2016-01-17.log

--- Log opened Sun Jan 17 00:00:07 2016
-!- PatrickRobotham [uid18270@gateway/web/irccloud.com/x-fbhkjeopmveieqpw] has joined ##hplusroadmap00:38
lkjhfralso, your scrollback hasn't been pushed out by messages00:57
lkjhfrhttps://irssi.org/documentation/settings/#scrollback_time00:58
lkjhfrthis means that scrollback is shrinking all the time, even if people aren't posting messages01:00
lkjhfrit shrinks down to scrollback_lines after scrollback_time01:01
-!- Gurkenglas [Gurkenglas@dslb-188-103-077-131.188.103.pools.vodafone-ip.de] has joined ##hplusroadmap01:13
kanzureaccelerating expansion of the scrollback does not bode well for cosmological outcomes01:21
lkjhfrfor five hours after 500 messages after the message from pasky that message wasn't yet 24 hours old, which means that it was pushed out by time01:33
-!- Houshalter [~Houshalte@oh-71-50-56-224.dhcp.embarqhsd.net] has quit [Quit: Quit]01:33
-!- drewbot [~cinch@ec2-54-162-205-126.compute-1.amazonaws.com] has quit [Remote host closed the connection]01:55
-!- drewbot [~cinch@ec2-107-22-60-23.compute-1.amazonaws.com] has joined ##hplusroadmap01:56
lkjhfrkanzure, is it because we can't use it as an argument against flooding?02:28
archelsgotta love it when people include the words "an accidental finding" in the title of their paper02:57
archels.title http://www.hoajonline.com/stemcells/2054-717X/2/302:59
yoleauxFulltext | Reversal of hair greying following autologous adipose mesenchymal stem cell transplantations: a coincidental finding | Stem Cell Biology and Research02:59
justanotherusermaaku: is there any reason I should be skeptical of opencog? I am looking into it's design, and I think it's kind of beautiful03:09
-!- chris_99 [~chris_99@unaffiliated/chris-99/x-3062929] has joined ##hplusroadmap03:09
justanotheruserI am a bit skeptical about all its moving parts.03:12
-!- Madplatypus [uid19957@gateway/web/irccloud.com/x-qbtvlkaboofsattp] has quit [Quit: Connection closed for inactivity]03:27
-!- pompolic [~A@unaffiliated/pompolic] has quit [Read error: Connection reset by peer]03:29
-!- pompolic [~A@unaffiliated/pompolic] has joined ##hplusroadmap03:30
archels.gc "deterministic.chaos"03:34
yoleauxarchels: Sorry, that command (.gc) crashed.03:34
-!- pompolic [~A@unaffiliated/pompolic] has quit [Read error: Connection reset by peer]03:35
-!- pompolic [~A@unaffiliated/pompolic] has joined ##hplusroadmap03:35
FourFire> What about those alien megastructures? Schafer is unconvinced. “The alien-megastructure idea runs wrong with my new observations,” he says, as he thinks even advanced aliens wouldn’t be able to build something capable of covering a fifth of a star in just a century.04:00
FourFireI find that claim to be too conservative...04:00
FourFireThe cryonics possibilities in europe were rather unfortunate last time I checked04:02
FourFire> <maaku> i wonder if one could reasonably harvest resources to create computronium from coronal mass ejections04:03
FourFirewith fusion, everything is possible04:03
FourFireyou just need to sacrifice a lot of energy to suppass Iron04:03
paskyxentrac: i have similar feelings about the backlog, but i honestly can't remember, you'll have to take a peek at the logs04:05
FourFire> <Houshalter> I believe that superintelligent humans would likely be bored, depressed, or wire headers. or some combination of them04:12
FourFireThe bored, depressed, wireheaders will promptly filter themselves out of the population04:12
FourFire> <Houshalter> maaku, i am. but i won't change inot a different person, even over a thousand years. i will learn a lot, but my brain structure and personality will be about the same04:13
FourFireHow old are you?04:13
FourFireLike, how long do you remember being conscious?04:14
lkjhfri like bored and depressed. much better than stupid and annoying04:17
FourFire> <streety> one issue with unburdening the immune system so it is free to "take care of more useful things" is the hypothesis that this is a primary cause of the increasing incidence of allergies and auto-immune diseases04:21
