2008-11-04.log

--- Day changed Tue Nov 04 2008
willPow3rit didn't work for me, no00:07
willPow3ri think the graph displays in that window, but i can't make it larger00:08
kanzure_Making it larger? Bah.00:13
kanzure_Just switch to ASCII instead of HTML.00:13
willPow3ri can't figure it out. where's the f'ing manual?00:19
kanzure_There is none yet. What is there to figure out?00:19
kanzure_Btw, in the end I'll probably do a webcast/video.00:19
kanzure_screencast/video.00:19
willPow3rthe MAKE function. for building CFGs. it doesn't do anything, just prints out "Match in 12681.dot with transfer"  but that number changes00:20
willPow3runless that *is* the function00:20
kanzure_Notice the sorry not implemented message.00:21
willPow3roh right00:21
kanzure_So, what this is going to do is convert the given graph that you have drawn, into possible designs.00:21
kanzure_So, a component that "converts" might be a particular engine.00:21
willPow3rthats a great fucking idea!00:21
willPow3rand it uses a database of parts/processes?00:22
kanzure_SKDB? yes, it is.00:22
kanzure_Yes.00:22
willPow3ryou're an industrial engineer right?00:22
willPow3rmakes perfect sense.00:22
kanzure_So this is also what I'm doing for synbioss/biobricks00:22
kanzure_if xp_prg ever finishes the work I gave him over a month ago.00:23
willPow3rbut you could feasibly datamine the entinre intarweb for useable parts/processes00:23
kanzure_no shit ;-)00:23
kanzure_Also the patent database.00:23
kanzure_"Reverse engineering of the entire patent database"00:23
willPow3rso skdb != serial killer database00:23
kanzure_plus seekda.com (maybe) to search for ..00:24
kanzure_uh00:24
kanzure_right.00:24
willPow3r;)00:24
kanzure_http://heybryan.org/shots/2008-11_02_supermetal_2.png <- What I made yesterday.00:24
kanzure_Today the doc for ADL was going over some Hod Lipson videos. 00:25
kanzure_He showed the evolved robotics stuff.00:25
willPow3rThe requested URL /shots/2008-11_02_supermetal_2.png was not found on this server.00:25
kanzure_Hod apparently had a simulator for showing different designs of bots, that moved around on the screen00:25
kanzure_http://heybryan.org/shots/ <- look for 'supermetal' on this page. Choose the second one.00:25
kanzure_Anyway, the doc commented that "this was the work of like three or four grad students over their time at his lab - many years"00:26
kanzure_and he pointed to the typical checkerboard OpenGL terrain00:26
kanzure_plus a cube on it and kinda twitching00:26
kanzure_Now, I don't mean to brag, or be an ass or anything,00:26
kanzure_but that's like tutorial 2 of NeHe00:26
kanzure_So, given that 'supermetal 2008' work that I did yesterday,00:27
kanzure_plus some twitching, I could probably mimic the same work in a relatively short amount of time00:27
willPow3rit doesn't seem that complicated really00:27
kanzure_Hod's group was then printing out the parts to the automatically generated bots via a 3D printer00:27
kanzure_plus some provided motors or something (which is where I personally get stuck)00:28
* kanzure_ fetches the video00:28
kanzure_oh, it's his ted talk00:28
willPow3rlink00:28
kanzure_Anyway, it's not a servo, but rather some sort of pneumatic system in the formfactor of a servo00:28
* kanzure_ gets up to move to the other machine00:28
kanzure-http://www.ted.com/talks/view/id/16500:29
kanzure-http://ccsl.mae.cornell.edu/research/golem/golem480x240.wmv00:29
kanzure_Anyway, I'm _pretty sure_ it's simple. Maybe I'm not seeing something here.00:29
kanzure_I don't understand the actuator for the tubular form factor or whatever. It somehow "crawls" but it's made out of a triangle. Not a wheel or anything. How is this possible?00:30
