fenn | yes they do | 16:48 |
epitron | i was talking to a professor who's the world expert in a subject, and he didn't want to touch the wikipedia page on it because it was so huge and bloated and wrong | 16:48 |
kanzure | Oh shit. There goes Opera. 423 tabs. | 16:48 |
fenn | unless you mean, pay them money | 16:48 |
epitron | refactoring text is like refactoring code.. it sucks :) | 16:48 |
epitron | hahah | 16:49 |
epitron | kanzure: impressive | 16:49 |
fenn | well, if he doesn't care enough to correct the primary source of information on his subject, that's his mistake | 16:49 |
epitron | kanzure: have you ever seen http://www.zotero.org/ ? | 16:49 |
kanzure | you bastard | 16:49 |
kanzure | giving me a link while Opera is down | 16:49 |
epitron | fenn: that's not the primary source. it's wikipedia :) | 16:49 |
kanzure | it's a primary source for us digital natives | 16:49 |
fenn | epitron: it's the first source people look at | 16:50 |
kanzure | things are topsie-turny in the internet age | 16:50 |
fenn | hence, primary source | 16:50 |
epitron | fenn: and it's not about not CARING... experts are busy people | 16:50 |
epitron | fenn: it's too much work to fix it | 16:50 |
fenn | pff | 16:50 |
kanzure | apparently too busy to fix fundamental knowledge problems | 16:50 |
fenn | do you know how much work it takes to write a journal article? | 16:50 |
epitron | a lot :) | 16:50 |
fenn | those pages and pages of references | 16:50 |
epitron | but the difference is you get paid usually | 16:50 |
fenn | well, maybe we should pay them to work on wikipedia instead of something nobody will ever see | 16:51 |
epitron | well, yeah, but then you have the problem with people reverting their page | 16:51 |
epitron | wikipedia doesn't really have a good trusted identity system | 16:51 |
kanzure | epitron: Ah. I opened it up in iceweasel/firefox. Yes, I've seen Zotero, I think I tried it. I dislike it. I don't want anything integrated into my browser (unless that's an extension, an extra part, which I'd gladly accept). Also, I don't want anything integrated into Mozilla's framework as its basis, or web services or anything. I want some *serious* programs. :) Thus, AutoScholar. | 16:51 |
kanzure | reverting pages doesn't matter | 16:51 |
kanzure | just keep a copy on your website | 16:51 |
epitron | kanzure: yeah, i hated the firefox-extension part of that.. but it looks nice | 16:51 |
epitron | i like the idea :) | 16:51 |
fenn | the history's still there anyway | 16:52 |
epitron | it saves the pages to your hard drive too | 16:52 |
fenn | wikipedia's a giant revision control system | 16:52 |
epitron | for permanent archival | 16:52 |
kanzure | epitron: So, did I ever explain to you my ideal paper-reading scenario? | 16:52 |
epitron | nope :) | 16:52 |
kanzure | I've been brainstorming for the past few years about a better way to read papers and absorb information. | 16:52 |
kanzure | Now, if Enki-2 was in here, he'd have a few things to say about his xu88 project (he's implementing Project Xanadu in prolog) | 16:52 |
kanzure | but anyway, the idea is that it's computationally complex to retrieve a paper | 16:52 |
fenn | kanzure: get a grad slave to summarize it for you :) | 16:52 |
kanzure | as in, you have to go click through 8 different links to get to a single paper | 16:52 |
kanzure | but ideally you can just autofetch with AutoScholar as an apt-get system for knowledge | 16:53 |
epitron | fenn: man, wikipedia's history for each page is usually huge... and hard to navigate... i.e. big changes and little changes look identical in the history view :) | 16:53 |
kanzure | and then you can "digest it" with minimized computational complexity | 16:53 |
fenn | epitron: there's a number of lines changed number (number!) | 16:53 |
kanzure | supposedly this digestion process is one of **refactoring** | 16:53 |
kanzure | I do not mind refactoring as long as it's positive progress | 16:53 |
kanzure | Now, most papers are in a PDF, | 16:53 |
epitron | fenn: but when you're scanning 50 pages of history, you want to have to read every single line? | 16:53 |
kanzure | so you have to click back and forth between PDF and an HTML page or something | 16:53 |
epitron | fenn: why not use colour coding or font size or something to represent the size of the change | 16:54 |
epitron | fenn: or colour coding to show reverts | 16:54 |
epitron | or SOMETHING :) | 16:54 |
kanzure | and that destracts from refactoring itself, which sucks, so I've decided I want kolourpaint+kpdf+kwrite all wrapped into one for an editing system to digest content. | 16:54 |
fenn | epitron: ok, maybe you should tell it to #wikipedia | 16:54 |
kanzure | not even that | 16:54 |
kanzure | that's an easy script | 16:54 |
kanzure | Wikipedia gives 130 GB data dumps daily | 16:54 |
kanzure | also see ##wikipeding for group-based collaboration on reading wikipedia | 16:54 |
epitron | hmm | 16:55 |
fenn | what, do diffs on the database dump? are there tools for extracting it back into a reasonable format like text files? | 16:55 |
epitron | so you want a streamlined knowledge-condensation pipeline? | 16:55 |
kanzure | fenn: something like that; epitron mentions other things like trying to figure out how much of a change, so I guess it could be text-based. I would like access to Amazon's "statistically unlikely phrases" tools so that we can do something like that. :) | 16:55 |
kanzure | epitron: Yes. I want to break up the images from the PDF directly and then allow rapid commentary, linking, doodling, connections, etc., not necessarily in any well-defined format, just something that can be (1) presented and (2) quickly made in real time (not necessarily edited later ... heh) | 16:56 |
kanzure | with minimal computational-complexity/clickage/eye-averting-movements | 16:56 |
fenn | kanzure: sounds like you want to get the original latex source? because it's (supposedly) like SGM | 16:56 |
fenn | SGML* | 16:56 |
epitron | have you ever read about Danny Hillis' idea for a next-geneartion emergent knowledge tool? | 16:56 |
epitron | http://www.edge.org/3rd_culture/hillis04/hillis04_index.html | 16:56 |
epitron | oops | 16:57 |
epitron | that's the addendum :) | 16:57 |
epitron | sec | 16:57 |
kanzure | Hrm. That'd be nice. But I don't mean like that. I don't care how the authors represented their data, I want to be able to quickly scratch out my own representations of the same information into various data structures, with a simple keyboard-interface. I.e., hit "j" to come up with a new table, then select image-areas to add into that table, and that's added to your page. | 16:57 |
kanzure | epitron: re: emergent knowledge tool, see Gunkel and ideonomy. David mentioned this guy to me the other day. http://ideonomy.mit.edu/ read the one that says Austin, TX, or just skim over it really | 16:58 |
epitron | oh wait, it's the right page :) | 16:58 |
epitron | nevermind | 16:58 |
epitron | you just have t scroll past the crap | 16:58 |
epitron | look for the word "introduction" :) | 16:58 |
kanzure | I also want to do automated citation/references so that I can download all of the papers related to some new field, and then read those as quickly as I want. | 16:58 |
fenn | gunkel is funny, it's like automated creativity.. a monte carlo method in some weird space | 16:58 |
kanzure | yes | 16:59 |
kanzure | epitron: Hillis seems to believe in information overload. | 16:59 |
kanzure | The problem is not the information; the problem is us. | 17:00 |
epitron | i think you should read this :) | 17:00 |
kanzure | fenn: I was trying to talk with him to see how he figured it out. | 17:00 |
kanzure | epitron: I am | 17:00 |
epitron | you're already judging | 17:00 |
kanzure | fenn: In other words, how did he do all that he did? He made many thousands of documents and so on, how'd he do this? It must have taken him hours of painstaking work. Unless he's an autist. But I emailed him and I looked at the way he writes and talks, chances are that he's not. | 17:00 |
kanzure | Not "extremely autistic", but perhaps partially in the repetition sense, but really. | 17:01 |
fenn | i havent even looked at the page yet, but just from the url "3rd_culture" sets off my wank detector | 17:01 |
kanzure | fenn: it's a good book | 17:01 |
kanzure | fenn: third culture just refers to traveling scientists | 17:01 |
epitron | god you guys are so judgy :) | 17:01 |
mech0r | kick them in the nads. | 17:01 |
epitron | that's why you're here | 17:01 |
fenn | epitron: it's my mechanism of perception, not active judgement | 17:01 |
kanzure | epitron: re: learning tools see http://supermemo.com as well as http://supermemo.com/articles/genius.htm or http://supermemo.com/articles/genius.html | 17:02 |
epitron | fenn: it's laziness is what it is :) | 17:02 |
fenn | yes | 17:02 |
fenn | how else am i going to keep up with all you supernerds | 17:02 |
kanzure | epitron: I'm pretty sure we should get Enki-2 back in here before we talk about Hillis' knowledge web ideas | 17:02 |
epitron | haha | 17:02 |
epitron | ok | 17:02 |
epitron | as long as we can all discuss in an orderly manner | 17:03 |
kanzure | also ... Joram Zutt. He wanted some automated teaching systems. He had some support from the World Bank last I checked. | 17:03 |
epitron | (that's a lot of people talkign at once) | 17:03 |
epitron | oh god | 17:04 |
mech0r | (so what's going on) | 17:04 |
epitron | world-bank funded automatic teaching machine | 17:04 |
epitron | that sounds like brainwashing :) | 17:04 |
fenn | "while jimmy wales' bazaar flourished and prospered, i continued to be stuck in my rut, plugging away at my cathedral..' | 17:04 |
kanzure | epitron: Joram Zutt was doing a self-replicator via atom holography. http://heybryan.org/projects/atoms/ | 17:05 |
kanzure | So he wanted (1) on-board nuclear fusion power source, (2) datalink, (3) computation & teaching, (4) atom holography. | 17:05 |
epitron | hahahaha | 17:05 |
epitron | that's a bit "out there" | 17:06 |
kanzure | yep | 17:06 |
kanzure | obviously I took up his ideas on atom holography | 17:06 |
epitron | i'm glad he's wasting world bank money on it | 17:06 |
kanzure | don't know about nuclear fusion power sources | 17:06 |
kanzure | well, he had support via a professor that recently died of cancer | 17:06 |
kanzure | which was learning-oriented more than anything else | 17:06 |
kanzure | I found Zutt by seeing a weird comment on Slashdot, by The_Laughing_God, and then going off to search for the Laughing God on the internet, only to find Zutt making fun of him on his website, so I clicked around and found his Bazaar Project C. | 17:07 |
fenn | there's a lot of unexplored territory re: fusion reactors | 17:07 |
kanzure | sure | 17:07 |
kanzure | Zutt's interest was in magnetically confined fusion reactors | 17:07 |
kanzure | which seems like an interesting way to do it | 17:07 |
kanzure | you have some massive plasma ball of flame, you use magnets to confine it | 17:07 |
fenn | that's the standard way to do it | 17:07 |
kanzure | ah | 17:07 |
fenn | it doesn't really work | 17:07 |
kanzure | I believe the project was LDX@MIT | 17:08 |
epitron | kanzure: robert bussard was working on that too | 17:08 |
fenn | after a terabuck and 50 years they now know exactly why it wont work | 17:08 |
epitron | for 11 years | 17:08 |
epitron | and made it work :) | 17:08 |
kanzure | yes, then he died | 17:08 |
kanzure | that bastard | 17:08 |