FourFirethere's some resaerch with strong evidence that absence of symbiotic gut bacteria increases allergic reactions04:22
-!- erasmus [~erasmus@unaffiliated/erasmus] has quit [Quit: Namaste]04:24
lkjhfrthe ones that constitute dental plaque are the same that are supposedly synbiotic as a gut flora, it doesn't mean that it's good idea to keep them on teeth04:24
FourFirelkjhfr, but maybe we can keep them off teeth and in the gut at the same time?04:26
lkjhfrand those are left on teeth only because people eat processed foods, eating sweet fruits like apples actually cleans the teeth04:27
cluckjhygiene hypothesis04:56
Diablo-D3well except the problem is05:07
Diablo-D3we keep eating high sugar foods, which causes bacterial growth of the wrong kind of bacteria05:07
Diablo-D3and then we compound the problem by using commercial detergents and floruide05:08
streetyIt is definitely important that the right bacteria be in the right place in the right quantity. Poor oral health causes problems but you don't want to swing too far in the opposite direction either05:16
-!- chris_99 [~chris_99@unaffiliated/chris-99/x-3062929] has quit [Quit: Leaving]05:24
-!- erasmus [~erasmus@unaffiliated/erasmus] has joined ##hplusroadmap05:29
-!- poppingtonic [~Thunderbi@unaffiliated/poppingtonic] has joined ##hplusroadmap05:33
-!- poppingtonic [~Thunderbi@unaffiliated/poppingtonic] has quit [Remote host closed the connection]05:55
-!- jtimon [~quassel@126.31.134.37.dynamic.jazztel.es] has joined ##hplusroadmap06:35
-!- AmbulatoryCortex [~Ambulator@173-31-155-69.client.mchsi.com] has joined ##hplusroadmap06:48
-!- CheckDavid [uid14990@gateway/web/irccloud.com/x-wyihyzthfwohhzfq] has joined ##hplusroadmap07:01
-!- jenelizabeth__ [~jenelizab@cpc76802-brmb10-2-0-cust399.1-3.cable.virginm.net] has quit [Quit: sleeeep]07:10
kanzurethat's strange, i don't ever remember being conscious07:25
kanzurebitfury 16nm asic chip demo video https://www.youtube.com/watch?v=3zPpj1JYw3807:27
Diablo-D3whos fabbing it07:31
Diablo-D3because no one is doing 16nm for small time stuff yet07:32
Diablo-D3all of the 14/16nm stuff is being used by intel, amd, more amd, nvidia, and samsung07:33
Diablo-D3oh and qualcomm for snapdragons07:35
Diablo-D3and the joint intel/micron factory is churning out 16nm flash at that size07:36
nshi bet i could fabricant 100 nomometers07:44
nshmeanwhile: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4081593/07:45
nsh.title07:45
yoleauxSoma-to-Germline Transmission of RNA in Mice Xenografted with Human Tumour Cells: Possible Transport by Exosomes07:45
nshall up in your central dogmas, killin' ur 'sumptions07:45
* Diablo-D3 fabricates 100 omnommeters07:45
Diablo-D3nsh: construct additional pylori07:46
nshokay, but i'm almost out of pylorum-bricks07:46
FourFirensh, really?07:47
nshnah, they're made out of common07:48
nsh.wa total mass of bacteria on earth07:48
yoleauxEarth: weight: bacteria: weight: (data not available)07:48
nshpft07:48
Diablo-D3heh, probably at least triple digit thousands of tons07:48
nsh~30-40% of earth-life is microbial07:48
nshand there's a whole heap of that kicking about07:49
nshkeep that wheel a-turnin07:49
FourFireanyone think there's any point in trying to produce this: http://memory.oyhus.no/ (patent is expired, I know the guy)07:49
nsh.title http://www.sdearthtimes.com/et0998/et0998s8.html07:49
yoleauxET 9/98: First-ever estimate of total bacteria on earth07:49
nshFourFire, good luck coding the microprocessor for wear-levelling that....07:50
-!- |node [uid125132@gateway/web/irccloud.com/x-tujvefywzxrjgvyz] has joined ##hplusroadmap07:50
FourFirensh, if it's cheap enough, I figure you could just something stupid like mirror everything internally in the device and test for dead cells upon read07:50
* nsh nods07:51
FourFireso if it's 10x cheaper, make it 1/3 the price and use the cells07:51
nshECCs go a long way07:51