willPow3rlooks like how an inchworm moves00:31
kanzure_An inchworm has muscles that 'stretch' via a molecular mechanism. Doesn't count.00:31
kanzure_I guess it's like rotating a screw. Rotate to the right to go away, to the left to go inward. 00:34
willPow3rthis is in the cornell vid?00:36
kanzure_It's in both I think.00:37
kanzure_One shows more of the simulations, the ted talk shows the physical implementations.00:37
willPow3rfor all of 5 seconds00:37
kanzure_heh'00:37
xp_prgkanzure_ it is coming along nicely as you have already seen :>00:37
kanzure_Where's the rest of it?00:37
xp_prgstill working on it, been quite busy00:37
willPow3rthey should have ran more evolutionary simulations than just "which ever bots move fastest get to move on"00:42
kanzure_I think they did.00:43
kanzure_Not sure. I would have hoped so.00:43
willPow3r"the only rule was that the machines that move faster will replace the ones that move slower"00:44
kanzure_That's the selection portion of the GA.00:44
willPow3rits hard for me to believe that this is state of the art robotics research.00:45
kanzure_Go to Japan for that. ;-)00:45
willPow3rits far too pragmatic.00:45
kanzure_But that's more "trying to make up AI".00:45
kanzure_Hm?00:45
kanzure_What do you mean?00:46
willPow3rthese robots could be designed with legos.00:46
kanzure_This is more about evolutionary algorithms than it is about robotics AFAIK.00:46
kanzure_But yeah, it's not "new materials" or new shapes and designs, that's for sure.00:46
kanzure_that's something fixable methinks00:46
willPow3rnanotech should be able to bridge the gap00:47
kanzure_The problem with letting it randomly generate shapes and new parts and so on, like in Supermetal, is that the possibility space explodes00:47
kanzure_Uh?00:47
kanzure_What gap?00:47
willPow3rthe servo/actuator model is far to coarse to be used seriously. we need to use something that has many, many more degrees of freedom00:48
willPow3rbut this is definitely a good starting place i suppose00:48
willPow3rthey're using an evolutionary process to weed out bad designs00:49
willPow3rbut the designs aren't elaborate enough to seriously be subjected to those kind of selective forces00:50
kanzure_On the contrary.00:50
kanzure_http://heybryan.org/~bbishop/docs/repo/ Check out this and the subdirectories. Anything in PNG too.00:50
kanzure_http://heybryan.org/~bbishop/docs/repo/CDD/00:50
kanzure_http://heybryan.org/~bbishop/docs/repo/CDDs/00:50
kanzure_one of those ..00:50
willPow3r45 KBps?00:51
kanzure_Hrm?00:52
kanzure_Anyway, there's giant graphs, lots of parts and so on.00:52
willPow3ryour upload speed.00:52
kanzure_hm.00:52
willPow3rwhat opens .cdd files?00:53
kanzure_Something commercial. conceptdraw or something. Avoid it.00:53
kanzure_Try the PNG. I took a few screenshots/exports when I was running conceptdraw.00:53
willPow3rall the files in CDD are .cdd...00:54
willPow3ri'll check png then00:54
kanzure_Sorry. Try CDDs. I don't know where I put it.00:54
kanzure_yeah.00:54
willPow3rtime for bfast. bbl00:55
willPow3rit would be funny if somebody pulled a bel-air during a ted talk01:29
kanzure_A what?01:29
willPow3ra bel-air.01:29
willPow3rhttp://encyclopediadramatica.com/Bel-Air01:30
willPow3rexcept remove all references to 4chan and /b/01:31
kanzure_So yeah, this isn't really working out for me.02:01
kanzure_There are about 46 different types of nodes, so at least *46 more graphs per depth into the trees that you go for possible graphs, so at 3 nodes it's already 46^3. 02:02
kanzure_I generated a million rules that could be used to transform the function structure (what the user provides) to possible configurations of parts and items, but.02:03