epitron | he died?? | 17:08 |
kanzure | Bussard? yes | 17:08 |
epitron | WTF | 17:08 |
epitron | bastard! | 17:08 |
kanzure | right after he got Navy funding for his ramjet. | 17:08 |
fenn | ramjet? | 17:09 |
epitron | mannn | 17:09 |
kanzure | wait | 17:09 |
epitron | what happened to his research? | 17:09 |
fenn | oh, the interstellar collector thing is totally separate | 17:09 |
epitron | i bet it was stolen and burned | 17:09 |
fenn | epitron: it's being continued by space-x i think | 17:09 |
epitron | did you ever see his google talk on magnetic confinement fusion? | 17:09 |
kanzure | maybe it was his tokamak or polywell | 17:09 |
kanzure | epitron: no, link please? | 17:09 |
fenn | polywell is bussard's baby | 17:09 |
epitron | really?? | 17:09 |
epitron | it was revolutionary | 17:09 |
kanzure | there's a good discussion forum on polywell | 17:09 |
kanzure | http://heybryan.org/forums.html has a link to polywell-forums | 17:09 |
epitron | yeah, it's the polywell | 17:10 |
kanzure | http://www.talk-polywell.org/bb/index.php | 17:10 |
epitron | that's the one he was talking about | 17:10 |
kanzure | here we go | 17:10 |
fenn | nobody knows anything about polywell except bussard's teams.. how can they have a forum on it | 17:10 |
epitron | bussard had like 500000 pages of research papers he'd generated in 11 years working at the navy | 17:10 |
mech0r | oh god... | 17:10 |
mech0r | TOO MANY LINKS | 17:10 |
kanzure | fenn: heh | 17:11 |
epitron | i know :) | 17:11 |
epitron | i wish these guys would summarize instead of linking | 17:11 |
kanzure | mech0r: I have over 12,000 in my bookmarks, and then I browse at least 500/day. | 17:11 |
mech0r | what the hell | 17:11 |
mech0r | go outside and play | 17:11 |
fenn | epitron: are you sure about that? | 17:11 |
epitron | hahaha | 17:11 |
kanzure | because if we don't link then you don't know where it is | 17:11 |
epitron | fenn: about what | 17:11 |
fenn | rather a summary than a link | 17:11 |
kanzure | a link is like a direct-access click to becoming better | 17:11 |
fenn | i can sit here and spout bs all day if you like | 17:11 |
epitron | while 5000000 links is like cache-thrashing :) | 17:11 |
mech0r | i'm not even loggint the chan usebhfsef | 17:11 |
kanzure | ? | 17:12 |
kanzure | usebhfsef? | 17:12 |
mech0r | yes. | 17:12 |
mech0r | exactly | 17:12 |
kanzure | what's it? | 17:12 |
epitron | it's ok everyone else is | 17:12 |
epitron | :) | 17:12 |
kanzure | epitron: there's a reason why academic papers have refs and links ;)O | 17:12 |
mech0r | it is exactly what it looks like | 17:12 |
kanzure | mech0r: ah | 17:12 |
epitron | kanzure: yes but academic papers are usually like, "There is a way to blow up frogs using nucleosynthesis [11]" | 17:12 |
epitron | and then you know what 11 is about without READING IT | 17:12 |
epitron | without disturbing the state of your short term memory | 17:13 |
kanzure | I see. | 17:13 |
kanzure | I guess fenn and I must have superior short term memory functionality then | 17:13 |
epitron | haha | 17:13 |
kanzure | because it's pretty easy for me to be able to go off on tangents and return successfully | 17:13 |
kanzure | that's how I do recursing through the internet | 17:13 |
kanzure | otherwise I'd just lose my way | 17:13 |
epitron | well you need to get over a hump in an area to be able to do that | 17:13 |
fenn | i try to actually read the line before the link, it usually has a summary | 17:13 |
epitron | you're giving me link to shit i've never even heard about before | 17:14 |
epitron | and it's not a summary, it's a 100 page detailed research paper | 17:14 |
fenn | epitron: welcome to the 21st century | 17:14 |
fenn | i suggest googling words you dont know, wikipeding | 17:14 |
epitron | i don't even know which direction i'm pointed when i read it ;) | 17:14 |
kanzure | the idea is to find stuff that you never heard of | 17:14 |
epitron | haha | 17:14 |
kanzure | that's the whole freaking point | 17:14 |
epitron | and i'm not saying there's anything wrong with that | 17:14 |
kanzure | why would you want to find something you already know about | 17:14 |
kanzure | you should know where that is ;) | 17:14 |
epitron | i'm saying that flooding someone with that is annoying | 17:14 |
epitron | SIGH | 17:15 |
fenn | kanzure: it's comforting to most people | 17:15 |
epitron | not something i already know about | 17:15 |
epitron | ok | 17:15 |
fenn | kanzure: you have to remember this is a learned behavior | 17:15 |
kanzure | fenn: what you and I do? | 17:15 |
fenn | whereas you are a 'natural' i guess | 17:15 |
kanzure | this is true | 17:15 |
epitron | when you're reading about NUCLEOBIOCHRONOSYNESTHESIA, you already know a bit about nucleobionetics and cyberchronodynamics, so it's easy to scan | 17:15 |
kanzure | but I don't remember learning this per-se | 17:15 |
kanzure | you know what, maybe I do | 17:15 |
epitron | it's not about ALREADY KNOWING IT | 17:16 |
kanzure | I remember that for many years I was staying in the same places on the internet | 17:16 |
epitron | it's about being familiar with the concept-space | 17:16 |
kanzure | and so I learned to do rapid clicking on certain features | 17:16 |
kanzure | and then I extrapoliated from there when moving to new programs | 17:16 |
fenn | epitron: see that's why i hate biology papers, they aren't made of latin roots. it's all people's stuffed animal names and totally made up words | 17:16 |
kanzure | because it was easy to see what problems were showing up in the "toolchain" | 17:16 |
epitron | fenn: hey you know what's good.. english roots :) | 17:16 |
kanzure | hurray, Opera crashed again | 17:16 |
fenn | epitron: and once you understand a few key concepts, it's all very easy | 17:16 |
epitron | disable plugins/javascript | 17:16 |
epitron | fenn: exactly | 17:17 |
epitron | nevermind | 17:17 |
epitron | i'm basically trying to teach you guys to teach and you don't wanna learn ;) | 17:17 |
kanzure | oh | 17:17 |
kanzure | well | 17:17 |
kanzure | what do you want us to teach? | 17:17 |
fenn | those concepts include: logic, abstraction, wave theory, empiricism | 17:17 |
epitron | kanzure: nevermind, i've already forgotten what brought me here | 17:17 |
kanzure | btw, a good reason for us to not teach is because we don't have a full overview of all of the information yet | 17:18 |
epitron | it was just the general idea of making the conversations more compact and ordered and simple | 17:18 |
* fenn highfives kanzure for successfully overwhelming epitron | 17:18 | |
epitron | instead of BLAH BLAH BLAH *50000 links* | 17:18 |
epitron | more focus | 17:18 |
epitron | pfft :) | 17:18 |
epitron | that's not hard | 17:18 |
epitron | my short term memory sucks | 17:18 |
kanzure | *MORE* focus? | 17:18 |
fenn | you do jump around a lot | 17:18 |
epitron | yeah | 17:19 |
kanzure | that might be true, but there are subtle connections that are worth observing | 17:19 |
kanzure | and actually | 17:19 |
epitron | you're very tangential kanzure :) | 17:19 |
kanzure | I'd prefer it if you guys call me on it when I jump | 17:19 |
epitron | ok | 17:19 |
epitron | i can do that | 17:19 |
epitron | i have a friend who does that too | 17:19 |
mech0r | totally isn't me. | 17:19 |
epitron | when i'm talking to him in person he can't paste me links though :) | 17:19 |
fenn | compchem browsers wiki semantic-web fusion all in the last hour | 17:19 |
kanzure | I am fairly certain that I do not jump; most of my thoughts and actions are very tightly integrated together | 17:19 |
kanzure | fenn: yeah | 17:19 |
kanzure | fenn: but those jumps were due to epitron, not me | 17:20 |
kanzure | I didn't bring up semantic webbing | 17:20 |
kanzure | but I did bring up autoscholar and eating articles re: the computational chem stuff in the first place | 17:20 |
epitron | kanzure: well, it depends on what the goal is. if the goal is a breadth-first communication of all knowledge you contain, then jumping is good. if you want to plan something, or discuss some kind of actionable direction, then jumping is bad :) | 17:20 |
kanzure | and the reason why we brought up computational chem on Wikipedia was because we are looking into the modeling software to narrow the search space | 17:20 |
kanzure | if I recall correctly, epitron was the one with the Wikipedia tangent ;) | 17:21 |
epitron | haha | 17:21 |
epitron | yeah | 17:21 |
epitron | i didn't intend it to be a tangent | 17:21 |
kanzure | right, we were already discussing the actionable item though | 17:21 |
epitron | i just wanted to make a little remark under my breath | 17:21 |
fenn | i'm too stubborn to scroll back and actually look | 17:21 |
epitron | :) | 17:21 |
epitron | but then fenn started questioning it | 17:21 |
epitron | and i had to explain it | 17:21 |
kanzure | fenn and I are figuring that we should go look for some software and narrow the search space for skdb and self-replication | 17:21 |
fenn | that's not what i think at all | 17:22 |
kanzure | which reminds me, I still don't know what particular functional parameters we need to search for / constrain in the first place (it's not the "properties" of materials that material scientists are playing with) | 17:22 |
epitron | what's skdb? | 17:22 |
kanzure | epitron: http://heybryan.org/mediawiki/index.php/Skdb -- a database of material projects | 17:22 |
kanzure | fenn: please explain | 17:22 |
fenn | predictive modeling of material properties is a big waste of time, especially since it totally ignores the economics of making said material | 17:22 |
kanzure | we don't need to *model* but rather generate | 17:22 |
fenn | same thing | 17:22 |
kanzure | so that we can know what materials we need that fit the specs for self-replication | 17:23 |
kanzure | then what's the alternative to knowing what to do | 17:23 |
fenn | you have a model of how atoms interact and how that gives rise to bulk properties | 17:23 |
fenn | er, that's not the alternative ^^ | 17:23 |
fenn | the alternative is gathering the societal engineering knowledge that's already out there | 17:23 |
kanzure | the models are the knowledge | 17:24 |
epitron | hmmm | 17:25 |
epitron | what does "societal engineering" mean? | 17:25 |
fenn | societal = in society at large | 17:25 |
fenn | like, no single person knows all of it | 17:25 |
epitron | distributed ?engineering | 17:25 |
kanzure | epitron: If we asked you to go make an analog circuit to do basic math, could you do it off the top of your head, from first principles? | 17:25 |
epitron | probably not | 17:25 |
kanzure | epitron: Chances are, the answer is no. Most of this knowledge has been "intuited" from many, many experiments that people have gathered all over the world. | 17:25 |
epitron | unless i was given a lot of time :) | 17:25 |
epitron | right | 17:26 |
fenn | even if you had a lot of time, you might get stuck in a rut forever | 17:26 |
epitron | emergent knowledge :D | 17:26 |
kanzure | especially in the case of chemistry, we just can't magically predict that certain materials can fit in our dependency loops (I think) | 17:26 |
* fenn rolls his eyes.. emergence.. fooey | 17:26 | |
epitron | synergetics! | 17:26 |
epitron | :D | 17:26 |
kanzure | yeah, so our idea is to collect the emergent model and apply it to make the design for the self-replicator | 17:26 |
kanzure | *the emergent knowledge | 17:26 |
epitron | ok | 17:27 |
kanzure | oh | 17:27 |
kanzure | fenn: I see | 17:27 |
