--- Day changed Fri Sep 26 2008
00:09 < fenn> now there's an idea.. i wonder how big the patents db is
00:10 < fenn> The database now offers the public an additional 1.7 terabytes of full-page images for the 4204863 US patents issued from 1790 through 1975. 
00:10 < fenn> delphion says 9TB
00:10 < kanzure> That's not bad.
00:11 < kanzure> procto, you could cover that right?
00:11 < kanzure> You have your startups doing data processing, right?
00:12 < fenn> i bet it could be slimmed down a lot. i think they use .tiff images and suchforth, and also keep separate documents for all the applications and revisions and miscellaneous crap nobody cares about except lawyers
00:13 < fenn> We have over 100 gigabytes of data which includes the full text of US patents from number 4000000 to 6857132 (most patents before 4,000,000 exist only as images, not searchable text). As more patents are published by the US Patent Office, we add them. We also have over 2 terabytes (a terabyte is 1000 gigabytes) of images so that you can view the illustration from the patents. (freepatentsonline.com)
00:13 < fenn> gzip + jpeg prolly get it all under 50GB
00:13 < kanzure> I'm not about to do an http rip on freepatentsonline
00:13 < kanzure> Where'd they get their db?
00:14 < kanzure> wgetting 100 GB just feels flat out wrong.
00:14 < kanzure> I've wget a max of 5 GB before, but the http overhead is kind of nasty.
00:14 < kanzure> tcp overhead too. yay.
00:15 < fenn> what would you do with a patents db anyway?
00:16 < kanzure> page segmentation algorithms on the PDFs or image files
00:16 < kanzure> reverse engineering with statistical analysis of part names
00:16 < kanzure> then automated generation of the reconstructed designs
00:16 < kanzure> based off of the probabilities of different hierarchical/graph-network-theoretic connections in the conceptual design
00:17 < kanzure> that sort of thing. :)
00:17 < fenn> just because it's patented doesn't mean the design is any good
00:17 < kanzure> it's more for fun than anything practical
00:17 < kanzure> this is true.
00:17 < kanzure> so, what about just the patents that are referenced in scientific literature or something
00:17 < kanzure> sometimes citations point to a patent
00:17 < fenn> and there's a trend to making it impossible to figure out wtf the patent is talking about, so you couldnt reconstruct the design anyway
00:17 < kanzure> bleh
00:17 < kanzure> well
00:17 < kanzure> it's not for one-to-one reconstruction
00:17 < kanzure> more for .. seeding :)
00:18 < kanzure> oh crap, this isn't an OCRed dataset is it?
00:18 < fenn> it is, but not before 1975?
00:18 < fenn> or something like that
00:18 < kanzure> typewritters can be hard to OCR sometimes.
00:25 < kanzure> I wonder if there's anything useful for seeding databases in there.
00:25 < kanzure> oh, if that was to work the parts would have to already be in the database
00:25 < kanzure> it's not exactly a modular approach, even with the citations that each patent have to other patents
00:27 < fenn> i decoded the binary in the star wars pic.. the first droid says "lol" and the second says "lulz"
00:28 < kanzure> what was the encoding
00:28 < fenn> imperial standard
00:28 < kanzure> i am fluent in over six million forms of communication, it's not an alliance code.
00:43 < kanzure> why did google get the patent db?
00:43 < kanzure> is the uspto discriminating against people?
00:43 < kanzure> http://www.uspto.gov/web/patents/pubs/ Office of Patent Publications
00:50 < kanzure> http://patft.uspto.gov/
00:52 < kanzure> on CD? http://ebiz1.uspto.gov/oems25p/index.html
00:52 < kanzure> ' Delivery of Documents: Certified copies of issued/published patent applications as filed and patent file wrappers may be delivered over the Internet or mailed to you on CDROM.  Internet delivery is also available for uncertified copies of patents and patent application publications. Other documents may only be delivered in paper form via U.S. Postal Service'
00:54 < kanzure> http://ebiz1.uspto.gov/vision-service/Product_Services_P/msgShowProductSets?category=P <-- I don't understand much of this.
00:55 < kanzure> http://patft.uspto.gov/help/notices.htm ' These databases are intended for use by the general public. Due  to limitations of equipment and bandwidth, they are not intended to be a  source for bulk downloads of USPTO data.   Individuals, companies, IP addresses, or blocks of IP addresses who, in  effect, deny service to the general public by generating unusually high  numbers (10,000 or more, roughly equivalent to viewing 1,000 patents) of daily databa
00:56 < kanzure> http://www.uspto.gov/web/offices/ac/ido/oeip/catalog/prices/priclist.htm
00:56 < kanzure> price list
00:57 < kanzure> What is all of this ? They hardly have descriptions.