FourFirestill, 10s of terabytes of flash in one device...07:51
Diablo-D3"The team thus found that the total amount of bacterial carbon in the soil and subsurface to be yet another staggering number, 5 X 10**17 g or the weight of the United Kingdom."07:51
Diablo-D3but it doesn't say the total weight damnit07:51
-!- dustinm [~dustinm@2607:5300:100:200::160d] has quit [Ping timeout: 255 seconds]07:53
-!- dustinm [~dustinm@105.ip-167-114-152.net] has joined ##hplusroadmap07:56
-!- AmbulatoryC0rtex [~Ambulator@173-31-155-69.client.mchsi.com] has joined ##hplusroadmap08:00
-!- AmbulatoryCortex [~Ambulator@173-31-155-69.client.mchsi.com] has quit [Ping timeout: 276 seconds]08:04
-!- Quashie [~boingredd@ool-18bccfe5.dyn.optonline.net] has quit [Ping timeout: 264 seconds]08:19
-!- Quashie [~boingredd@ool-18bccfe5.dyn.optonline.net] has joined ##hplusroadmap08:26
-!- chris_99 [~chris_99@unaffiliated/chris-99/x-3062929] has joined ##hplusroadmap08:28
-!- c0rw|away is now known as c0rw1n08:37
-!- PatrickRobotham [uid18270@gateway/web/irccloud.com/x-fbhkjeopmveieqpw] has quit [Quit: Connection closed for inactivity]08:40
-!- yashgaroth [~yashgarot@2602:306:35fa:d500:f5e0:f867:a11d:8d52] has joined ##hplusroadmap09:01
-!- chris_99 [~chris_99@unaffiliated/chris-99/x-3062929] has quit [Remote host closed the connection]09:06
lkjhfrand... all of this carbon is fixed by plants that grow out of soil09:06
kanzurehttps://petertodd.org/2016/soft-forks-are-safer-than-hard-forks09:07
kanzurehttp://earlz.net/view/2016/01/16/0717/analyzing-the-56-million-exploit-and-cryptsys-security09:07
lkjhfri expected that firt link to be about plastic vs metal forks09:09
maakujustanotheruser design no, code yes. They make too many compromises, and the implementation is not self reflective09:14
-!- CheckDavid [uid14990@gateway/web/irccloud.com/x-wyihyzthfwohhzfq] has quit [Quit: Connection closed for inactivity]09:19
-!- chris_99 [~chris_99@unaffiliated/chris-99/x-3062929] has joined ##hplusroadmap10:04
-!- Pathseeker [~Pathseeke@176.4.98.38] has joined ##hplusroadmap10:44
-!- augur [~augur@c-73-46-94-9.hsd1.fl.comcast.net] has quit [Ping timeout: 260 seconds]10:49
-!- Madplatypus [uid19957@gateway/web/irccloud.com/x-zkladkvhktltwwwh] has joined ##hplusroadmap10:52
-!- augur [~augur@c-73-46-94-9.hsd1.fl.comcast.net] has joined ##hplusroadmap10:56
-!- ArturShaik [~ArturShai@212.97.26.254] has quit [Ping timeout: 276 seconds]11:15
-!- erasmus [~erasmus@unaffiliated/erasmus] has quit [Ping timeout: 264 seconds]11:42
-!- Jaw_ [~Jawmare@unaffiliated/jawmare] has quit [Quit: Leaving]11:48
-!- Pathseeker [~Pathseeke@176.4.98.38] has quit [Quit: Leaving]12:15
justanotheruserThe implementation doesn't follow their specification maaku?12:17
maakujustanotheruser: ben goertzel has written a >1k pages book, in two volumes, called "engineering general intelligence". you can find pdfs online12:18
justanotheruserYes, I have downloaded vol 212:19
justanotheruserin what sense isn't their implementation self reflective12:19
maakuthat outlines "CogPrime" which is an abstract AGI design based on the integrative approach -- a bunch of "narrow AI" algorithms working together in concert on a single shared backbone in order to patchwork-style cover all the bases12:19
maakuthat is a solid idea12:19
maakui take some issue with a few of their implementation choices, e.g. PLN instead of a more accurate probabalistic graph update mechanism a la Pearl12:20
maakuor using DeSTIN instead of the much more capable deep learning tools that are coming out of Google/Hinton12:21
maakubut that's not so important12:21
-!- |node [uid125132@gateway/web/irccloud.com/x-tujvefywzxrjgvyz] has quit [Quit: Connection closed for inactivity]12:21
FourFireok, trying to figure out open source, PCIe passthrough virtualization software, being more confused...12:21
maakuthe big issue with OpenCog (as a proposed implementation of CogPrime) is that all these algorithms are implemented in C++ and Python in a way that is totally opaque12:22