epitron | so it's like danny hillis' knowledge web kinda | 17:27 |
fenn | wait.. huh? what's "the emergent model"? | 17:27 |
kanzure | fenn: I just forgot about my conclusions on skdb; you're right, we do have to bruteforce it. Like I was saying earlier. I must have forgotten what I was thinking. Sorry. Any room for computational models in the database anyway? | 17:27 |
kanzure | fenn: note I said "emergent knowledge" | 17:27 |
kanzure | fenn: I corrected myself. | 17:27 |
fenn | what's the emergent knowledge | 17:27 |
kanzure | the stuff we are gathering | 17:27 |
epitron | do you know what emergence is? | 17:28 |
fenn | epitron: no, actually | 17:28 |
epitron | haha | 17:28 |
fenn | it seems to be one of those words that's impossible to define | 17:28 |
epitron | so you phooeyd it because you didn't know it? | 17:28 |
kanzure | fenn: it's the societal knowledge, that's what I mean. | 17:28 |
epitron | it's not at all | 17:28 |
epitron | people just throw it around | 17:28 |
kanzure | anyway, | 17:28 |
epitron | here's a simpel example of emergent behaviour | 17:28 |
fenn | i phooeyed it because it's a buzzword that doesnt mean anything | 17:28 |
epitron | BEES and FLOWERS | 17:28 |
fenn | COEVOLUTION | 17:28 |
kanzure | emergence/mutation/insight are all on the same level | 17:28 |
epitron | yes | 17:28 |
epitron | or mass behaviour of bacterial colonies | 17:28 |
epitron | each bacterium only obeys local rules | 17:29 |
fenn | epitron: find me an organism that isn't co-evolved | 17:29 |
kanzure | re: insight, http://heybryan.org/thinking.html | 17:29 |
epitron | but all the local rules combine to create more complex global behaviour of the colony | 17:29 |
epitron | that's emergent behaviour | 17:29 |
fenn | epitron: find something, anythign at all, that isnt emergent | 17:29 |
epitron | when two organisms' co-adaptation creates a higher-level structure | 17:29 |
epitron | sigh | 17:29 |
epitron | i'm just explaining what the word means | 17:29 |
epitron | it is a useful term | 17:29 |
fenn | no, it isnt | 17:29 |
epitron | well how would you have expressed this without the word: 17:26 < kanzure> yeah, so our idea is to collect the emergent model and apply it to make the design for the self-replicator | 17:30 |
fenn | so, you will continue saying emergent and i will continue asking you precisely what you mean | 17:30 |
kanzure | okay, | 17:30 |
kanzure | so without the word 'emergent' | 17:30 |
epitron | i want fenn to try :) | 17:30 |
epitron | he's going to realize it's hard | 17:30 |
kanzure | no he's not | 17:30 |
epitron | haha | 17:30 |
kanzure | he knows a good way to reword it | 17:30 |
kanzure | and he's already used it to some extent | 17:30 |
epitron | well, what word does fenn like? | 17:30 |
kanzure | 'societal' ;) | 17:30 |
fenn | epitron: i dont know what he meant because the words didnt convey enough information | 17:30 |
epitron | ._o | 17:31 |
kanzure | that's why we called it skdb | 17:31 |
epitron | but societal means too many things | 17:31 |
kanzure | fenn: so, I want to get back on topic for just a sec | 17:31 |
epitron | in societal | 17:31 |
epitron | adj : relating to human society and its members; "social | 17:31 |
epitron | institutions"; "societal evolution"; "societal forces"; | 17:31 |
epitron | "social legislation" [syn: {social}] | 17:31 |
epitron | emergent is a more focussed term | 17:31 |
epitron | it just means "greater than the sum of the parts" | 17:31 |
fenn | that's what synergy means | 17:31 |
epitron | bucky has better more technical terms in his book synergetics | 17:32 |
epitron | :) | 17:32 |
kanzure | fenn: I realized (again) that modeling isn't necessarily going to be the most useful approach, so if we want to throw in various materials and their interactions with each other, who's to say that this information is useful as well? | 17:32 |
kanzure | Material properties are near meaningless for what we want to do ... we don't care what the properties do, as long as they allow the dependency-loop to be stable. | 17:32 |
fenn | kanzure: right, that's what i mean by 'economics' | 17:33 |
kanzure | huh? | 17:33 |
kanzure | economics == money stuff? | 17:33 |
kanzure | how does that come into this? | 17:33 |
fenn | the dependency loop is the cost | 17:33 |
kanzure | I don't get it. | 17:33 |
fenn | cost of complexity, cost of material scarcity | 17:33 |
kanzure | that's not what I am talking about at all | 17:33 |
kanzure | stability in the sense of functioanlity | 17:33 |
kanzure | *functionality | 17:33 |
kanzure | I don't care how much it costs to make the machine | 17:33 |
kanzure | how much other people 'value' the materials and other BS like that | 17:34 |
fenn | that's not what i mean at all | 17:34 |
kanzure | instead? | 17:34 |
fenn | i'm picturing a clanking replicator sitting in a clay pit | 17:34 |
kanzure | right | 17:34 |
fenn | if you come up with some material made out of rare earth metals and liquid helium, what good does it do the replicator? | 17:34 |
kanzure | you just said "come up with some material" | 17:35 |
kanzure | so are we back to material modeling again? | 17:35 |
kanzure | or not? | 17:35 |
fenn | 'if your comp-chem material finder program' comes up with.. | 17:36 |
fenn | i guess you could seed the search space with raw materials that are available | 17:36 |
kanzure | I thought that's what we were talking about when we mentioned constraining search space | 17:36 |
kanzure | for our programs. | 17:37 |
fenn | i just think it's too much like 'AI' | 17:38 |
fenn | if you give a bunch of ants parts to make a bridge, and show them how to do it, they arent going to build a bridge | 17:38 |
fenn | computers are similar | 17:38 |
kanzure | we can constrain the parts-space, and then we can constrain the functionality-space to certain types of functionality that allow the materials to process each other | 17:39 |
kanzure | which would mean that the ants would focus on certain functions of the parts, not certain 'properties' of the parts (like, this one is red) | 17:39 |
fenn | ok, what are the 'functions' of raw materials then? | 17:40 |
kanzure | my guess is the same batch processing reactions found in extractive metallurgy | 17:40 |
kanzure | which is what we need so that we can find which materials can make which for our d-loops | 17:40 |
kanzure | (you mentioned last night that many such batch methods are based off of each other for inspiration anyway) | 17:41 |
fenn | this has a bit of fractal nature to it | 17:41 |
kanzure | recursion wouldn't be replication | 17:41 |
kanzure | would it? | 17:42 |
fenn | the functionality available depends on which materials are available, and the materials available depend on which processes are available | 17:42 |
kanzure | the functionality available of the final machine depends on which materials are present in the final machine, and the materials that built in are dependent on which processes are available due to the functionality of the final machine, which would be used to generate more functionality for a second machine | 17:43 |
kanzure | same thing, said with more expressive terminology | 17:43 |
fenn | but more than that, you have to know what you're going to do with the materials before you even know what they are | 17:43 |
kanzure | right, then you have to find materials that can match those "functional-specs" | 17:44 |
kanzure | but those functional-specs are broad | 17:44 |
kanzure | since they depend on the materials | 17:44 |
fenn | i think there's circular logic in there, not just recursion | 17:44 |
kanzure | the functional-specs for replication with clay and sand etc. is constrained in comparison to functional-specs with say all minerals | 17:44 |
kanzure | so we have circular logic, recursion, fractals, d-loops, ... | 17:45 |
fenn | where is this all rooted? if it isn't rooted somewhere it's just a big fractal loop that is impossible to get to | 17:45 |
kanzure | is it rooted at the starting materials that *we* (the First Builders) give to it? | 17:46 |
kanzure | or do you mean a root in the abstraction? | 17:46 |
fenn | i mean, it connects to reality somewhere (or it doesnt) | 17:46 |
kanzure | I wonder if nonrooted systems can be designed | 17:46 |
kanzure | it connects to reality re: its environment | 17:46 |
fenn | there might be some uber-cool nano diamond mechanosynthesis replicator, but i dont know how to get there from here | 17:47 |
fenn | not just 'reality' (which is an abstraction) but the present, you and me and anyone else that's listening | 17:48 |
kanzure | please elaborate | 17:48 |
kanzure | it's certainly relevant to our projects | 17:48 |
fenn | well, if it were 1750 and we were in our laboratory, i could say 'look around, jenkins, these are the materials we have to work with' | 17:49 |
fenn | but the world's becomes considerably more abstract and fluid since then | 17:49 |
kanzure | but in the 1750s lab, you can't necessarily prove that those materials can be assembled in some combination to make a replicator | 17:49 |
fenn | i can have any chemical or device drop-shipped by tomorrow, if i can convince a certain person it's a good idea | 17:49 |
fenn | on the other hand, i do have a bunch of junk in my lab | 17:50 |
fenn | the reason i wanted autogenix was so i can say 'heres the junk i have, can i make this?' | 17:50 |
kanzure | but you have to be able to specify all material properties, functional properties, and on and on and on | 17:51 |
fenn | not 'all' | 17:51 |
fenn | it's not a mathematical proof, it's just a formalization | 17:51 |
kanzure | it seems that autogenix would be best for already designed projects | 17:51 |
kanzure | h, | 17:51 |
fenn | i would use a 555 as it's intended (a one-shot) even though the transistors as they are arranged are capable of doing weird op-amp stuff | 17:52 |
fenn | i dont expect a computer to come up with analog circuitry | 17:52 |
fenn | i DO expect it to be able to put together functional parts like digital circuitry | 17:53 |
kanzure | in the case of digital circuitry, all circuits are really just proposition statements and then some processing to compute a result | 17:53 |
fenn | if i dont tell it that some particular shape in an aluminum block is a microwave cavity, it's not going to figure that out | 17:53 |
kanzure | in our replicator design task, it's more than just 'logic' since the propositions are 'inputs' and the output has to be the entire proposition-deciding engine | 17:54 |
kanzure | yeah | 17:54 |
kanzure | *since the propositions are 'inputs' (materials) | 17:54 |
fenn | (although someone might write a libmicrowave module to do that) | 17:54 |
fenn | more than just materials, any design that already exists, no? | 17:55 |
kanzure | huh? | 17:56 |
kanzure | for starters, a replicator that can take input materials and make itself is good | 17:56 |
fenn | i propose that this assemblage of wires and glass is a computer | 17:56 |
fenn | autogenix is the logic engine | 17:57 |
kanzure | oh | 17:58 |
kanzure | I was talking about the replicator itself, the inputs to the replicator are materials | 17:58 |
fenn | libmicrowave is a simulator, it generates statements about the functionality of a certain shaped conductor | 17:58 |
kanzure | but if you are asserting that the replicator itself has to be autogenix, then that's interesting | 17:58 |
fenn | no, that' snot what i meant | 17:59 |
