2008-04-23.log

--- Day changed Wed Apr 23 2008
kanzurehttp://heybryan.org/mediawiki/index.php/2008-04-22 <-- Works. 00:00
fennTrue is capitalized, no endwhile00:56
kanzurehuh01:02
kanzureokay01:02
fenni have a custom css that blanks out all page coloring, so i'm not exactly bouncing up and down for inline syntax hilighting01:03
kanzureit was more something for me to do :)01:03
fenni dont really understand this line: yaml.load("!!python/object:__main__.Hero")01:04
fennwhat is Hero supposed to be?01:04
kanzurea class01:04
fennshouldnt that come from a plugin?01:04
kanzureassume that it was defined before the line starting with 'try'01:04
kanzureyes01:04
kanzurebut this was from our logs01:04
kanzurealso, 01:04
kanzureI don't know how to specify a class otherwise01:04
kanzureas in, what does it need to be?01:04
kanzure!!python/object:huh?01:04
kanzureI mean, it has to be in __main__, right? else it's defined as a python-provided type, right?01:05
kanzureI don't know01:05
fennno, you just say the name of the class01:05
fennif its in your namespace it will work01:05
kanzurepython has namespaces?01:05
fennif you imported the class from a module it's module.Class01:05
kanzurehuh01:05
kanzureconvenient01:05
fenngood manners in python is not to 'pollute' the main namespace with imported modules01:06
fenngonna have to actually read this yaml spec i think :)01:12
kanzureyou didn't read it?01:13
kanzureheh01:13
kanzureikeljfald;fj;akfjladkjfa01:13
kanzureslacker01:13
* kanzure goes to sleep (<--- is a slacker too)01:13
fenni'm slow01:13
kanzureI used to be able to program while in school, on paper, but these days I rely on external libraries to such an extent that if I don't have the specs and a few search engines opened, ehh01:13
fennprogramming on paper is not very useful01:14
fenni mean, once you've got the library figured out the job is 90% done01:14
kanzureoh?01:14
kanzurehttp://www.multicians.org/andre.html01:14
fennso, "polymath" doesnt have anything to do with math01:14
fennit just means, "not stupid" i guess01:15
kanzureyou may be interested in http://eugen.leitl.org/polymath.mbox01:16
kanzurean old mailing list01:16
kanzureCarl Feynman used to post, among others01:16
fennof course the last 10% takes another 90% of effort01:17
fennpareto principle01:18
fennoh, here we go http://en.wikipedia.org/wiki/Ninety-ninety_rule01:19
fennusenet should have had a moderation system like slashdot01:22
kanzure'Members are: me, Robin, James Rogers, Tim May, Peter McCluskey, DDFR,01:22
kanzureGreg Burch, Damien B., Perry, Hal, Richard Schroeppel, Ted Kaehler,01:22
kanzureEugene Leitl, Carl F., Amara, Romana, Anders, Curt Adams, and Ralph01:22
kanzureMerkle.  (plus Brin.)'01:22
-!- marainein [n=marainei@220-253-110-236.VIC.netspace.net.au] has joined #hplusroadmap01:22
kanzureDavid Brin, Carl Feynman, Eugene Leitl, Damien Sullivan, Anders Sandberg, all on the same list?01:22
kanzureThat's a deal.01:22
* kanzure needs sleep01:24
fenngnite01:24
-!- kanzure [n=bryan@cpe-70-113-54-112.austin.res.rr.com] has quit ["Leaving."]01:24
-!- marainein [n=marainei@220-253-110-236.VIC.netspace.net.au] has quit ["Ex-Chat"]04:54
-!- marainein [n=marainei@220-253-110-236.VIC.netspace.net.au] has joined #hplusroadmap09:23
-!- marainein [n=marainei@220-253-110-236.VIC.netspace.net.au] has quit ["Ex-Chat"]10:12
-!- kanzure [n=bryan@cpe-70-113-54-112.austin.res.rr.com] has joined #hplusroadmap17:46
-!- Splicer [n=p@h126n3c1o261.bredband.skanova.com] has joined #hplusroadmap18:30
Splicerkanzure: You are setting up a videocast?18:32
kanzureSplicer: Are you John Fields?18:46
Splicerno18:47
Splicergot a webcam?18:47
kanzureNot yet, but that's cheap.18:48
kanzureYes, I'm going to be doing some videos, it's something that I have been putting off. I can do much more in less time via vid.18:48
Spliceryou´re becoming one of my heroes18:48
Splicerwho is John Fields?18:49
kanzureHe emailed me about his podcast for biohacking, so I mentioned my vid casting ideas, and told him I'd be willing to host the podcast.18:49
Splicer.. cool18:49
Spliceri'm the guy on the whois record of my site18:50
kanzureI haven't bothered to check.18:51
kanzureSometimes it's better to *not* know who you are, so that way you have some leeway.18:52