00:58 < kanzure> http://www.uspto.gov/web/offices/ac/ido/oeip/catalog/products/pp-o2w-2.htm#GrantXML5364ol XML data sets? really?
01:06 < kanzure> grr, the design repo kicked me off, I've been downloading "your session has expired" for hours.
02:19 < kanzure> http://www.eurekalert.org/pub_releases/2008-09/usmc-cie091508.php Another case of nanotubes + brain stimulation.
02:23 < ybit> there are several databases that are being grabbed
02:24 < ybit> is there a list of everyone?
02:24 < ybit> perhaps it would be good to make an hplusroadmap torrent tracker?
02:24 < kanzure> http://blog.wired.com/gadgets/2008/09/growing-bamboo.html <-- Growing bikes. ;-) Apparently it's like the aluminum frame bikes.
02:25 < ybit> it would save me time looking through the logs to find interesting things being downloaded
02:25 < kanzure> re-ask question
02:25 < ybit> i saw that mentioned by "kim" on some mailing list but she didn't have a link
02:25 < kanzure> do you want me to start indexing my downloading sprees?
02:25 < kanzure> No, that was on diybio
02:25 < kanzure> and it came with a link :)
02:26 < ybit> ah
02:26 < ybit> is there a list of what everyone here is downloading?""" was the question i believe
02:26 < ybit> hm, not a bad bike
02:26 < kanzure> there's no master list file yet
02:26 < kanzure> archive is kind of hairy and large, untamed
02:28 < kanzure> matweb, biobricks, portions of ncbi, human genome, decodeme files from everyone who has released their data upon the internet, a ridiculously large amount of stuff taken from lab servers, 
02:28 < kanzure> there's a lot of nature in there
02:28 < ybit> ot, concerning star wars: http://www.youtube.com/watch?v=Sv5iEK-IEzw
02:28 < kanzure> is that the one that blew up?
02:29  * kanzure finds a browser that will play flash
02:29 < procto> I'm really not sure how well the connections between the bamboo tubes can hold
02:30 < kanzure> a 'wicker bike' doesn't sound all that great
02:31 < procto> well, bamboo is very strong
02:31 < procto> but a bent bamboo's weak point is the bend
02:38 < kanzure> ybit: I seem to remember you downloading stuff as well.
02:52 < ybit> kanzure: just the git repo
02:52 < ybit> other than that, most files have been deleted
02:53  * kanzure is still looking for anything other than "first come first serve" to land/matter/energy
03:00 < kanzure> and you want me to deploy a torrent tracker?
03:01 < ybit> just saying it would helpful for everyone
03:02 < ybit> if enough peopel gather diff. DBs than a torrent tracker would be nice
03:02  * ybit still can
03:02 < ybit> 't figure out what's with my server
03:03 < kanzure> your server? where?
03:03 < ybit> the second power supply didn't even get a response from the machine
03:03 < ybit> my home
03:03 < ybit> laying beside me right now
03:03 < kanzure> I meant in the DNS lookup tables
03:04 < kanzure> as in, give me the IP or URL, foo'
03:04 < ybit> *@dyndns.org sometime
03:04 < ybit> http://www.youtube.com/watch?v=_370hPM8Cfg
03:05 < ybit> speaking of dyndns, met someone who knew what kde and octave was
03:06 < ybit> he runs open solaris, i'm guessing for zfs, though he was talking about mathematica
03:06  * ybit wonders if he likes java
03:07 < kanzure> I'm not familiar with zfs, is it a fs?
03:07 < ybit> yeah
03:07 < ybit> http://en.wikipedia.org/wiki/Zfs
03:10 < ybit> nsh: did you ever get any feedback to your thoughs on LETS?
03:10 < ybit> (and similar systems)
03:23 < kanzure> Hey gene.
03:23 < gene> hell
03:23 < gene> o
03:24 < gene> yes
03:25 < gene> what is the name of the nanotech IRC server?
03:28 < kanzure> Nanotech IRC server?
03:30 < gene> yeah
03:32 < kanzure> I haven't heard of it.
03:32 < kanzure> Anybody have the old arxiv torrent links?
03:32 < ybit> the closest thing i can come up with is #u-labs on undernet
03:33 < ybit> gene ^
03:46 < gene> isn't arxiv free?
03:48 < kanzure> gene: They do not allow crawls.
03:57 < gene> dang it
03:57 < gene> my mouse is messed up
03:58 < kanzure> http://uniteddiversity.com/nano-powered-electric-brompton-folding-bike/
04:01 < gene> cool
04:01 < gene> I want to do that except with a powered skateboard
04:52 < gene> So who can figure out if this works or not:  http://blog.wired.com/defense/2008/09/chinese-buildin.html
04:55 < gene> I have my money on does not work
05:09 < kanzure> fenn: we could title the paper, "Reverse engineering of every patent ever." 