FourFireso far I've rounded it down to kvm, and Xen(server)12:22
maakuto the running mind that is. it can't self-improve12:22
lkjhfrwould it be easy for the ai to self-improve if it didn't know c++ or python in the first place?12:29
lkjhfror is it about hot-plugging, which not many people even think about these days, not understanding the process of linking12:33
maakulkjhfr: no i don't mean anything like that. the context you are probably missing is that CogPrime involves shared data structure for its knowledge base, which includes an executable specification12:38
maakuto accomplish evolutionary learning of new programs, for example12:38
maakuthe way you would close the loop in cogprime is to implement all these algorithms in that language, within the system12:39
maakuso it can use the same concept-learning techniques to learn better implementations of itself12:39
justanotherusermaaku: The program could have its own source code known though? I don't see why you would need to shove the source code into every openCog application12:42
maakujustanotheruser: sure, that's just the boil-the-ocean approach12:42
maakujustanotheruser: we'd have to take an AI all the way to the level of budding computer scientist in order to reliably make any changes to its own source code12:45
paskywe can't self-improve (in that sense) either, is that really an issue?12:46
justanotheruserwhats the problem with that?12:46
maakuwhereas the alternative is to represent the AI algorithms within its own concept-learning language, so it learns newer algorithms for itself in the same way that it learns better representations and better classifications12:46
maakuyou get a feedback loop very early on that way12:46
maaku(and an architecture like GOLUM is an example of how you would do so with relative safety)12:47
xentracpasky: it's a significant difference: that's the difference between us staying at the same level of intelligence for three million years and slightly-subhuman artificial intelligences turning themselves into extremely-superhuman ones within a matter of years, months, days, or seconds12:47
paskybut that's not that realistic, is it?12:48
maakudays or seconds was not realistic12:49
maakubut the basic idea is12:49
FourFireok so apparently Nvidia support is shitty for all alternatives12:50
maakua hard takeoff on the order of months is plausible in a breakout scenario (low probability anyway), years under controlled circumstances12:51
xentracnobody knows what is realistic12:51
xentracand my limited experience programming things that do optimization or search is that apparently small improvements can have surprisingly large effects when they affect the base of an exponential function12:52
lkjhfrit doesn't look like that book is free12:53
maakuxentrac: nonsense. it only takes longer because there's a limiter. what would that limiter be?12:53
lkjhfrdoes this mean that there is a language that in which programs can be represented which ai will already know?12:54
maakuargument from incredibility is not valid. why would it take longer?12:54
maaku(i know it won't happen in seconds because of computational limits, and not days or weeks because of logistical limits, but why not months?)12:54
lkjhfrs/language that in/language in/12:54
maakulkjhfr: yes, that is a component of the CogPrime design12:55
xentrachmm, you seem to be thinking that I think a hard takeoff in months is implausible12:55
maakuit has a built-in programming language called 'combo' -- my own design has a much more expressive VM language, but same principle12:55
xentracbut I was saying the opposite: that a hard takeoff is entirely plausible, and that not only months but even seconds are plausible timescales12:55
maakuxentrac: i thought that's what you said in response to me. i think months is the lower end of plausible but not probable12:56
lkjhfris that language already designed? how strict is it?12:56
maakuxentrac: eh. you can count the number of cpu/gpu cycles available in 1 day. it's a finite number12:56