fenn | there needs to be some defined interface between autogenix (clear defined functionality) and a material property simulator | 17:59 |
kanzure | what part of autogenix are we talking about | 18:01 |
kanzure | the database that lists packages? | 18:01 |
kanzure | the package-finder to find packages that have certain properties? | 18:02 |
fenn | the package-finder | 18:02 |
fenn | i think | 18:03 |
fenn | there's a big pool of packages, one strategy is to build every possible tree based on the functionality of the packages (forward chaining) | 18:04 |
fenn | the other strategy is to pick the goal and find packages that provide that functionality until we get back to something that we already have or that satisfies our optimization constraints | 18:04 |
kanzure | one way to do it might be to first find the longest path, and then cut down on packages to simplify it to some necessary extent | 18:05 |
fenn | how do you simplify it? why not just pick the shortest path first? | 18:06 |
kanzure | so we need to grow the database like crazy, then find longest path, then cut back, get back a list of "requested packages" that might make things easier | 18:06 |
kanzure | I suppose you could | 18:06 |
fenn | or pick 'a' path | 18:06 |
kanzure | I thought that we wanted more dependencies, so that we can ensure that we probably have a possible machine within it? | 18:06 |
fenn | then find the local minimum | 18:06 |
fenn | you could walk down the dependency trees randomly, leaving bread crumbs | 18:08 |
kanzure | log files? | 18:08 |
fenn | no, just an algorithm | 18:08 |
fenn | to find the loops that may be replicators | 18:08 |
kanzure | a bread crumb = algorithm? | 18:09 |
fenn | bread crumb is a piece of data in RAM that says 'i've already been down this path' | 18:09 |
fenn | you're assuming that a path from A to B exists | 18:10 |
fenn | and that you can just 'see' it | 18:10 |
fenn | but a computer doesnt know that | 18:10 |
kanzure | ah, okay, a path finding algorithm is easy though | 18:10 |
kanzure | these have been proven | 18:11 |
kanzure | but we need to be able to make a 'path' in the first place | 18:11 |
kanzure | and this will be based off of something in computational materials modeling | 18:11 |
fenn | not necessarily | 18:11 |
kanzure | I still doubt it's the *properties* of the materials that we need to model | 18:11 |
kanzure | but rather the functionality. | 18:11 |
kanzure | whatever that means | 18:11 |
fenn | you are assuming that you have to create the material properties and thus 'functionality' from first principles | 18:11 |
fenn | which i think is probably impossible | 18:11 |
fenn | but we have huge databases of parts and materials and their properties already | 18:11 |
fenn | and those databases are based on the real world, empirical data | 18:12 |
kanzure | right, I don't care if we don't do it from first principles | 18:12 |
andares | Hi guys. | 18:12 |
kanzure | can you show me a database that, say, has a formalized entry for clay + sand batch processing already? | 18:12 |
kanzure | that's the sort of 'functionality' that I am referring to. | 18:12 |
fenn | no, i dont know anything about geology or even if such a database exists | 18:12 |
kanzure | didn't you just say we have databases of parts already ? | 18:12 |
-!- zedrdave [n=zedrdave@dan75-1-87-88-125-171.dsl.club-internet.fr] has joined #hplusroadmap | 18:12 | |
kanzure | that do these things? | 18:12 |
fenn | yeah but nobody is interested in making replicators | 18:13 |
andares | Why not? | 18:13 |
fenn | or digging in the dirt | 18:13 |
fenn | because nobody's ever made a replicator | 18:13 |
-!- zedrdave [n=zedrdave@dan75-1-87-88-125-171.dsl.club-internet.fr] has left #hplusroadmap [] | 18:13 | |
andares | Well, they've made replicator-like things. | 18:13 |
andares | Like those new 3d inkjet printers. | 18:13 |
andares | Though those are far from optimal. | 18:13 |
fenn | well, things are changing i guess | 18:13 |
fenn | hence why we're here | 18:13 |
andares | fenn, I looked up fpgas yesterday. | 18:14 |
andares | After you were talking. | 18:14 |
epitron | FPGAs! \o/ | 18:14 |
fenn | by 'nobody' i mean, huge decades-old organizations with more money than they know what to do with, have not made databases relevant to building replicators from raw minerals | 18:14 |
kanzure | Hi zedrdave -- I noticed you mentioned an interest in ai, comp sci, and bio. At the moment we're discussing http://heybryan.org/mediawiki/index.php/Self-replication | 18:14 |
kanzure | blah, he left | 18:14 |
kanzure | fenn: ah | 18:14 |
andares | epitron, have you done any work with them? | 18:14 |
epitron | not really.. just VLSI simulation | 18:14 |
epitron | and lots of thinking about what could be done with them :) | 18:15 |
andares | fenn, is it possible to make a neural network that uses fpga nodes though, instead of complex activation functions? | 18:15 |
kanzure | sure | 18:15 |
andares | Well, using commonly-available FPGAs. | 18:15 |
kanzure | fenn: so, I still don't know what this sort of data would look like | 18:15 |
fenn | andares: i have no idea | 18:15 |
kanzure | I don't even know where to collect it | 18:15 |
andares | For instance, I'm not sure how you'd do neural weights. | 18:15 |
-!- zedrdave [n=zedrdave@dan75-1-87-88-125-171.dsl.club-internet.fr] has joined #hplusroadmap | 18:16 | |
andares | You'd need a variable transistor or something, no? | 18:16 |
epitron | neural nets are probably better suited for the more complex chips... the one where they have many tiny units that do a more complex function than an fpga gate | 18:16 |
fenn | andares: frankly, i think neural networks is crap pseudo-science, but i probably shouldnt tell you that | 18:16 |
epitron | and they have better routing too :) | 18:16 |
kanzure | Hi zedrdave -- I noticed you mentioned an interest in ai, comp sci, and bio. At the moment we're discussing http://heybryan.org/mediawiki/index.php/Self-replication -- which I plan to help boostrap intelligence augmentation later, a subject close to ai | 18:16 |
andares | fenn, well, ANNs might be, but they've been put to productive use. | 18:16 |
kanzure | andares: please go read up on computational neuroscience first | 18:16 |
epitron | it's not nerual nets, it's backprop that sucks :) | 18:16 |
epitron | the training algorithm is everything | 18:17 |
andares | epitron, moreso the computational limits of conventional computers. | 18:17 |
fenn | andares: you might not be aware that transistors are analog devices | 18:17 |
andares | The training algorithm works okay. | 18:17 |
kanzure | andares: I'd teach you myself, but I haven't really sufficiently explored it all yet | 18:17 |
andares | fenn, I am aware. | 18:17 |
epitron | the training algorithm has been proven to be very limited :) | 18:17 |
andares | kanzure, I'm continuing to look it up. | 18:17 |
epitron | mathematically equivalent to something else well understood and also limited | 18:17 |
andares | epitron, true, I tried to train a net to square numbers, which failed. | 18:17 |
epitron | andares: you should look into restricted boltzmann machines | 18:17 |
kanzure | andares: http://heybryan.org/mediawiki/index.php/Computational_biology -- you can go play with GENESIS today or something, that should be loads of fun | 18:17 |
epitron | oh i remember now | 18:17 |
andares | Then asked some AI people, who ranted at me that "it wasn't a practical application." | 18:17 |
andares | Oh, hopfield nets. | 18:18 |
andares | fun. | 18:18 |
epitron | the problem with neural nets is that: a) they're VERY SLOW to train many layers, and b) they need LOTS OF LABELLED DATA... like.. millions of labels | 18:18 |
fenn | once, i tried to peel hardboiled eggs with a sledgehammer.. | 18:18 |
andares | epitron, that's a problem of the hardware though, not the math. | 18:18 |
epitron | restricted boltzmann machines are awesome because they use the data itself as the labelled data | 18:18 |
fenn | then my mother ranted at me that it wasnt a practical application of a sledgehammer! | 18:18 |
epitron | they train themselves by learning to generate the data they're given | 18:18 |
epitron | so basically they're pattern recognizers | 18:19 |
andares | epitron, you can do that with MLPs too. | 18:19 |
epitron | not the same way though :) | 18:19 |
andares | Just make a network with training input = training output, then have the nodes constrict in the middle. | 18:19 |
epitron | you should look at this talk -- | 18:19 |
andares | But lots of things fall through the cracks if it can't totally reform input. | 18:19 |
kanzure | fenn: obviously we're not going to have our computers coming up with "functional creativity" for reusing systems and so on | 18:19 |
kanzure | fenn: that's like saying "Insert magical Ai here" | 18:19 |
epitron | http://youtube.com/watch?v=AyzOUbkUf3M | 18:19 |
kanzure | however, if that's what people would contribute modules for, that's fine | 18:19 |
fenn | kanzure: do you really expect people to use those lists of links? like computational_biology | 18:19 |
epitron | andares: that's a great talk | 18:19 |
kanzure | fenn: yeah, why not? | 18:19 |
kanzure | fenn: I did / still do | 18:20 |
epitron | you use them because you know what they are :) | 18:20 |
fenn | well, the url's usually dont mean anything | 18:20 |
kanzure | no I don't | 18:20 |
andares | Okay, looking up genesis. | 18:20 |
kanzure | I don't know what the hell they are ;) | 18:20 |
zedrdave | erm. I'd love to take part in that conversation, but I don't think I can conduct normal daily activities and follow more thant 5% of what's written :P | 18:20 |
epitron | ok well, i guess you're just a mutant then :) | 18:20 |
kanzure | zedrdave: Heh. Alright :) | 18:20 |
epitron | less is more! | 18:21 |
zedrdave | but if I could say smth: to the person who failed to do supervised learning on a sq function: most likely bad parameters. | 18:21 |
zedrdave | any decently parametered NN with one hidden layer can do it. | 18:21 |
epitron | zedrdave: you seen boltzmann machines yet? | 18:22 |
epitron | (restricted) | 18:22 |
zedrdave | mmnyea... | 18:23 |
kanzure | fenn: again, please see my objection to your sledgehammer example. You'd need an ai for a computer to do such 'novel uses of objects'. So I guess those 'novel uses of materials' are what we would have people insert in the wiki, right? | 18:23 |
zedrdave | what about them? | 18:23 |
zedrdave | mostly simulated annealing in an NN, afaik. | 18:23 |
andares | arr firefox crashes. | 18:23 |
kanzure | fenn: and then we'd need a computer to be able to "use" and "apply" those novel uses of materials, to do path-following | 18:23 |
* zedrdave is really not sure of where all this is going. but is going with it for the sake of it :) | 18:24 | |
kanzure | zedrdave: Andares is doing NNs for the sake of learning, it seems. | 18:24 |
andares | And flash breaks again. | 18:24 |
fenn | heh kanzure i was referring to the use of NN's for squaring numbers | 18:25 |
zedrdave | well, doing NN for the sake of learning is prolly a good idea. | 18:25 |
kanzure | fenn: oops. Well. I think my misinterpretation was useful. | 18:25 |
fenn | sure | 18:25 |
zedrdave | certainly more sound than doing NN in the hope of building tomorrow's emerging artificial consciousness... | 18:25 |
andares | fenn, I did the squared numbers test because I was looking at the structure of MLPs and realized I didn't think it was possible. | 18:25 |