05:09 < kanzure> Everybody would be all over it.
05:32 < fenn> fail to see how epoxy drenched hemp is more "green" than aluminum
05:34 < fenn> anyway bamboo bikes are old news: http://www.bmeres.com/index.htm
05:34 < fenn> (i went through a bamboo phase a couple years ago)
05:36 < kanzure> heh :)
05:37 < kanzure> bamboo hut and everything?
05:37 < fenn> geodesic domes
05:37 < fenn> i never came up with a good enough strain relief coupling for the vertices
05:38 < fenn> too many conflicting requirements.. i was overengineering things
05:38 < fenn> main problem seems to be supply though
05:38 < fenn> if you have to order the shit from china it doesn't make it any better than just getting electrical conduit from the scrap yard
05:39 < fenn> (even though bamboo grows like a weed here)
05:40 < kanzure> overengineering?
05:40 < kanzure> How far over?
05:40 < fenn> ok backing up a step, i was trying to make a pillow dome, which consists of a rigid geodesic dome frame sheathed in inflatable plastic traingular bubbles
05:41 < fenn> the bubbles were supposed to press tightly against the frame, which should be straight and smooth to make a good seal
05:42 < fenn> so if i did something like, wrap a rubber tube around the bamboo and then put a split piece of conduit over that, the frame tube is no longer the same diameter throughout
05:42 < fenn> so it wouldnt seal right
05:43 < fenn> maybe something like that but inside out might work (silicone caulk over a stud on the inside)
05:43 < fenn> also my bamboo wasn't straight, not sure how that happened
05:44 < fenn> harvested it myself
05:45 < fenn> not much tensile strength in the radial direction though, so the stud would easily split the bamboo, which wouldnt happen with an outer pipe
05:46 < kanzure> oh, so were you just iteratively making changes that required you to redesign the other components?
05:46 < fenn> i was trying to fit bamboo into a scheme that i had worked out for metal tubing
05:46 < fenn> and it wasnt instant enlightenment
05:48 < fenn> if 'change to bamboo' is an iteration, then yeah i guess
05:48 < kanzure> if it only stopped there then nevermind :)
05:49 < fenn> ffs you'd think google could make a simple webpage that loaded properly
05:50 < fenn> in one browser it reloads over and over, and in the other it says The page - www.google.com/?hl=en - does not exist. 
05:56 < gene> epoxy can be made from plants
05:56 < fenn> anything carbon based can be made from plants
05:57 < gene> yeah
05:57 < fenn> burning diesel made from soybean oil is worse for the environment than petroleum
05:57 < gene> plants make resins
05:58 < fenn> the hemp "burl" on that bike does look really neat though
06:01 < gene> http://www.suscomp.com/index.htm
06:01 < gene> 96% vegetable oil UV cure resin
06:02 < gene> hmmmm
06:02 < gene> wonder if I could use a similar process to electrospin MMA without acetone
06:02 < gene> so I could do it in my dorm
06:07 < fenn> can't you just buy some nanotubes?
06:08 < gene> I'm not interested in making nanotubes
06:08 < fenn> or whatever you call a pmma filament
06:08 < gene> probably not
06:08 < fenn> or write to some researcher and ask for a sample
06:08 < gene> not 8 nm nanofiber 
06:09 < gene> it's also quite easy to make
06:09 < fenn> apparently not
06:09 < gene> you take some PMMA dissolve it in acetone
06:10 < fenn> it doesn't dissolve very well btw
06:10 < gene> yeah I know
06:10 < fenn> methylene chloride works way better, but good luck getting that
06:10 < gene> you have to sonicate it
06:10 < gene> but, I have an interesting solution to that
06:10 < fenn> i got some acrylic cement on ebay for ~$25
06:11 < fenn> it contains methylene chloride and maybe some acrylate monomers
06:11 < gene> then it can probably be electrospun
06:11 < gene> put solution in needle charge up needle
06:11 < gene> direct needle at tin foil ground
06:12 < gene> of course you have to provide some pressure to the needle and adjust the flow rate
06:12 < gene> which might be my problem
06:13 < fenn> thickness is probably some function of viscosity, voltage, and distance between needle and plate
06:15 < gene> you have to adjust the flow rate in the needle with some special high accuracy needle squeezer
06:15 < gene> of course if it may not require high accuracy
06:16 < gene> http://www.coe.drexel.edu/ret/personalsites/2006/Sees/researchfocus.html
06:16 < gene> electrospinning set up
06:44 < fenn> zoinks talk about old news bamboo bikes: http://www.wisconsinhistory.org/wlhba/articleView.asp?pg=1&id=10367&pn=0
06:46 < fenn> one hundred years ago you could buy a bike of hickory, elm or even bamboo. Bamboo bicycle companies claimed wood was lighter and better than steel, and so it was. In 1899 some of these companies folded with investors claiming to have been "bamboozled".