Splicerthus Splicer18:52
kanzureHm.18:52
kanzurehttp://www.med.wayne.edu/news_media/streamingmedia/curriculum/GradPrograms/PYC_7010-IBS_7050/index.asp 18:52
kanzureMolecular neuropsychopharmacology vids18:53
Splicerfuck.. they are vids... i didn´t see18:53
Splicerthank you ;)18:55
Spliceri was listening to the berkley genetics podcasts in the background18:55
kanzureSuperkuh: http://linux.die.net/man/1/mmsrip -- Opera -> select all links -> copy -> vim -> replace all beginnings of the lines with wget (or do a quick awk script?) -> execute this script -> extract all MMS:// URIs -> use mmsrip. Source package: http://werewolf.freshrpms.net/rpm/mmsrip/19:08
kanzureoops, I was telling him that in #biology19:08
SplicerI added the neurpsychopharmacology vids to biopunk with a nudge to you19:21
Splicerhttp://www.biopunk.org/post63.html#p6319:21
kanzureHold on. I'll be writing a script to help rip the vids.19:22
Splicer;)19:22
SplicerI think the link from the site still has to be to their streamed version though.19:24
kanzureSure, but.19:26
Spliceryeah.. I know19:27
kanzureI am ripping them to my 500 GB data server. 19:27
Splicercan I ask you something... do you remember everything you read?19:27
kanzureThat's my own personal secret.19:28
Splicerhehe19:28
kanzureSplicer: Heh, as I wrote that I realized I know why your name is not John.19:28
Splicerwhy is that?19:29
kanzureI was distracted when I asked the question. Sorry 'bout that.19:29
Splicerah.. I signed some mail whit my name i think19:30
kanzureAnd the jabber handle.19:30
Splicerthat's IM right?19:31
Spliceryeah, sorry19:31
Spliceri really enjoy lectures on video... they are one of the few things that are better on video than irl19:32
kanzureyes19:37
kanzureThe server is a piece of crap. It doesn't even have a recent libc.19:43
Splicerfor #biology?19:45
Splicertime to sleep19:47
Splicercu19:47
-!- Splicer [n=p@h126n3c1o261.bredband.skanova.com] has quit []19:48
kanzurehttp://heybryan.org/docs/mol-neuropsychopharm.txt19:48
-!- Aulere [n=dragon_d@131.229.176.252] has joined #hplusroadmap21:07
kanzureHi Aulere.21:10
kanzurefenn: Check the three-forwarded emails I sent to hplusroadmap. They describe where I'm at tonight - run levels, and then how to get agx-get plugins to interface with the user for giving them options, without having to enforce too much rigidity in the future possibilities of skdb metadata files.21:11
Aulerehi kanzure21:13
kanzureWhat's up?21:13
Aulerewriting a research papaer21:13
Aulereer paper21:13
kanzureheh :)21:13
kanzureOn what?21:13
Aulerethe neuroactivational differences between regular and irregular verb tenses21:14
Aulereone study is using MEG21:14
Aulereand I had the random thought, spurred by previous discussion here,21:14
Aulerehmm, if MEG were in theory used on computers, would it be able to tell what each computer chip did?21:15
Aulereand each connection and whatnot.21:15
Aulere(so I'm not a hardware engineer ;-) )21:15
Aulerewhadda you think?21:15
kanzureSorry, was writing an email to hplusroadmap.21:22
kanzureAulere: First, if MEG can go down to the resolution of nanometers and so on for the resolution of the circuits that computers are using, that's a big step. Second, you would be going about it all wrong, because you could see what's happening to the signals and so on, but not the abstract design of the processor ... you see?21:23
AulereYes, that's what I was thinking too, which makes me wonder at the validity of using the method at all21:23
Aulerefor the brain.21:24
kanzureYES!21:24
kanzurebut it's actually still useful21:24
kanzurethere's a pretty good chance that if some general region is activated, heh21:24
kanzurein a computer circuit it still works the same way - wouldn't you be surprised if a chip programmed to do scientific calculations suddenly activated the sound card? ;)21:24
kanzureSame way with the human brain thanks to functional specificity.21:25
Aulereheh 21:25
kanzureBut this doesn't mean mind reading. ;) Far from it.21:25
Aulereyeah.21:26
kanzureTo be able to get the design of the processor from MEG readings, you would have to record the processor going through *ALL* possible states and then correlate everything into some very large logic matrix. Since the chips these days are 64 bit, that means you have at least 2^64 possible inputs. That's physically impossible to reverse engineer.21:26