kanzure | fenn: you mean, I misinterpreted your sledgehammer example, right? | 18:25 |
andares | I wasn't actually going to use a neural net to square numbers for any reason other than fun/science. | 18:26 |
zedrdave | MLP = multi-layer perceptron? | 18:26 |
kanzure | zedrdave: hehe. Yeah, building ai sounds dubious. Eliezer and Goertzel haven't seemed to make much progress re: OpenCog, and I'm just not seeing how their plan works out overall. Henry Markram's stuff is awesome, but. | 18:26 |
fenn | kanzure: you misinterpreted what i was saying, but correctly understood that humans will be submitting novel uses of materials | 18:26 |
andares | zedrdave, yeah | 18:26 |
kanzure | fenn: okay | 18:26 |
kanzure | fenn: we need a way to formalize novel functions so that the computer can use them automatically as it's finding the shortest path | 18:27 |
kanzure | how would we do that, without restraining the possible uses? | 18:27 |
zedrdave | andares: MLP have been proven to be very good approximators for pretty much any math function you can think of. | 18:27 |
fenn | kanzure: however, simulations can also provide lists of 'functionality' code objects | 18:27 |
zedrdave | you need a good cut-off function. sigmoids for ex. | 18:27 |
kanzure | I guess the computer could just say "Humans! Please make a functional interface between these two packages. Might be a shorter path than <something else>. Good day." | 18:27 |
andares | zedrdave, well, since I've implemented some ways to fit input and output for the network, maybe I'll give squaring another try. | 18:27 |
andares | But with straight input/output it's impossible. | 18:27 |
fenn | kanzure: that's where graph visualization comes in | 18:28 |
zedrdave | erm. once again... you need decent parameters. | 18:28 |
andares | Exactly. I didn't think about that at the time. | 18:28 |
zedrdave | *possible* doesn't mean easy. and certainly nt fast in all cases :) | 18:28 |
fenn | kanzure: also the ignore tool, ignores dependencies of a certain type, or pretend pretends that you already have some package (or something like that) | 18:28 |
zedrdave | also probably the right number of hidden neurons. | 18:28 |
fenn | s/package/functionality/ | 18:29 |
zedrdave | note that learning this type of quadratic function is a very poor use of a NN... | 18:29 |
zedrdave | can be done, but pretty much driving screws with a hammer. | 18:29 |
kanzure | fenn: That *does* sound good. Now I am wondering whether or not we can have it so that when you submit a 'novel functionality', you're really submitting a new type, and this new type (just like C++ | 18:30 |
kanzure | objects) can have different preprogrammed ways to interface with other objects. Probably just a collection of scripts. And the compiler will come around and *fail* if it can't do something yet, so that's where we do blackboxing. Ah, makes sense (EOL). | 18:30 |
fenn | kanzure: did you read the autogenix format draft? http://fennetic.net/pub/irc/autogenix-format-v0.0.2.txt | 18:30 |
kanzure | yes | 18:30 |
zedrdave | problem with MLPs is that they are universal function approximators. so in essence, if you can put it into undergrad math, chances are it can learn it. | 18:30 |
kanzure | I don't remember if it accounts for all of this, fenn. | 18:30 |
zedrdave | *but*, it's also subject to all sorts of very annoying dimensionality issues. and the bias-variance is horrible... | 18:31 |
zedrdave | so, not a very good way to train in machine learning. | 18:32 |
fenn | kanzure: right, functionality is like types, but i think its more like a python attribute than a C++ type | 18:32 |
fenn | because a car and a truck are different types, but they have the same functionality | 18:32 |
kanzure | fenn: can you explain python types to me, then? I need to go read about them. | 18:32 |
fenn | they both provide moveHeavyShit() | 18:32 |
kanzure | http://docs.python.org/lib/types.html | 18:32 |
fenn | python is strongly typed, meaning you can't cast anything | 18:33 |
kanzure | (and users would contribute moveHeavyShit(), of course) | 18:33 |
kanzure | you can't cast? yikes | 18:33 |
kanzure | I love casting :) | 18:33 |
fenn | casting is very necessary if you cant access attributes directly | 18:33 |
kanzure | int jitterbugs = (int *) someOtherClassOfData; | 18:33 |
fenn | so you have an object someObject, you want an int from it | 18:34 |
fenn | you'd do something like jitterbugs = someObject.number | 18:34 |
fenn | uh poor example and not sticking too close to reality, but number would be the built-in method that returns an 'int' object | 18:35 |
fenn | s/object/attribute/ | 18:35 |
fenn | one sec, lemme get a link | 18:35 |
kanzure | yes, I can see how that works | 18:35 |
kanzure | but this is sounding like we are falling into OOP neverland | 18:36 |
fenn | http://en.wikipedia.org/wiki/Duck_typing | 18:36 |
zedrdave | if I may ask a silly question. what's OOP gotta do with self-replication? :) | 18:36 |
fenn | zedrdave: when you're designing the cad system to build the replicator | 18:37 |
fenn | zedrdave: it's a fun recursive problem :\ | 18:37 |
kanzure | zedrdave: Kinematic self-replication, i.e. macroscale stuff, not DNA and below-Brownian-diffusion limits stuff (definitely *above*). | 18:37 |
fenn | i think atomic scale replication would actually be easier | 18:37 |
kanzure | zedrdave: So, we figure that there are such things as 'dependency loops' in the design, where one package has to be able to rely on the other available materials to make itself and the other packages. | 18:37 |
zedrdave | I don't think compiled language are your best options for that type of task (within this decade) :) | 18:37 |
kanzure | zedrdave: Right. We are doing a giant database and then will be doing path-finding through the database to find a "closed-loop", indicating a set of materials that can make each other (given certain inputs) | 18:38 |
fenn | zedrdave: definitely not, in fact it's not even a program, but rather a wiki-ish database with attached code modules | 18:38 |
zedrdave | erm... I'm still quite hazy on what you are trying to achieve... care to sum it up from the ground up in less than 3 sentences (I did glance at your wiki, but frankly don't have time to read 50 pages right now) | 18:39 |
zedrdave | self-replication, genetic programming, modular programming, OOP? | 18:39 |
fenn | say you want your rapid prototyper to spit out a fully functional robot, you need to make all the components of the robot or else have them on hand | 18:40 |
zedrdave | erm. that's not how most of robotics work :) | 18:40 |
-!- Enki-2 [n=weechat@c-71-234-190-248.hsd1.ct.comcast.net] has joined #hplusroadmap | 18:40 | |
fenn | the program/database we are making will determine whether it's possible to make the robot with the tools and components and materials you have | 18:40 |
zedrdave | and also, what level "component" are you talking about. | 18:40 |
zedrdave | mmn... ok, starting to see a bit better what you are inkling at. still quite blurry though. | 18:41 |
kanzure | yes, it is blurry | 18:41 |
kanzure | otherwise we'd have the solution! | 18:41 |
fenn | it's 'emergent'! | 18:41 |
kanzure | hah | 18:41 |
zedrdave | erm. I contend you don't have a problem either. | 18:41 |
kanzure | zedrdave: Really? Then please design us a self-replicating machine. | 18:41 |
fenn | yeah :( | 18:41 |
kanzure | please | 18:41 |
zedrdave | ah ok. that's easy. | 18:42 |
fenn | we dont have a specification for a problem he means | 18:42 |
kanzure | zedrdave: please elaborate on our design | 18:42 |
zedrdave | I'd do it in Prolog, but I'll even do it in PHP for you :) | 18:42 |
kanzure | *your | 18:42 |
kanzure | huh? | 18:42 |
kanzure | it's a physical machine | 18:42 |
fenn | replicating information is trivial | 18:43 |
zedrdave | mmnyea. that's what I thought... | 18:43 |
kanzure | is mmyea == Office Space ref? | 18:43 |
zedrdave | so, erm, going for physical electronic replication? | 18:43 |
kanzure | zedrdave: http://reprap.org/ has a start to a project like this, but they can't entirely self-replicate all of their parts (and we doubt they ever will, the way they are doing it). | 18:44 |
zedrdave | no. it's me expressing a mix of confusion, slight incredulity and very hazy understanding... | 18:44 |
fenn | i find that a lot of AI researchers are totally clueless when it comes to building stuff, or the complexity involved | 18:44 |
zedrdave | fenn: then again. define "building stuff". | 18:44 |
kanzure | zedrdave: Well. Electronics are not necessary, but it'd really help out with the internal logic of the final replacing machine, I think. | 18:44 |
fenn | zedrdave: putting together pieces of physical matter | 18:44 |
fenn | usually with a goal in mind | 18:45 |
zedrdave | for all my practice in AI, I wouldn't have the faintest idea of how to put together a car engine... | 18:45 |
zedrdave | oddly enough, I don't consider that a hindrance to my goals :) | 18:45 |
kanzure | zedrdave: Putting together an engine can be easy, it's just a set of instructions, like a program. | 18:45 |
fenn | i think its the root of the symbol grounding problem | 18:45 |
kanzure | but you have to have the tools available to express the program | 18:46 |
zedrdave | silly theoretician that I am, I tend to think that AI is more paper than oil and cogs... | 18:46 |
kanzure | not just a keyboard but also the CNC shop, the lathe, the metal, | 18:46 |
fenn | in a physical control system there is no symbol grounding problem. the variables represent measured quantities | 18:46 |
kanzure | (also the smarts to never start the ignition within 50 feet of the engine ..) | 18:46 |
zedrdave | okay. if I may: | 18:47 |
zedrdave | there is *one* (count it: one) and only one reason why you sometimes need to leave your math model and computorial simulation behind and go dig for some actual cogwheels: | 18:48 |
fenn | the difference between theory and practice is much greater in practice | 18:48 |
zedrdave | the impossibility to build fully realistic complex system simulations. | 18:48 |
kanzure | we are not interested in simulations | 18:48 |
fenn | also there is the limits of computater power | 18:48 |
zedrdave | but then you are more of a physicist than an AI researcher. | 18:49 |
kanzure | although simulations can be useful, and indeed entertaining, they are not reality | 18:49 |
kanzure | how could they ever replace the fundamental reality that they represent? No matter how precise? | 18:49 |
zedrdave | if thermodynamics affect your model, then you're probably in the wrong direction. | 18:49 |
kanzure | hm, I don't know whether or not thermo is a factor or not | 18:49 |
fenn | kanzure: electronics simulations have gotten pretty close, because it's well understood and amenable to modeling | 18:49 |
kanzure | fenn: surely thermo has its place in our model, but it's more like just an optimization thing | 18:49 |
zedrdave | well, they're the root of what makes the difference between reality and simulation. | 18:50 |
zedrdave | you can modelize a crapload of things nowadays. | 18:50 |
kanzure | fenn: simulating electronics requires, at the moment, electronics in the first place! But that doesn't mean that running a virtual circuit on a physical circuit makes the 'virtual circuit' just as real as the physical circuit it's running on .. layers of abstraction, layers of abstraction. | 18:50 |
kanzure | yeah, I agree that we can do lots of models and so on | 18:50 |