06:52 < fenn> "design" apparently means bashing your head against the wall over and over again: http://gp.co.at/stayfocused/jano/pix/jano_presketches_760px.jpg
07:08 < kanzure> fenn: and thus the helmet was born.
07:30 < fenn> i got some snail mail today with a glossy photo of olaf sporns on it, apparently he's being named a 'distinguished faculty member'
12:50 < kanzure> http://groups.google.com/group/openmanufacturing 
12:50 < kanzure> I'm still impressed with the high amount of 'convergenceness' going on there
12:50 < kanzure> feels extropic :)
12:54 < kanzure> hrm, where did  http://openp2pdesign.org/ come from ?
12:54 < kanzure> http://redesignme.com/ - "where everybody is a product (re)designer"
16:08 < kanzure> 30667
16:18  * kanzure just got an email from the Singularity Institute asking me to view their rough draft "to be signed by luminaries" announcing their concerns for ai
16:18 < kanzure> I dislike how SIAI is just a front for Eli's cult
16:31 < kanzure> http://heybryan.org/2008-09-26_ai_singularity_statement.html <- I am not impressed.
16:39 < kanzure> http://cosmeng.org/index.php/Order_of_Cosmic_Engineers:Prospectus
16:40 < kanzure> Everything's too staggard. 
16:44 < kanzure> What's the solution? 
17:06 < kanzure> http://www.buddhism-dict.net/ddb/indexes/term-en.html digital dictionary of buddhism
17:07 < kanzure> http://www.acmuller.net/yogacara/articles/buddhist_steps.html Buddhist steps to an ecology of mind
17:07 < kanzure> oops, ignore
17:09 < kanzure>  ??? Deva of increase and growth;
17:10 < kanzure> Does anyone know how to do romanizations?
17:13 < kanzure> http://en.wikipedia.org/wiki/Personal_growth <-- why does this link over to open source religion?
17:36 < kanzure> oh, which reminds me
17:37 < kanzure> whenever nsh gets back, somebody tell him to go look up hypereconomy and (the now dead) sasha
17:52 < kanzure> http://www.aepalizage.com/lhp/aid.html <-- This reads like it's straight out of Neverness.
17:52 < kanzure> "Left Hand Path"
17:52 < kanzure> uh, fappage?
17:53 < kanzure> http://www.aepalizage.com/lhp/genesis.html yay poems
17:56 < kanzure> allen institute has released spinal cord map according to ##neuroscience
18:19 < kanzure> http://www.ucadia.com/default.htm Haha. "UCADIA = Complete Economic-Political-Social-Spiritual-Scientific model of the world". 
18:19 < kanzure> hehe, these guys are always fun.
18:26 < kanzure> ' http://www.cnn.com/2008/WORLD/asiapcf/09/24/nkorea.nuclear.plant/index.html#cnnSTCText
18:26 < kanzure> The key passage therein is "Last week, a South Korean news agency reported that North Korea was restoring a reactor at Yongbyon nuclear complex and no longer wanted to be removed from a U.S. list of countries that sponsor terrorism."'
20:17 < UtopiahGHML> hi #hplusroadmap
20:18 < kanzure_> Hey there.
20:18 < kanzure_> Okay. So compatibility re: formats. 
20:18 < kanzure_> http://repository.designengineeringlab.org/ (free registration required) has a partial file format, but it kind of sucks.
20:19 < kanzure_> Basically I've been assuming that we'll just go with git + ikiwiki + YAML, where YAML would be for 'metadata' much like the .deb format (specifically the .dsc files inside of it)
20:19 < UtopiahGHML> any API to submit files, check for coherence, simulation, etc...?
20:19 < kanzure_> file submission is basically via git
20:19 < UtopiahGHML> ok
20:19 < kanzure_> simulation is a conceptual difficulty at this point though
20:19 < kanzure_> Consider a frankenstein construction that you design
20:20 < UtopiahGHML> do you have a documented full process that has already produced a result?
20:20 < kanzure_> Say you have a smooth surface and a curved concave metal plate; how do you know their interconnection and where would this be specified?
20:20 < kanzure_> There are results, but these results are not a result of SKDB / our work
20:20 < kanzure_> as it turns out, the lab that I'm currently sitting in has done some related stuffs
20:20 < kanzure_> there's about 20 posters outside these walls showing the automated redesign and simulation of various physical shapes for rockets
20:21 < kanzure_> (nose cone optimization)
20:21 < UtopiahGHML> ok, do they have relations with TechShop as I said before or similar places?
20:22 < UtopiahGHML> do you actuall intend to have such a physical location?