kanzure(luckily we can visually analyze the gates with microscopes, instead of having to rely on the electrical signal data ... phew)21:26
Aulereyes, and not just the design, but the function too.21:27
kanzureThe function is known via society - i.e., people telling us how to program it, whether by them writing books or physically hitting us over the head when we make a typo re: their instruction set architecture (ISA). :)21:28
Aulerelol21:28
Aulereresearchers are trying to reverse-engineer the function of the brain using MEG21:29
kanzure100 billion neurons with over a few trillion connections, not to mention the routes possible between all of these connections. Yeah, right. hehe21:29
kanzureThis is where self-replicating machines would come in, for example21:30
Aulereheh21:30
kanzureYou could either do an exponential growth program where you try out possible experiments or something, or you could do exponential growth and try to bruteforce engineer 'intelligence'/brain from the ground up, but both of these are suspect and we *must* significantly constrain possibility space, especially in the case of wetware brain experiments.21:30
kanzurePlus, Godel would argue that no brain can understand itself. But the neuroscientists should understand this already.21:31
AulereYes, good ole Godel...21:31
kanzurehttp://www.miskatonic.org/godel.html21:31
AulereSo the study is specifically involving spatiotemporal brain mapping the magnetic changes on the order of ms21:31
kanzureWith respect to what ?21:32
Aulere?21:32
kanzureMaybe I am misunderstanding.21:33
kanzureSpatiotemporal means "xyz + time" dimensional components. Unless you mean there's a region of the brain for spatiotemporal information processing?21:33
kanzureand that they are studying this region?21:33
Aulerexyz + time21:34
kanzureThey are doing this over the whole brain ?21:34
Aulerepretty much - following the cortical stream of activation from the occipital lobe onward after viewing verbs.21:35
Aulerewell, after viewing and then deciding the past tense.21:35
kanzureThat would be even more interesting once we get the Allen Institute to complete the human brain atlas re: gene expression throughout the brain. Then we can target particular receptors with very, very specific drugs.21:35
Aulereinteresting. 21:36
kanzurethe Allen Institute has done such for the mouse, already21:36
kanzurehttp://brain-map.org/ I think21:36
Aulerethe current application is to validate or invalidate connectionist models21:36
kanzure'The Anatomic Gene Expression Atlas (AGEA) is a data-driven, interactive three-dimensional atlas of the adult mouse brain based on Allen Brain Atlas ISH gene expression images. AGEA characterizes the organization of the mouse brain as derived exclusively from gene expression data without reference to classical neuroanatomy. AGEA, which is based on approximately 4,000 coronal gene sets, allows anatomic specification and browsing ba21:36
Aulereinteresting21:37
Aulereunfortunately as of yet, mice are non-verbal. 21:39
Aulere:)21:39
kanzureDouglas Adams would disagree.21:39
Aulereheh21:39
kanzurehttp://hitchhikers.wikia.com/wiki/Mice21:39
kanzureMice are merely the protrusion into our dimension of hyper-intelligent pan-dimension beings who, unbeknownst to the human race, are the most intelligent species on the planet Earth. They spent a lot of their time in laboratories running complex experiments on man.21:40
kanzureAt the outset, they were so fed up with the constant bickering about the meaning of life which used to interrupt their favourite game of Brockian Ultra Cricket, that they decided to sit down and solve their problems once and for all.21:40
kanzureThey were the creators of Deep Thought, a stupendous super computer the size of a small city, to tell them the Answer to Life, The Universe and Everything. When seven and a half million years later it was realised they didn't know they question to the answer they'd been given, a second computer, of such infinite and subtle complexity that life itself would form part of it's operational matrix, was created to work out the Ultimate 21:40
Aulereoh yes21:40