kanzure_But if the idea is to apply a million rules to each applied million rules, and go on for a few depths, that's many, many trillions of results..02:03
kanzure_I need a simple way of extracting interesting specimens for now.02:03
kanzure_The problem is that if the user chooses something ridiculously exotic, as if the rules are not evenly distributed for all 46 options, then it's all shot to hell, if you're doing a small 150-rulecount system.02:04
willPow3ruse a weighted rule system perhaps?02:05
kanzure_*if the case is that the rules are not evenly distributed for all 46 options (i.e., if the generated ruleset just so happens to be centering around one section)02:05
kanzure_Hrm?02:05
kanzure_I need a balance between the number of possible options in the ontology versus expected interesting stuff turning up, for now.02:06
kanzure_So maybe only 10 types of nodes, which would allow depth, but there's very little breadth here. You wouldn't get a weird "gasoline line connected to a parachute". :-/02:07
kanzure_Meh. Maybe it's better that there's too many rules to explore. 02:09
kanzure_Aha.02:16
kanzure_The idea was always that the users would build up a library of interesting rules for their own personal use. 02:16
kanzure_But since the newbie will not have a set of rules, just let it be random. 02:16
kanzure_If there turns out to be a rule of thumb for figuring out interesting rules, then I'm sure they'll start putting those into the db and having their fun02:16
kanzure_Ok. Food, star wars, then some programming.02:17
wrldpchttp://d.scribd.com/docs/11eq9u1ilefgk830irm6.pdf05:06
wrldpc^TSIN, Kurzweil05:06
wrldpcDidn't know if you guys had this in PDF yet.05:07
wrldpcIf anyone has Ending Aging in PDF holler or scribd it.05:07
drazakI wonder how much it costs to double barrel sequence a whole human genome05:14
willPow3ri accidentally a whole human genome05:43
willPow3rthe *whole* thing06:00
splicer_you ate a whole human genome?06:12
willPow3rno, i accidentally it06:13
willPow3ri dont know if i should be worried or what06:14
splicer_just on purpose it next time, you should be fine06:16
willPow3rnot sure, because if you it, then you're supposed to or else the consequences 06:17
splicer_there is always the risk of that of course. and we don't want that.06:18
willPow3rthats what it says in the. and you're not supposed to everything you, as they say.06:20
splicer_ah.. they, the ones doing both it and that.06:22
ybithttp://www.youtube.com/watch?v=Li5nMsXg1Lk "pinky and the brain on parts of the brain"06:24
wrldpclol07:06
kanzure_wtf?13:31
kanzure_"Out of socket memory"13:31
fennanother buzzword to add to your list: http://en.wikipedia.org/wiki/Knowledge-based_engineering14:12
UtopiahGHMLas opposed to emotion based engineering14:12
fennRTFA14:15
fennthis makes me sad: http://commons.wikimedia.org/wiki/Commons:Screenshots14:46
UtopiahGHMLtry http://www.fractaluniverse.org/ instead14:47
fennemotion-based physics :(14:48
fenneveryone knows the universe is patterned after the flying spaghetti monster14:52
fennheh first time i've come across a wingnut championing a "flat sun" theory14:59
facefacekanzure_, have you seen this wrt. your journal dump? http://www.myjournals.org15:09
faceface"Wiki for *executable* English" over databases?15:42
UtopiahGHML?15:44
UtopiahGHMLanother rewrite of SQL?15:44
fenngellish?15:50
facefaceaparently information is 'in google', but I didn't query yet15:56
facefacehttp://www.nature.com/news/2008/080903/full/455022a.html15:56
facefacecan you see that link?15:56
facefacethere reference is in the comments part15:57