20:22 < UtopiahGHML> or sth in this style
20:24 < kanzure_> Yes, physical locations and all that-- 
20:24 < UtopiahGHML> (sorry if you don't like the comparison but it's the only things of this style I know ;)
20:24 < kanzure_> I have lab space arranged here, I just need to drop the $10k for it
20:24 < kanzure_> Tech Shop -- see also Fab Lab
20:24 < kanzure_> I don't actually have any TechShop connections. I have a few FabLab connections, but more is always nice.
20:30 < kanzure_> So there's still a few issues to hash out. Simulation is a pain in the neck because it's confusing me a lot. I simply don't know how to make things "simulatable" since usually the interconnection of two components requires doubly domain specific knowledge. Rarely do the parts have a common public interface (physically/chemically/mechanically ...).
20:31 < UtopiahGHML> do you have an example in mind?
20:35 < UtopiahGHML> do you expect OpenCog do derivate such models?
20:38 < kanzure_> No, I don't expect that at all.
20:39 < kanzure_> An example of why simulation is hard. Say your design file shows that an egg beater and a gear are to be combined. I have told you absolutely nothing about how this is to physically occur .. there's a difference in the possible ways you could combine them. And each has drastic implications for the function of the device.
20:42 < UtopiahGHML> if it's just a matter of combinations it's only a matter of available computation resources
20:43 < UtopiahGHML> Im not saying it's evident but it's doable
20:43 < UtopiahGHML> mutation/selection/mutation/selection...
20:52 < kanzure_> It's not combinations
20:52 < kanzure_> That's part solvable, it's what I do
20:53 < kanzure_> yep, it's all just evolution and such
20:53 < UtopiahGHML> well I do "believe" in evolutionary algo a lot ;)
20:53 < kanzure_> I suppose I need to go find somebody who has taken an engineering mechanics class and figure out what sort of abstract calculations are possible, because it's just going to be painful to write an algorithmic way of checking that things stick together correctly
20:54 < UtopiahGHML> got friends who are good in math and who did mech engineering
20:54 < UtopiahGHML> if you problem is correctly formulated I can ask them
20:54 < UtopiahGHML> (they do optimization on simulation for french companies, not always "nice" objects)
20:55 < kanzure_> Given two parts that are theoretically able to be interfaced, i.e. the output of the first is of the correct type for the input of the second, how do you make sure you generate a goemetry, or any other substrate, accurately and precisely, so that you may run simulations of the functionality? For instance, in the case of a vehicle, the simulations would be CFD.
20:55 < kanzure_> What's a "non-nice" object?
20:55 < kanzure_> Also, that other message might have cut off.
20:56 < UtopiahGHML> weapons
20:56 < UtopiahGHML> doesn't change their skills though but...
20:56 < UtopiahGHML> they use simulation for shock resistance
20:56 < kanzure_> with weapons I can't imagine them needing to do different types of simulations - always the same sort of calculations about the barrels and the load efficiencies, shock, etc.
20:56 < kanzure_> yeah
20:57 < UtopiahGHML> yep but they optimize pieces fit for shock resistance
20:57 < UtopiahGHML> they design pieces with such problematics
21:00 < kanzure_> Say you have the connection between a screw and two metal plates. This has very defined characteristics that necessites the directionality of the components; this would be important when generating 3D representations of the model. But a non-3D example would be the kinetics of two reactants in a chemical equation. If you don't know how the probabilities of interaction, then you're making some rather big assumptions that you shouldn't be
21:00 < kanzure_> I'm just repeating mysefl.
21:00 < kanzure_> Hoping that there's some generalization I can come up with.
21:01 < UtopiahGHML> well you need a more general model that connects the two, a universal model (universal at least for the ones used)
21:01 < UtopiahGHML> maybe you could have some model ontology/hierarchical repository
21:03 < kanzure_> the ontology kinda already exists, go check out the repository website ;-)
21:03 < kanzure_> http://repository.designengineeringlab.org/
21:03 < UtopiahGHML> ok ok, I won't ask you anything for at least 1 week
21:03 < kanzure_> (btw, that's not our repository - but it sort of is - since it's a joint project with my lab. So.)
21:03 < kanzure_> hah
21:03 < kanzure_> Why?
21:03 < kanzure_> but your idea of using the ontology to come up with the interconnection details might be useful. I'm not entirely sure how to make it happen.
21:03 < UtopiahGHML> I wont ask stupid questions then :)
21:03 < kanzure_> Stupid questions are good.
21:04 < kanzure_> I'll tell you when you get stupid.
21:04 < UtopiahGHML> :P
21:04 < UtopiahGHML> no but I need to sleep over it a bit more then Ill come up with stupid question and smart suggestions
21:04 < kanzure_> :)
21:05 < UtopiahGHML> and if I don't... OpenCog will
21:05 < kanzure_> Tell me what you think OpenCog would have to do with this
21:05 < kanzure_> From what I can remember, the docs didn't explain much, but this was back in February that I looked
21:06 < UtopiahGHML> well so far... not a lot it's still Work In Progress
21:06 < kanzure_> but *what* is WIP?