fenn | kanzure: it is just as real if it's a digital circuit | 18:50 |
zedrdave | what you can't really modelize well, for *purely* numeric reasons, is the atomic level (and even the macro level, in most case) of thermodynamic interactions. | 18:50 |
fenn | a processor in an FPGA is just as real as a hardwired processor | 18:51 |
kanzure | fenn: perhaps intrinsically | 18:51 |
zedrdave | but once again. if you're doing AI. that's really not much of your problem. | 18:51 |
kanzure | fenn: FPGAs are a different story, since those are physical gates that are switching connections and whatever | 18:51 |
zedrdave | but I get the feeling here we're saying AI when we actually mean Robots. | 18:51 |
fenn | fine, an emulator running in an emulator ad infinitum | 18:51 |
kanzure | I don't know what we are talking about any more. | 18:52 |
kanzure | fenn: we lost track | 18:52 |
kanzure | we were on a roll with something | 18:52 |
* fenn blames hazy-ai-researcher | 18:52 | |
kanzure | oh, right, duck typing | 18:52 |
zedrdave | but I'm sure glad you kids like doing that sort of unrewarding robot-elbow-grease tuning stuff... | 18:52 |
fenn | attributes are provided by simulations? is it possible? | 18:52 |
zedrdave | I need people like you to spend hours turning my cool models into rather lame but visually pleasing walking robots :) | 18:53 |
fenn | attributes can be provided by humans (fairly smart ones at least) | 18:53 |
kanzure | uhh | 18:53 |
fenn | zedrdave: perhaps you should join #robots if that's what you're interested in | 18:54 |
fenn | er, ##robotics | 18:54 |
kanzure | he seems to be focused on abstractions only | 18:54 |
zedrdave | erm, no, that's precisely what I was saying :) | 18:54 |
kanzure | he also seems to think the simulation is more important than the reality of the situation | 18:54 |
kanzure | O.o | 18:54 |
zedrdave | robots are cool and all. but they're just a way to let the hoi polloi understand what your model does in reality. | 18:54 |
fenn | zedrdave: that's bullshit | 18:55 |
kanzure | anyway, fenn - yeah, so we were talking about duck typing, it looks good to me (just skimmed the wikipedia article) | 18:55 |
zedrdave | real AI people don't care about cogwheels ;) | 18:55 |
mech0r | hello zedrdave | 18:55 |
mech0r | hello Enki-2 | 18:55 |
fenn | zedrdave: without robotics your fancy AI will go nowhere. | 18:55 |
mech0r | hello andares | 18:55 |
kanzure | without physical implementation, in general | 18:55 |
kanzure | even if that implementation is in your head | 18:55 |
kanzure | when you think about it. | 18:55 |
zedrdave | fenn: sure, if you read what I said above: there is a point where you may want robotic models to test certain things. | 18:55 |
zedrdave | but note that a good 90% of AI has strictly nothing to do with robots. | 18:56 |
fenn | not for testing, for building, advancing technology and science and all that | 18:56 |
kanzure | entropy maximization ;) | 18:56 |
fenn | yeah, stupid stars dont even know how to properly burn hydrogen | 18:56 |
zedrdave | and a good 90% of robotics has nothing to do with AI either. | 18:56 |
kanzure | fenn: heh, look how much waste is left over! (i.e., us) | 18:56 |
fenn | they just explode every few kilo-aeons | 18:57 |
zedrdave | also, not using robots doesn't mean it's "all in my head"... there's this fancy tool called computer that does a very good inbetween :) | 18:57 |
kanzure | yes, but you're not building | 18:57 |
kanzure | huh | 18:57 |
kanzure | fenn: you were right about the engineering distinction | 18:57 |
kanzure | of course, building models is building, yes | 18:57 |
zedrdave | kanzure: you are building just the same. | 18:57 |
zedrdave | the only difference is your environment. | 18:58 |
fenn | kanzure: uh, which engineering distinction again? | 18:58 |
kanzure | fenn: you said something about how ai researchers don't understand the distinctions in engineering and their semantics or whatever | 18:58 |
kanzure | zedrdave: that's true | 18:58 |
kanzure | zedrdave: but we're still working on our model | 18:58 |
kanzure | so I don't see where you are going with this | 18:58 |
fenn | kanzure: it was just a general stereotype | 18:58 |
zedrdave | well then, I fail to see really what it should care about real world matter at this point. | 18:58 |
zedrdave | if it's a model, it's paper and possibly computers. | 18:59 |
kanzure | "it should care about" ? what is it? | 18:59 |
kanzure | *what is it "it" ? | 18:59 |
zedrdave | your model. | 18:59 |
fenn | if the model is a model, the model is paper and possibly computers | 18:59 |
kanzure | because we want a proof of possibility? | 18:59 |
zedrdave | you build your model using fully abstract concepts. possibly software tools. | 18:59 |
fenn | a model cant be made from fully abstract concepts, that's impossible | 19:00 |
zedrdave | fenn: that's the very definition of a model :) | 19:00 |
fenn | yay! do i get a cookie? | 19:00 |
zedrdave | doesn't mean you do not integrate physical constraints. | 19:00 |
zedrdave | fenn: no. the very definition is what you describe as "impossible". | 19:01 |
kanzure | you can't separate from reality -- that's thermodynamically impossible | 19:01 |
zedrdave | models put together abstract concepts that appropriately abstract what you are eventually trying to achieve. | 19:01 |
kanzure | context, context! | 19:01 |
fenn | An abstract model (or conceptual model) is a theoretical construct that represents something, with a set of variables and a set of logical and quantitative relationships between them. | 19:01 |
zedrdave | kanzure: it's not only very possible, very desirable but also ultimately the only way to solve anything. | 19:01 |
zedrdave | if a computer can't solve most themodynamics, neither will you. | 19:02 |
fenn | in this twisted tortured definition we see the word "something" - this "something" is the non-abstract | 19:02 |
kanzure | zedrdave: You need to read some Egan. Even if you just (supposedly) came up with the most abstract thing ever, it's still *related to you*, sitting there shuffling neurotransmitters around in your head. | 19:02 |
* zedrdave sighs ever so little | 19:02 | |
zedrdave | I don't think we really need to revert to tautologies just yet. | 19:03 |
kanzure | (and thus my interest in self-modification re: neuroengineering and why we might need self-replicators for combinatorial testing, or simulations as fenn suggests) | 19:03 |
zedrdave | yes, whatever you do, whatever you think, it's run on a physical substrate. | 19:03 |
fenn | simulations provide the fodder for combinatorics | 19:03 |
fenn | they arent the same thing | 19:03 |
zedrdave | but... and that's pretty much day one of AI... there's the idea that those concept have an existence separate from their physical substrate. | 19:04 |
fenn | sure, that's just abstraction | 19:04 |
zedrdave | and that's what you call a model. | 19:04 |
fenn | an abstraction without a physical substrate is useless | 19:05 |
kanzure | most of our substrates seems to be (1) brains and (2) computers. | 19:05 |
kanzure | *seem | 19:05 |
zedrdave | then there's the next (often mixed together for obvious reasons) step of putting down that idea into smth a bit more concrete. like an equation on a piece of paper or a PoC software program. | 19:05 |
fenn | or at least a lower level of abstraction, that eventually works its way down to the real world | 19:05 |
kanzure | fenn: yeah, but the abstraction is always real (you just created the idea, no? and ideas have a physical substrate) | 19:05 |
kanzure | (even though ideas have not been fully quantified) | 19:05 |
zedrdave | if that works and the next step (turning it into whatever, a robot for your sake) doesn't. you need better models. and better roboticists. | 19:05 |
kanzure | "I am confident that this idea can be quantified." | 19:06 |
zedrdave | (and probably no AI to being with) | 19:06 |
fenn | kanzure: yes, at this stage in our understanding of possibility space :) | 19:06 |
kanzure | hehe | 19:06 |
zedrdave | kanzure: okay. I don't really think discussing Platonic theories of Ideas will do much to help settle the debate. | 19:07 |
fenn | no | 19:07 |
zedrdave | bottom line is: *of course* you need to get to a hardware implementation at some point (depending on the problem), but in roughly 99.9999% of cases, there is strictly no reason to jump directly there. | 19:08 |
zedrdave | if you can't put it on a piece of paper first, then you are going at it wrong. | 19:09 |
zedrdave | that, or the entire scientific community of this planet has been going wrong at it for a couple centuries now :) | 19:09 |
kanzure | if it's in your head then it has a physical substrate already | 19:09 |
kanzure | sigh | 19:09 |
fenn | the scientific community constantly deals with phyical reality | 19:09 |
fenn | if you believe laws and textbooks and all that crap is science, you're mistaken | 19:09 |
zedrdave | erm. apparently I do. it's done me pretty good so far :) | 19:10 |
kanzure | I don't know what's going on anymore | 19:10 |
fenn | maybe you should have said 'mathematicians' | 19:10 |
kanzure | what are we talking about? | 19:10 |
zedrdave | and precisely, the work of scientists is to *abstract* physical reality | 19:10 |
kanzure | you cannot escape into abstraction completely | 19:10 |
kanzure | but I already brought up this point | 19:11 |
zedrdave | putting models back into physical shapes and ensuring it works... is not science research nor AI. it's got a name. it's called *ENGINEERING* | 19:11 |
fenn | engineering goes the other way too | 19:11 |
fenn | it's just applied science | 19:11 |
zedrdave | sure... | 19:11 |
fenn | (as if there were a non-applied science) | 19:12 |
fenn | oh, yeah its called math | 19:12 |
kanzure | but you still do math engineering | 19:12 |
kanzure | how do you think we train people to become mathematicians? | 19:12 |
kanzure | we are working with their brains to support math-models and other programs | 19:12 |
fenn | and there's lots of starving mathematicians-in-becoming out there, in the rat-race-sweatshop | 19:13 |
fenn | inventing new math isnt helping them to become mathematicians | 19:13 |
fenn | sad but true | 19:13 |
* zedrdave is less and less sure of where this all is going... | 19:13 | |
zedrdave | nor what that point about mathematicians was. | 19:13 |
fenn | whatever your goal is, it involves physical reality at some point | 19:14 |
zedrdave | (oh yea, and for the record, you stand way better chance to make millions if you are a mathematicians than a budding robotmakers, these days) (not that it really matters) | 19:14 |
fenn | uh.. show me a millionaire mathematician please | 19:15 |
zedrdave | fenn: once again, that's pretty much carpet-bombing fishes in a 3nm barrel... | 19:15 |
kanzure | fenn: that might be possible, a cool quest. Shall we googlestalk? | 19:15 |
fenn | i can think of one: wolfram | 19:15 |
kanzure | ooh, but I thought he was trained in physics? | 19:15 |
zedrdave | may I ask one question.... | 19:15 |
kanzure | his first article when he was 16 was on QED | 19:15 |
zedrdave | fenn: how old are you? | 19:15 |
kanzure | and he knew Feynman, so I'm guessing physics | 19:15 |
fenn | yes, but he got rich off of publicity and media | 19:15 |
kanzure | fenn: I agree with "it involves reality at some point" :) | 19:16 |
fenn | zedrdave: why do you ask? | 19:16 |
kanzure | fenn: yeah, and then his Mathematica package. Big bucks. | 19:16 |