facefaceBig data: Wikiomics - http://www.nature.com/news/2008/080903/full/455022a.html15:57
nshLow input, high throughput, no output16:15
facefacehave you guys seen this?17:22
facefaceIn Google we trust17:22
facefaceThis paper considers information safety and accuracy in the digital age using Google as an entry point.17:22
facefacehttp://www.firstmonday.org/issues/issue11_4/shaker/17:23
facefaceits a really interesting paper.17:23
UtopiahGHMLfirstmonday usually have interesting questions17:25
UtopiahGHMLwithout reading it Im pretty sure the conclusion will be to, as always, cross check information and have multiple sources of information and the danger of having some information blackhole like Google.17:26
UtopiahGHML(but they don't provide DOI :(17:29
facefaceUtopiahGHML, don't know where its published17:38
facefaceUtopiahGHML, the results are a bit more nuanced17:38
facefacebut its interesting to read what they write about trust.17:38
facefacewhat is that registry that SMW can be 'published' to?17:39
kanzure__Screenshots are subject to the copyright of the displayed program? waaaht??17:44
kanzure__faceface: Are those for open access journals only or what?17:45
facefacewhat?17:45
kanzure__myjournals.org17:45
facefaceoh... don't know the deets17:45
facefaceI think its for any journal... like a login aggrigator...17:45
facefaceperhaps17:45
kanzure__They are asking users for usernames and passwords?17:46
facefaceanyone know how to use Chanserv17:46
kanzure__Good god, people fall for this shit?17:46
* kanzure__ makes a clone ASAP17:46
facefacehehehe... did you get that email from me asking you to join some company?17:46
kanzure__Maybe. Let me check.17:46
facefaceI gave them my gmail addressbook and they went to town17:46
facefacenevermind if you did!17:46
kanzure__Recent email?17:46
facefacesome time earlier this year17:46
kanzure__ThirdGen?17:47
kanzure__Anyway.17:48
facefaceyo dudes... http://www.google.com/base/a/dan.bolser/1084695/D1239101131598809032618:11
facefacefeedback ... good bad or otherwise...18:11
facefaceits a 'journal club' of this... http://biology.plosjournals.org/perlserv/?request=get-document&doi=10.1371/journal.pbio.0060175&ct=118:12
facefacel8r18:12
bkeroIsn't biobricks a wiki?18:13
bkeroSorry, not biobricks, http://partsregistry.org/Main_Page18:13
facefacecrap... can you even see that18:13
bkeroYes, I have the PDF loaded18:14
kanzure__Biobricks is a wiki, but there's a database behind it that they like to hide.18:14
kanzure__I mean, not just the mediawiki db.18:14
bkeroHave you seen the parts registry?18:18
kanzure__Me?18:18
kanzure__Are you kidding?18:18
bkeroHeh18:18
kanzure__http://heybryan.org/books/biobricks/18:18
kanzure__http://heybryan.org/graph/18:18
bkeroMoreover talking to faceface18:18
kanzure__http://heybryan.org/graph/synbioss/18:18
kanzure__Are you kidding?18:18
kanzure__Dan works in bioinformatics18:18
bkeroOk18:19
kanzure__:)18:19
kanzure__xp_prg: Do you have it done yet?18:24
xp_prgkanzure still working on it, want me send you what I have so far?18:51
xp_prgto send you18:51
kanzure__yes?18:53
xp_prgok18:53
xp_prgsent it19:09
UtopiahGHMLanybody used DBpedia? Id like to have a list of the US stats with name, area and coordinates20:24
UtopiahGHMLactually maybe there is a mashup to directly query DBpedia, process it with R and display a nice graph all online, no?20:27
xp_prgkanzure__ did you see it?20:58
kanzure_xp_prg: Does this fully (re)generate the SBML?21:08
kanzure_ScienceDirect just called me.21:08
* kanzure_ grins21:08
xp_prgyes it does21:11
willPow3rkanzure_, prank call21:11
xp_prgbut it is not complete21:11
xp_prgits about 70% there21:11