21:06 < kanzure_> Magical ai beans?
21:06 < UtopiahGHML> but if you need to manipulate ontology, do ANN optimization, inferences based on models... could help there
21:06 < kanzure_> I've been out of the loop on Ben's work for a while now.
21:07 < UtopiahGHML> even if it "only" provide a coherent framework for the different AI tools
21:07 < kanzure_> is that what it's doing?
21:07 < kanzure_> http://heybryan.org/mediawiki/index.php/Computational_neuroscience
21:07 < UtopiahGHML> that's already a big gain even if the "AGI vision" doesn't get realized soon.
21:07 < kanzure_> If you can come up with a framework for all of those tools listed on that page, then I'll throw it into my 'mind-on-a-disc' Linux CD.
21:07 < UtopiahGHML> :)
21:07 < UtopiahGHML> I guess you know how to use a wiki
21:08 < kanzure_> uh, yes :)
21:08 < UtopiahGHML> http://seedea.free.fr/wiki/index.php?n=Seedea.AImatrix
21:09 < UtopiahGHML> started that few days ago, obviously not populated at all, didn't take time for it yet
21:09 < UtopiahGHML> but you get the idea
21:09 < kanzure_> hm - http://mathforum.org/~ken/bayes/bayes.html - automated doc classification w/ perl :)
21:10 < kanzure_> if I was back home I'd show you my ai bookmarks.
21:10 < UtopiahGHML> this would evolve to an RSS repository on key AI labs, etc.. etc... etc....
21:11 < kanzure_> http://heybryan.org/bookmarks/
21:11 < kanzure_> Check the OPML file.
21:11 < kanzure_> I keep track of a good number of AI RSS feeds.
21:11 < UtopiahGHML> but so far, IMHO, OpenCog seems to be the AI unifying framework with the best potential
21:11 < kanzure_> well, by "keep track" i mean glance over, because it's too painful to actually read all of the BS
21:11 < UtopiahGHML> (and that's why Im on the channel)
21:11 < kanzure_> :)
21:12 < kanzure_> That's a good reason to be there.
21:12 < UtopiahGHML> Cognitive scaffolding =
21:12 < UtopiahGHML>     * Cognitive proxy (now instantiated as a browser+plugins+wiki+pugins+physical interface package)
21:12 < UtopiahGHML>           o Intellectual augmentation
21:12 < UtopiahGHML>                 + Artificial intelligence
21:12 < UtopiahGHML>                       # Algorithms
21:12 < UtopiahGHML>                             * Heuristics, Mimetics (cognitive science, game theory, ...)
21:12 < UtopiahGHML>                                   o Statistics
21:12 < UtopiahGHML>                                         + Mathematics 
21:12 < UtopiahGHML> maybe it's close to your mindCD except being online
21:14 < kanzure_> re: brain augmentation
21:14 < kanzure_> http://heybryan.org/docs/neuro/ brain implants
21:14 < kanzure_> http://heybryan.org/2008-08-15.html more recent work on that
21:14 < kanzure_> http://heybryan.org/~bbishop/docs/open-rtms/
21:14 < kanzure_> http://heybryan.org/recursion.html
21:14 < kanzure_> http://heybryan.org/humancortex.html
21:14 < kanzure_> http://heybryan.org/buildingbrains.html
21:15 < UtopiahGHML> I dont read that fast without my model being actually instanciated fully (yet) ;)
21:16 < UtopiahGHML> I guess you've seen « corticogenèse in vitro » (Nature 17th of
21:16 < kanzure_> The trick to reading faster is to assume you've read it.
21:16 < UtopiahGHML> August) already
21:16 < kanzure_> I might have. What was it about?
21:16 < kanzure_> Was this blood v. activity?
21:16 < UtopiahGHML> "implanted neurons are able to connect to the host brain to build specific neural circuits in the cortex"
21:16 < kanzure_> uhh
21:16 < kanzure_> link? :)
21:16 < UtopiahGHML> http://www.ulb.ac.be/homepage.html http://www.lemonde.fr/sciences-et-environnement/article/2008/08/18/du-cortex-cerebral-fabrique-in-vitro-a-partir-de-cellules-souches_1084950_3244.html
21:17 < UtopiahGHML> (don't have the DOI of Nature's article)
21:17 < kanzure_> for the record,
21:17 < kanzure_> I have every single Nature paper ever
21:17 < kanzure_> except things from, uh, August, I think.