zedrdave | just curious, really... | 19:16 |
fenn | oh, he wrote mathematica? i guess i should have known that | 19:16 |
fenn | zedrdave: 25, you? | 19:16 |
zedrdave | oh yea, and as for mathematicians: any math geek who cares to make a buck goes into finance math and makes more in a year than any engineering job will ever pay in a decade. | 19:17 |
zedrdave | fenn: a bit more... | 19:17 |
fenn | number! | 19:17 |
zedrdave | 28 if you really care :| | 19:17 |
fenn | it's only fair, you see | 19:17 |
zedrdave | I was just wondering if you were of the same age bracket as kanzure... that's all :) | 19:18 |
fenn | not the economic system, our little information-scarcity game | 19:18 |
fenn | um, am i? | 19:18 |
zedrdave | not that I know of. | 19:18 |
kanzure | I am 18. | 19:18 |
fenn | i really dont know anymore | 19:18 |
fenn | it seems to be related to how long people have been allowed to do what they want | 19:18 |
zedrdave | that's not the same bracket... | 19:19 |
zedrdave | not that it should matter all that much. | 19:19 |
kanzure | yeah, I've been on the net for ages, I feel much older than I should | 19:19 |
zedrdave | sure... | 19:19 |
fenn | some guy who's been working in the ford plant all his life might as well be a high school graduate | 19:19 |
zedrdave | well, I pretty much was using the net before you were born. so I guess that makes me really ancient. | 19:19 |
kanzure | zedrdave: haha | 19:19 |
fenn | yeah it does | 19:20 |
fenn | how did you begin (end up?) in that situation? | 19:20 |
kanzure | zedrdave: yeah, based off of the way the internet develops (rate-wise), that's a difference of maybe 6 billion websites | 19:20 |
zedrdave | also I can pretty much see myself in your position around your age... | 19:20 |
kanzure | I've had people tell me "Son, I was on the internet before you were born." and that really sets things into perspective | 19:20 |
zedrdave | with way more enthusiasm than actual time to sit down and think of things in rational terms :) | 19:20 |
kanzure | that's bullshit, we are pretty clear in what we are thinking about | 19:21 |
zedrdave | oh, I'm sure. | 19:21 |
kanzure | don't pull the "you're young and clueless" shit | 19:21 |
zedrdave | nope. | 19:21 |
zedrdave | I wasn't... | 19:21 |
zedrdave | although..... :D | 19:21 |
fenn | zedrdave: what's the holy project you're working on then? maybe we can nip at its ankles | 19:21 |
zedrdave | fenn: see... that's the thing. | 19:21 |
zedrdave | I *did* have holy projects and all when I was 18. | 19:22 |
zedrdave | and then reality hit me in the head. | 19:22 |
fenn | and now you're a jaded old bastard that likes to attack peoples' dreams | 19:22 |
zedrdave | now... I'm criticizing the way you go at it. | 19:22 |
zedrdave | which I know is a futile thing to do. but you kinda got me into it. | 19:22 |
fenn | you dont even know what we're doing | 19:23 |
fenn | hell *I* dont even know what i'm doing exactly | 19:23 |
zedrdave | what I mean is that trying and tackling a "grand project" from the ground up, rarely ever worked for anybody. | 19:23 |
kanzure | right, but that's why we avoid blackswans | 19:23 |
kanzure | that's why we rely on community dynamics | 19:23 |
kanzure | (i.e., contributers to the database) | 19:23 |
kanzure | *contributors | 19:23 |
zedrdave | but I know it never stopped me from trying when I was 18. so I wouldn't try to stop you :) | 19:23 |
fenn | zedrdave: i want a program that can spit out a robot, pray tell how i should go about it | 19:24 |
zedrdave | i guess my humble experience has always been that, trying to tackle smaller, more obvious things, always eventually leads to tackling the big picture. except you have a much clearer idea of what you are doing by then. | 19:24 |
zedrdave | fenn: afai understood, you want way more than that... | 19:25 |
zedrdave | you want a robot that is capable of spitting out a robot. right? | 19:25 |
fenn | yes of course | 19:25 |
zedrdave | because as for a program that spits out a robot... it's got little to do with AI :) | 19:25 |
kanzure | fenn: for when we're done with zedrdave, check out http://heybryan.org/mediawiki/index.php/Egan_quote :) | 19:25 |
fenn | and hopefully it's capable o fdoing more than just spitting out another robot | 19:25 |
zedrdave | depending on what you want that spit out robot to look like, of course. | 19:26 |
fenn | i.e. not just a burning piece of paper | 19:26 |
zedrdave | also. that may sound a bit contemptuous, but I swear it's not: | 19:28 |
zedrdave | you may want to go and read some old-fashioned dead greek guys for a change. | 19:29 |
fenn | i'd rather get some work done actually | 19:29 |
zedrdave | Plato and Aristotle would do much to talk you out of that silly habit of quoting the "brain <-> idea" connection as the ultimate "abstraction doesn't exist" argument... | 19:29 |
zedrdave | and that's smth very important to understand, because it's what the whole of AI is built upon. | 19:30 |
zedrdave | the idea that thinking isn't just a matter of cogwheels. | 19:30 |
fenn | and look at where it's gotten us | 19:31 |
* fenn gestures at the broad expanse of intelligent agents out there, in the wild of the 'net | 19:32 | |
zedrdave | now, I won't even try to postulate one of the countless theories on how that might happen, 'cause I'd hate to blow the big ongoing AI riddle, but one thing is sure: you've got brain on one side, you've got symbols on the other.... in the middle is a whole lot of magical pixie dust that we don't quite understand... but it's ostensibly enough to make a clear line between the two. | 19:32 |
zedrdave | if by agents, you mean the MAS kind of agent, they are the least intelligent sort of "intelligence" you could think of. | 19:33 |
fenn | what does MAS stand for? | 19:33 |
zedrdave | multi-agent systems | 19:33 |
kanzure | Hm, I am pretty sure that I understand the anthropic problems with thought and the subjective viewpoint, but this does not mean that we still can't investigate the brain (it's *right there* - just hack into it). Plato genuinely thought of some "other realm" that was inaccessible, yet somehow he could magically access it, so whatever. | 19:33 |
zedrdave | perhaps as a whole, if that (most so-called agents are low-level reactive automata with a fancy name).. | 19:33 |
zedrdave | kanzure: didn't say not to hack into the brain. | 19:34 |
zedrdave | of course we should. and it's certainly gonna help understand a lot. | 19:34 |
fenn | i think plato's "ideals" are cultural information artifacts, nothing to do with the nature of reality | 19:34 |
zedrdave | I just take exception to the argument "it's thought out by the physical brain, so it's physical"... | 19:34 |
zedrdave | fenn: I took Plato because I think it sums up things elegantly... | 19:34 |
fenn | plato didnt have moral/cultural relativism, so he couldnt see | 19:35 |
zedrdave | but pretty much anybody who's been giving it some thoughts across the ages has come down in the same direction... | 19:35 |
zedrdave | not least among them: Kant. | 19:35 |
kanzure | you can't possibly propose the existence of an inaccessible void since that's untestable | 19:35 |
fenn | well you can, but by definition it doesnt matter | 19:35 |
zedrdave | kanzure: "inaccessible void"? | 19:35 |
fenn | see: god, angels on a pin, etc | 19:36 |
zedrdave | c'mon now... what are we, medieval scientists? what's next "aether"? | 19:36 |
kanzure | Plato was simply dead wrong. Tegmark and Egan have made modifications to the platonic realm ideas, saying that "*we are* mathematics", but how does that help progress things? | 19:36 |
kanzure | exactly | 19:36 |
kanzure | that was Plato's idea | 19:36 |
kanzure | that there was some realm of mathematics that he could get ideas and concepts from | 19:36 |
zedrdave | no. that was Plato's antiquated formulation, if that. | 19:36 |
fenn | zedrdave: does information exist outside of the physical? | 19:36 |
kanzure | information physics is fun. | 19:37 |
zedrdave | the idea is much more obviously that of an intangible realm detached from physical reality. | 19:37 |
fenn | kanzure: its much like thermodynamics isnt it? | 19:37 |
kanzure | fenn: yep. Algorithmic information theory (Chaitin), Godel, Kolmogorov, etc. | 19:37 |
zedrdave | fenn: do you need physical reality to ensure that algebra holds together? | 19:38 |
kanzure | Salthe (ecology, information), then Smolin (epitron, that's your que) re: LQG and the connections between gravity and information and so on | 19:38 |
fenn | zedrdave: i dont know, i cant seem to remember what it's like without physical reality | 19:38 |
kanzure | let's experiment | 19:38 |
kanzure | let's blow up every person in the galaxy | 19:38 |
kanzure | and then test algebra | 19:38 |
fenn | and the galaxy too | 19:38 |
kanzure | ooh, good idea? | 19:38 |
fenn | and the rest of the universe | 19:38 |
zedrdave | the fact that you cannot *conceive* of purely abstract thought doesn't mean it doesn't exist. | 19:39 |
zedrdave | that's the trick of it. | 19:39 |
zedrdave | I don't know... try zen koans, maybe. | 19:39 |
kanzure | heh :) | 19:39 |
kanzure | I was looking those up the other day | 19:39 |
kanzure | the trick is that they are still running on your brain, your physical substrate | 19:39 |
zedrdave | methink that's your next best try. short for taking massive doses of ritalin and sitting with Russel's history of philosophy. | 19:40 |
kanzure | zen koans rely very much on incubation insight, mutation, and so on | 19:40 |
kanzure | <-- massive doses of Adderall. | 19:40 |
zedrdave | no shit. I sure couldn't tell. | 19:40 |
kanzure | Maybe I need to learn to hide it. | 19:40 |
zedrdave | zen koans rely very much on the concept of shortcutting that connection you insist on making between neural substrate and abstract thinking. | 19:41 |
kanzure | fenn: http://heybryan.org/mediawiki/index.php/Chat_minutes#2008-03-29 some summaries | 19:41 |
kanzure | you cannot escape into abstraction/simulation | 19:41 |
kanzure | Egan claims that you can, in his book on simulations. I believe it was the one about simulations of people on a chip, and suddenly the chip is destroyed but the simulation is still running | 19:42 |
zedrdave | since the philosophical formal route seems to escape you still, I could only recommend trying the absurdist koan way... | 19:42 |
kanzure | http://heybryan.org/mediawiki/index.php/Neverness/1 has some of my notes on koans | 19:43 |
zedrdave | for the love of shiva, if you don't stop quoting a science fiction author to back up your scientific arguments, I'll soon starting to quote from Ron Hubbard's... | 19:44 |
zedrdave | care to take a personality test? | 19:44 |
kanzure | maybe not right now, I want to get back to work, but if you drop a link I'll take a peek soon | 19:44 |
zedrdave | okay. let me hook up that machine... | 19:44 |
zedrdave | we'll be assessing your thetan count shortly | 19:44 |
kanzure | uhh | 19:45 |
kanzure | let's not do scientology | 19:45 |
zedrdave | hey. you started it with the sci-fi guys. | 19:45 |
kanzure | so what? | 19:45 |
zedrdave | if you can have your sci-fi guy, I can have mine. | 19:45 |
kanzure | yes, but I was citing from Wikipedia on koans | 19:45 |
kanzure | if you think it's wrong, just edit the damn thing | 19:46 |
kanzure | that's the whole point :) | 19:46 |
kanzure | fenn: you alive? | 19:46 |
zedrdave | I wasn't referring to the koan part :P | 19:46 |