UtopiahGHML(failure, Freebase doesn't have the area, I have the list of states and other useful info but not that one :\21:20
kanzure_if it's not complete then how would you say it fully regenerates it?21:27
kanzure-https://www323.livemeeting.com/rm/elsevier/webJoin?id=UsabilityTest&pw=presenter&role=present21:31
kanzure-http://concepts.elsevier.com/sciencedirect/2009-1/UsabilityTest_2009_1.html21:32
kanzure-http://concepts.elsevier.com/sdrevamp21:33
kanzure_These people are not secure.21:40
kanzure_user: demo / pass: feed8ack21:40
kanzure_gasp! googleable usernames and passwords!21:40
willPow3rit certainly isn't palin's yahoo email21:40
kanzure-http://concepts.elsevier.com/sciencedirect/2009-1/ <-- teh hell?21:41
kanzure-What are they doing here?21:42
kanzure-Hrm.21:42
willPow3rlooks like getting ready for next year21:43
kanzure-hm.21:44
kanzure-must corrupt their md5 scheming system21:44
kanzure-muuust21:44
kanzure_http://oxzaps6.elsevier.co.uk maps to http://concepts.elsevier.com/21:48
kanzure-xpharm-oxzweb5.elsevier.co.uk/21:50
kanzure-https://secure.elsevier.co.uk/ane-server/login21:53
kanzure-shratz and skiddaw..21:54
kanzure-http://www2.elsevier.co.uk/~tony/rss_demo/ sample.rdf21:56
kanzure-rse13.ritense.com21:59
kanzure-http://rse12.rittense.com/22:00
xp_prg4kanzure?22:30
kanzure_Yes?22:34
xp_prg4so did the email help you?22:34
kanzure_Did you send another one?22:35
xp_prg4no22:35
xp_prg4did the one I sent you help you?22:35
kanzure_No. I asked you what's remaining.22:36
kanzure_You said 30% and didn't tell me.22:36
xp_prg4what's remaining is to add math rate functions and prompting the user for input etc...22:36
kanzure_Which math rate functions?22:39
xp_prg4you specify rates for a reaction etc...22:41
xp_prg4just compare the testmodel.xml and the one I gave you that tells what is missing22:41
kanzure_User input supplies the rates?22:42
xp_prg4yup22:43
xp_prg4so does that help you kanzure_ or not?23:01
kanzure_xp_prg4: I don't see "complex specifics" support.23:24
kanzure_I also don't see "binding specifics".23:24
kanzure_Nor anything about protein inducers23:24
kanzure_or "bound complexes" for each of the inducers23:25
kanzure-http://voiced.device.mst.edu/groups/voiced/23:35
kanzure_ /whois kanzure__23:44
kanzure_I wonder how drug commercials are made these days. Does the advertizing firm go to the scientists that published the papers and ask, "Okay, great, now what's the best way to fuck this up in the ad?"23:51
kanzure_"Using a powerful ingredient, this product strengthens bone marrows by __magic elves__."23:51
proctoI made contact today with a guy who runs a clinic in ethiopea23:52
proctowe discussed orgs like diybio producing blueprints of medecines and medical equipment23:52
kanzure_The clinics don't do the ads.23:52
proctoto be produced on-site in africa23:52
kanzure_procto: You should talk to Guido about that. He's been popping up on the diybio mailing list recently, but he's more from Joseph Jackson's side of things on the internet.23:53
willPow3ryeah, why bring the slave labor here?23:53
kanzure_I met him with Joseph u pin CA.23:53
kanzure_procto: If I gave him schematics, would he put them to use?23:55
proctonot yet23:56
proctoI just introduced him to the idea of fablabs, etc.23:56
kanzure_Also, if you talk with Andrew Hessell and convince him to show us something working from his automated diy biotech toolchain *cough*, then that's good too.23:57
kanzure_Ah.23:57
kanzure_*Hessel23:57
proctothing is, he's a fellow student here23:57
proctoso I have close contact23:57
proctoanyways, off to my math midterm review session23:57

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