21:17 < UtopiahGHML> well, you did assume you've read it anyway I guess :P
21:18 < UtopiahGHML> (hmmm the ULB link changed, the lab didn't keep an up to date page...)
21:18 < kanzure_> Is there one in English?
21:19 < UtopiahGHML> the one in Nature
21:19 < kanzure_> erm
21:20 < UtopiahGHML> author Nicolas Gaspard
21:20 < kanzure_> An intrinsic mechanism of corticogenesis from embryonic stem cells
21:20 < UtopiahGHML> maybe with http://dev.ulb.ac.be/pvdhlab/
21:20 < kanzure_> http://www.nature.com/nature/journal/v455/n7211/full/nature07287.html
21:20 < UtopiahGHML> yep that one
21:21 < UtopiahGHML> ehhh they updated it?!
21:21 < UtopiahGHML> ok corrections
21:21 < kanzure_> http://search.cpan.org/~jhi/Graph-0.84/lib/Graph.pod#Comparing
21:21 < kanzure_> Hm. Looks like all of my work is done for me.
21:22 < kanzure_> woot, plus isomorphisms.
21:22 < UtopiahGHML> isn't it the goal of the Internet?
21:22 < kanzure_> To do *my* work specifically?
21:22 < kanzure_> Yes, yes it is.
21:24 < kanzure_> hm, the isomorphic functions don't return the areas of the graph that are isomorphic
21:24 < UtopiahGHML> (but yes brain augmentation is part of my tree but not in the cognitive scaffolding, not sure it's required)
21:25 < kanzure_> UtopiahGHML: be back in a second; in the mean time, amuse yourself: http://heybryan.org/quotes.html
21:27 < kanzure_> http://qosip.tmit.bme.hu/~retvari/Lemon-Graph.html a combinatorial optimization and graph lib
21:31 < UtopiahGHML> do you have an API/Lib repository that you can query when needed?
21:31 < kanzure_> http://search.cpan.org/~azs/Graph-ModularDecomposition-0.13/ModularDecomposition.pm is this what I need?
21:31 < kanzure_> what do you mean, UtopiahGHML?
21:31 < kanzure_> Yes, I do have a server, if that's what you want to know
21:32 < UtopiahGHML> nop
21:32 < UtopiahGHML> http://www.catonmat.net/blog/searching-and-mining-open-source-code-from-the-web/ configured for your needs is what I meant
21:33 < kanzure_> grr, I might need a regexp for graph searching
21:33 < kanzure_> hm?
21:34 < kanzure_> just looks like a demonstration of krugle and friends
21:34 < UtopiahGHML> have a look at the references in the presentation maybe
21:35 < UtopiahGHML> they provide tools for Eclipse&Co to go further than that.
21:37 < kanzure_> yay, this is what I need: http://search.cpan.org/~jhi/Graph-0.84/lib/Graph/UnionFind.pm
21:38 < UtopiahGHML> what are you actually trying to code
21:41 < kanzure_> At the moment I'm just writing some stuff to show the union and intersection of different graphs
21:41 < kanzure_> it's just a small exercise really, somewhat to pass time.
21:45 < UtopiahGHML> ok
21:45 < UtopiahGHML> well Ill pass time sleeping
21:46 < UtopiahGHML> bye
21:46 < kanzure_> bye
22:02 < kanzure_> hm, I need a way to do WYSIWYG-HTML with directed graphs. I know there's HTML::TreeLayout or something, but that's more for parsing than building pages. I guess there should be something somewhere .. CGI and all that.
22:10 < kanzure_> 302606 <--- HTML templating as tree rewriting
22:11 < kanzure_> http://perlmonks.org/index.pl?node_id=302606
22:12 < kanzure_> http://kobesearch.cpan.org/htdocs/HTML/html_base.pod.html ah, it was for HTML 2.0
22:19 < kanzure_> and for HTML 4.0 --> http://search.cpan.org/~duncand/HTML-EasyTags-1.071/lib/HTML/EasyTags.pm
22:20 < kanzure_> How (un)fortunate that HTML doesn't have cyclic subgraphs. Oh well. ;-)
22:32 < fenn> not unfortunate, if it did then the page would be infinitely long
22:36 < kanzure_> I mena those old OOP diamondoid loopers that you're supposed to avoid in your UMLing
22:39 < fenn> ya there's a reason you're supposed to avoid them
22:39 < kanzure_> turtle eating his own tail?
22:40 < fenn> something like that
22:46 < kanzure_> fenn: http://heybryan.org/~bbishop/docs/repo/
22:46 < kanzure_> it just finished.
22:47 < fenn> ah i thought it was going to be the matweb repo :)
22:47 < kanzure_> alas :(
22:47 < kanzure_> let me check on that while I'm thinking about it
22:47 < fenn> those files are bigger than one would expect
22:48 < kanzure_> fscking images are stored in them
22:48 < kanzure_> this doesn't explain the roller coaster's size though
22:48 < kanzure_> 100+ MB, wtf?