Aulereanswer 21:40
Aulereyes, 4221:40
Aulere(that first yes was to the former paragraph, the last one to the latter)21:41
AulereI lost my Adams virginity in an elevator shaft in Indiana with about 8 other friends21:48
AulereWe would stay up after midnight taking turns reading ;-)21:48
kanzureWere you stuck?21:48
Aulerenah21:48
AulereIt was an elevator shaft, but no elevator was ever put in it.21:48
kanzureUsually losing your virginity and shaft don't have an elevator in common with the two, but I digress.21:49
AulereLOL21:49
AulereOh yes, so literal, I forgot21:50
kanzurenah, just joking :)21:50
Aulere:)21:50
kanzureit's really just my way of wasting time and not making some rather important design decisions for agx-get21:50
kanzureor not. dunno. fenn, you around?21:51
Aulerewell, I'd best get back to my paper. there are several other studies to incorporate.21:52
kanzureHTTP archives are back up and running. :-)22:37
kanzurehttp://heybryan.org/pipermail/hplusroadmap/2008-April/thread.html22:37
kanzurehttp://heybryan.org/pipermail/hplusroadmap/22:37
kanzureproblem:23:15
kanzureuser interface can be updated via downloading new plugins which contain new classes and new methods for dealing23:16
kanzurebut23:16
kanzurethese plugins are ran when 23:16
kanzureah, only when there's actually that datatype23:16
kanzuretherefore there's no conflicts23:16
* kanzure is stupid23:16
kanzuredidn't I know this last night?23:18
kanzureNow what? 23:19
kanzureagx-get should be able to fetch an updated package of all metadata files23:19
kanzurethen it should be able to search through them via tagging or something23:19
kanzureactually, search is not necessary per-se - definitely not an original feature (but tagging would be great to implement right off, of course)23:19
kanzureanyway, it should have an install function23:19
kanzurewhich would grab the repository and get all of the data - 23:20
kanzureor not all of it, really, just the parts that the user specifies either through the cli parameters or through user input with agx-get asking the user23:20
kanzureso any parameter that agx-get does not natively understand should be assumed to be understood by the program that parses the (skdb-file)'s metadata file.23:20
kanzureanything that is not understood by agx-get would be passed as a parameter to that program,23:21
kanzureand that program will have some internal logic to deal with it23:21
kanzurebut the output of that program that is devoted to interpreting that type of metadata file?23:21
kanzurethe output shouldn't be simple print statements23:21
kanzureit should be filtered back through agx-get, yes?23:21
kanzurethis way agx-get can be updated to different versions like a cli, a gui, whatever23:21
kanzureI was originally thinking of there being a limited number of types of possible interactions with the user23:22
kanzureI think there would *have* to be if agx-get is supposed to be the layer in between the specific program interpreting the metadata file and the user ...23:23
kanzureso I think HTML forms might be a good example spec to work off of23:23
kanzurethere has to be input boxes, select boxes, radio buttons (choose from a variety of options, perhaps a few simultaneously), large textboxes, input methods where you can open up a new shell or a new instance of vim or your other favorite text editor, ...23:23
kanzureright?23:24
kanzureand then cli v. gui implementation can be left up to agx-get23:24
kanzurehttp://en.wikipedia.org/wiki/Adapter_pattern23:27
kanzurehttp://en.wikipedia.org/wiki/Facade_pattern23:47
fenni didnt get any email - am i on the mailing list?23:56
kanzureoh23:58
kanzurehttp://biohack.sf.net/ is the signup form23:58
kanzureignore lots of what I said tonight, there are a few ideas that I just stumbled upon in the background here23:59
kanzureneed to code an interface layer for the plugins to do their thing, so I'm thinking that there are two options: (these plugins are for agx-get, btw, for interpreting the yaml when agx-get doesn't have the type on hand)23:59
kanzure1) yaml to be sent to specify what sort of user io has to be filled in23:59
kanzure2) a generalized gui/cli interface layer23:59

Generated by irclog2html.py 2.15.0.dev0 by Marius Gedminas - find it at mg.pov.lt!