zedrdave | your infatuation with Egan quoting... | 19:46 |
kanzure | it's a good line :) And it makes sense. | 19:47 |
zedrdave | so does most of dianetics. | 19:47 |
zedrdave | OK, I take that back. it makes very little sense unless you are on acid. | 19:47 |
kanzure | haha | 19:47 |
zedrdave | but it belongs to the same aisle of your local library... | 19:48 |
zedrdave | tempting as it would be to build an entire set of scientific arguments on the work of a minor sci-fi writer, I don't think that's the best way to go at convincing people. | 19:49 |
zedrdave | otherwise I'd have long developed that Anne Rice angle on Superstring theory... | 19:49 |
kanzure | argument from authority? | 19:50 |
kanzure | get out. | 19:50 |
zedrdave | depends on the authority. | 19:51 |
zedrdave | if the authority is that of said sci fi writer... I may take issue. | 19:51 |
zedrdave | call me biased, I tend to prefer my science from the mouth of scientists. | 19:51 |
zedrdave | similarly, when I am looking for entertainment, I rarely rely on the work of Mssrs. Bohr and Einstein. | 19:52 |
kanzure | I don't think you understand. We are not talking about science any more. We are talking about philosophy. You missed the transition when we started claiming that virtuality does not require a reality. | 19:52 |
zedrdave | sure. but sci-fi is not philosophy either. | 19:52 |
kanzure | mathematics and philosophy have a fine line behind them | 19:52 |
kanzure | but I am not sure we are making any progress any more | 19:53 |
zedrdave | and for that matter, philosophy of science require at least perfect mastery of both fields. | 19:53 |
kanzure | I need to go do something. | 19:53 |
zedrdave | oh no, we aren't. and we wouldn't be anyway. which is why I'd rather discuss the form. :) | 19:53 |
kanzure | There is no matter without form, and no form without matter. | 19:53 |
zedrdave | but yea... good luck and call me when you are about to turn the thing on. :D | 19:53 |
-!- zedrdave [n=zedrdave@dan75-1-87-88-125-171.dsl.club-internet.fr] has left #hplusroadmap [] | 20:05 | |
kanzure | fenn: for STM machine, SpeedEvil in ##electronics suggests a steel wire with a perpendicular electromagnet, which will make the steel wire act like a spring and move the STM tip a bit. | 20:06 |
-!- Enki-2 [n=weechat@c-71-234-190-248.hsd1.ct.comcast.net] has quit [Read error: 104 (Connection reset by peer)] | 21:01 | |
-!- Enki-2 [n=weechat@c-71-234-190-248.hsd1.ct.comcast.net] has joined #hplusroadmap | 21:07 | |
kanzure | Maybe I should apologize for zedrdave. Thought he might do something good, like offer some computational neurosci knowledge. ;) | 21:09 |
fenn | i'm not sure why AI researchers are always so arrogant and abrasive | 21:10 |
fenn | i havent read egan's people on a chip story, but it sounds more like a refute than seriously entertaining the existence of pure abstraction | 21:12 |
kanzure | What I did is I had a young friend of mine go read it for me | 21:12 |
fenn | i was thinking the many-worlds hypothesis could lead somewhere | 21:12 |
kanzure | and then I listened to his report | 21:12 |
kanzure | And he basically said that Egan was arguing that the abstractions could exist without a substrate. I didn't like the sound of that. I don't know if it was a refute (I sure hope so) | 21:13 |
fenn | last AI researcher I had the pleasure of arguing constantly with would also refer me to the beauty of zen koans and such | 21:13 |
fenn | as if it were some sort of proof of his correctness | 21:14 |
kanzure | fenn: I think referring to zen koans is a way of saying "you need to mature" | 21:14 |
kanzure | "and understand that you can't bruteforce it, you can't create entirely new knowledge on the spot" | 21:15 |
kanzure | That's not my interpretation of the koans themselves; just how people point you to them, rather. | 21:15 |
fenn | hmm? you cant create new knowledge? i'm not sure about that - depends on your definitions | 21:16 |
kanzure | I do not know how to put it into words, but it's the same fundamental concept of why we need skdb for example | 21:17 |
kanzure | because you need all of those trials and errors accumulated from 'societalized knowledge' | 21:18 |
fenn | ah, synthesis | 21:18 |
fenn | Creatively or divergently applying prior knowledge and skills to produce a new or original whole. | 21:19 |
kanzure | right | 21:20 |
* fenn strokes his black box. this here's my knowledge synthesizer.. | 21:20 | |
kanzure | you need both recombination and mutation | 21:21 |
kanzure | Sometimes when I am in my little philosophizing sessions (more often with paper and pen, rather than on a noisy computer) I find myself wondering if I am a "mutation harvesting engine" (my brain, at least) | 21:21 |
kanzure | since I do have lots of thoughts that others might consider mutated, defective, varied. | 21:22 |
fenn | music is to frequency envelopes and sequences in a synthesizer as creativity is to ideonomics and ?? in a knowledge synthesizer? | 21:22 |
kanzure | frequency envelopes in a synth? | 21:23 |
kanzure | hm? | 21:23 |
fenn | you know, the algorithms used to decompress music | 21:23 |
kanzure | "as creativity is to ideonomics and ?? in a knowledge synthesizer" I'm tempted to answer with downloading | 21:24 |
fenn | the new digital synthesizers are mostly inverse fourier transforms | 21:24 |
fenn | downloading? | 21:24 |
fenn | maybe i'm off in context-free land, sry | 21:25 |
fenn | ideonomics is combining lists of items that can apply to each other with a function or relation | 21:25 |
kanzure | yes | 21:26 |
kanzure | if frequency envelopes are just music decompression algorithms, then you need a decompression algorithm for creativity, no? | 21:27 |
kanzure | and that's usually "downloading" of previous thoughts/ideas/content/etc | 21:27 |
fenn | i keep thinking of FPGA's and how the step going from logic and symbols to gates and buses is called synthesis | 21:31 |
kanzure | compiling? | 21:33 |
fenn | there is also a reverse process, going from a circuit layout to an abstract principle or idea | 21:33 |
kanzure | decompiling? | 21:33 |
fenn | sure, but in computer programs it's a linear process, whereas in an fpga the signals interact with each other due to physical closeness (emergent effects) | 21:34 |
fenn | they're no longer black boxes | 21:34 |
fenn | a computer program can be black boxes from start to end | 21:34 |
fenn | ok this isnt going anywhere | 21:35 |
fenn | i will put it on my idea shelf | 21:36 |
kanzure | Jef keeps on telling me that it's not just ideonomics, it's mutation too | 21:36 |
fenn | mutation, how? | 21:37 |
kanzure | recombinant DNA / ideonomics and stuff like strategically searching out good value to add to your whole (ideally you are aiming for the whole to be greater than the sum of its parts) | 21:37 |
kanzure | mutation kicks you out of a local rutt when you get stuck | 21:37 |
kanzure | if your entire population is basically the same and they are generally not solving the problem they need to be | 21:37 |
kanzure | a single mutation will change their "playing cards" | 21:37 |
fenn | hm, that's not how i understand ideonomics at all (maybe i dont understand it) | 21:37 |
kanzure | with ideonomics you don't have a DNA to ideas of course | 21:38 |
fenn | but you could | 21:38 |
fenn | with a human ideonome project | 21:38 |
kanzure | right, but you have to root it somewhere | 21:38 |
kanzure | and no matter how you root it, you will cut off your possibility space to some extent | 21:38 |
kanzure | and mutation is supposedly the way to never be completely stuck | 21:39 |
fenn | that's why it's the HUMAN ideonome project | 21:39 |
fenn | ^^ root | 21:39 |
kanzure | do you mean something like http://test.canonizer.com/ ? | 21:39 |
kanzure | (I think the answer is no.) | 21:39 |
fenn | no idea what that's about | 21:39 |
kanzure | Brent's Canonizer project is to get everybody's opinions on every subject | 21:39 |
kanzure | so it's a massive database of opinions and positions of different people | 21:39 |
fenn | who cares | 21:40 |
kanzure | heh | 21:40 |
fenn | i bet you'd be disappointed how few opinions there really are | 21:40 |
fenn | i'd guess about 8^2 = 64 | 21:41 |
fenn | .. anyway.. | 21:41 |
fenn | simulated annealing.. blah | 21:41 |
kanzure | (2008-03-11 12:23:46) Jef: recombination is about exploiting synergies that have already proved workable within particular contexts and are thus likely (but not certain) to work again in a larger context. Mutation is simply for randomly bumping the system out of any ruts it may have settled into. | 21:42 |
kanzure | and then there's stuff like the Red Queen's Race | 21:43 |
kanzure | a.k.a that you better hope the whole is greater than the sum of the parts if you intend to go anywhere | 21:43 |
fenn | are synergies then used as a heuristic in the search? | 21:45 |
kanzure | heuristics are important for search in the social knowledge domain and recombination etc. etc., but I would think to throw 'synergies' over to the "whole is greater than the sum of its parts" idea, | 21:46 |
kanzure | because otherwise you're casting synergies as a type of astrology | 21:46 |
kanzure | "Hm. The moons are aligned today! All is in synergy." | 21:46 |
fenn | i meant the specific theme of, say, lathe and mill | 21:47 |
fenn | or, teepee and horse | 21:47 |
-!- Enki-2 [n=weechat@c-71-234-190-248.hsd1.ct.comcast.net] has quit [Read error: 104 (Connection reset by peer)] | 21:47 | |
fenn | i cant even think of any good synergies | 21:47 |
kanzure | how about ecological synergies? | 21:48 |
kanzure | the flowers/bees? | 21:48 |
fenn | Once person B sits on the shoulders of person A, they are more than tall enough to reach the apple. | 21:51 |
kanzure | "Stand on the shoulders of giants" | 21:51 |
kanzure | But are they really giants? I once asked Jef (or maybe Tony) if it was just a tower of ants. | 21:51 |
fenn | yeah, ants | 21:51 |
fenn | some ants are bigger or smaller | 21:52 |
kanzure | but ants can do cool things collectively | 21:52 |
kanzure | go knock over an ant hill and watch | 21:52 |
-!- Enki-2 [n=weechat@c-71-234-190-248.hsd1.ct.comcast.net] has joined #hplusroadmap | 21:54 | |
fenn | http://www.rwgrayprojects.com/synergetics/s01/p0100.html | 21:55 |
kanzure | fenn: maybe the ?? is "intention" ? | 21:56 |
fenn | i think it was a poor analogy to begin with | 21:57 |
kanzure | fenn: interesting link | 21:57 |
fenn | a synthesizer is just a circuit that combines multiple musical "voices" | 21:57 |
fenn | like, you can modulate one oscillator with another, or with noise, or direct input | 21:58 |
fenn | in ideonomics you can modulate one idea with another | 21:59 |
fenn | but (like music) it doesnt mean anything unless someone is listening | 22:00 |
fenn | synergetics is a huge impenetrable book that i've always wanted to read | 22:00 |
kanzure | I wanna get back to work. | 22:01 |
fenn | it's like a fat encyclopedia sized book | 22:01 |
kanzure | " I like what bronowski said on free will and determinism; they are misunderstandings of history. At any point in time we move forward into an area that is generally known but whose boundaries are uncertain in some calculable way." | 22:10 |
kanzure | from #biology | 22:10 |
kanzure | http://scholarly-societies.org/ | 23:26 |
kanzure | a list of just what the name says. | 23:26 |
Generated by irclog2html.py 2.15.0.dev0 by Marius Gedminas - find it at mg.pov.lt!