22:49 < fenn> well the images are in ascii so there's some overhead
22:49 < kanzure_> 32808
22:50 < fenn> for i in ./* ; do echo $i | grep -v > $i ; done
22:50  * kanzure_ has to ls for the largest file number by the typical bounded trial and error
22:50 < fenn> er, s/grep -v/grep -v CDATA/
22:50 < kanzure_> I think there's some theorem in number theory or something,
22:50 < kanzure_> that shows that I only have to do it at most 10 times for any number that I'm bounding
22:50 < fenn> or soething like that :P
22:50 < kanzure_> erm, proportional to the number of digits or something
22:51 < kanzure_> this time it took me 14 guesses
22:51 < kanzure_> though I started by assuming it was past 30000
22:51 < fenn> standard deviation?
22:52 < kanzure_> erm. maybe.
22:52 < fenn> i dunno what you're talking about really
22:52 < kanzure_> I'm still not familiar with standard deviations
22:52 < kanzure_> uhm
22:52 < kanzure_> so I have them all in files
22:52 < kanzure_> the latest downloaded file is where it's at
22:52 < fenn> with names like air_hawg.repo
22:52 < kanzure_> matweb, matweb
22:52 < kanzure_> you mentioned matweb, so I'm checking matweb's download
22:52 < fenn> oh
22:53  * fenn has high mental inertia
22:53 < fenn> especially first thing in the morning
22:53 < kanzure_> heh
22:54 < fenn> how big is design repo if you gzip it?
22:56 < kanzure_> odd, I know how to use zip but not gzip for compression
22:57 < kanzure_> oh
22:57 < kanzure_> need to tar it, don't I?
22:57 < fenn> tar -zcvf repo.tgz repo/
22:58 < kanzure_> will take a few moments.
22:58 < kanzure_> crap I'm stupid. repo/ is for a reason .. dir preservation. 
22:59 < fenn> did you just tarbomb yourself?
22:59 < kanzure_> no, I was still building the tar
22:59 < kanzure_> also, is it still a tarbomb if all of the filenames have a standard extension?
22:59 < kanzure_> *shrug*
23:00 < fenn> depends if you have other filenames with that extension in your dir already :)
23:00 < fenn> yet another case for versioning file systems
23:00 < kanzure_> can't count the number of times I've had to go digging through very long lists of files for "hey, I *think* that's new .."
23:01 < fenn> ls -lart
23:01 < kanzure_> the number is high, though
23:01 < kanzure_> ls -lart helps?
23:01 < fenn> shows newest files last
23:01 < fenn> unless your tar preserved dates :(
23:01 < kanzure_> gzip preserves dates, fwiw
23:01 < fenn> depends what switches you give it
23:02 < kanzure_> wtfpancakes why is there repo.tgz in my repo.tgz file
23:02 < kanzure_> grr
23:02 < fenn> it's a recursive backup :)
23:02 < fenn> that way when you delete the original archive you have an extra one left over
23:03 < kanzure_> huh? 
23:03 < fenn> (yes i'm being silly)
23:03 < kanzure_> No, that's because of my mistake
23:03 < kanzure_> yeah
23:03 < kanzure_> okay, just checking.
23:04 < kanzure_> 373 MB to 500 MB.
23:04 < kanzure_> http://heybryan.org/~bbishop/docs/
23:04 < fenn> i thought it was 100MB before compression?
23:05 < fenn> guess not
23:05 < kanzure_> that was a single file in there
23:07 < kanzure_> next on my list should be automated web design, since that's something I can test with
23:07 < kanzure_> or, figuring out which subcomponents are commonalities between those parts in the repo
23:08 < kanzure_> not that it gives you any real extra info.
23:08 < kanzure_> blah. anyway, I'm off to fetch a burger and doodle death stars
23:08 < fenn> enjoi
23:28 < fenn> just discovered this http://beagleboard.org/
23:28 < fenn> looks eminently wearable
23:38 < fenn> omfg they hired a guy in a dog suit
23:41 < kanzure> that dog suit under 'Latest News' is from Invader Zim.
23:41 < fenn> dirty dirty dog: http://www.flickr.com/photos/jadon/2740273801/
23:42 < kanzure> meh, it happens
23:42 < fenn> there are an unusually large number of women in these photos
23:43 < fenn> i wonder if it's observer bias or just that companies have women to staff their booths
23:52 < kanzure> I still think this is comedic genius: http://neuroblastoma.omgwallhack.org/media/img/dump/instructions.jpg
23:59 < kanzure> http://geekz.co.uk/lovesraymond/archive/so-i-married-a-kernel-programmer "Everybody loves Eric Raymond"