maakuand given my familiarity with the types of algorithms involved I highly doubt anything meaningful could be accomplished in that timeframe12:57
maakui admit that's an argument from gut feeling, but we're talking orders of magnitude difference...12:57
xentracright. what I can't count is the minimal number of cpu cycles needed to improve an AI algorithm. I am sure your estimate is more informed than mine12:57
xentracbut I think there's an enormous amount of uncertainty12:58
maakuxentrac: for each of the components (e.g. deep learning neural net training, combinatorially complex probabalistic reasoning, etc.) these sorts of things often accomplish single results with hours of work on large clusters of workstations13:00
maakuan AGI self-improvement cycle would involve such work to consider such a change, then equivalent or more work to evaluate changes in various simulated environments13:00
maakuand it will probably consider many options before making a single incremental change, so multiply by some small constant13:01
lkjhfrso... opencog is free, but the book is not13:01
maakuxentrac: add those up and we're already talking about days or a week of effort per cycle13:02
maakunot all improvements will be paradigm shifts, but some will be. we're still talking many weeks, a few months before it surpasses its creators assuming it doesn't break and need to be debugged which is the real most likely outcome13:04
lkjhfrit's almost like linux and minix :)13:04
xentracyeah, things being broken is always the mots likely outcome :)13:04
maakuso this back of the envelope calculation arrives at a number that is six to seven orders of magnitude more than "seconds"13:05
maakuwhich is why I don't give much credence to hard takeoffs13:05
maakuthat changes as computing advances though (an argument for doing AGI now, not later)13:06
xentracyou could just as easily say that the fact that deep-learning NN training takes hours across a large cluster is an argument that AGI is not yet practical13:07
maakuIn 30 years that number will have reduced to seconds.13:07
FourFireended up giving up, going to try KVM Lubuntu Host, Lubuntu Guest13:08
xentracyou need either much better hardware (your 30 years) or much faster software to get that number down before seconds-scale hard takeoff is possible, agreed13:08
lkjhfrbut by then there will be invented new algorithms that everyone will want to embed in theirs ai13:08
xentracbut the much faster software might be possible13:08
FourFiremaaku, if you can make an application well optimized then you might be able to reduce the problem to ASICs13:09
FourFireof FPGAs13:10
FourFirecurrent old lithography is only going to get cheaper13:10
xentracmuch faster hardware in much less than 30 years is also conceivable but less likely13:11
maakuxentrac: getting a little off topic, but this is the core of what bugs me about MIRI and FAI x-risk activists generally. The argument is made that we should stop work on AGI and instead work on solving the much harder FAI -13:12
lkjhfrhttp://goertzel.org/EGI_vol1..pdf.zip http://goertzel.org/EGI_vol2..pdf.zip13:13
maaku- but the 30 years spent doing that would mean there's no capacity left for safe engineering13:13
-!- chris_99 [~chris_99@unaffiliated/chris-99/x-3062929] has quit [Quit: Leaving]13:13
maakuIf you are worried about AGI x-risk, then do what you can to get AGI written -now-, ASAP13:13
drethelinyou seem to have completely missed the point13:17
xentracI understand your argument, and it may be correct13:18
-!- [dpk] [~dpk@xn--ht-1ia18f.nonceword.org] has joined ##hplusroadmap13:18
xentracit is contingent on hardware improvement being at least a necessary, if not a sufficient, AGI success factor13:19
-!- dpk [~dpk@xn--ht-1ia18f.nonceword.org] has quit [Ping timeout: 246 seconds]13:19
-!- [dpk] is now known as dpk13:19
maakudrethelin: me? perhaps so13:21
-!- Pompolic_ [~A@unaffiliated/pompolic] has joined ##hplusroadmap13:22
-!- pompolic [~A@unaffiliated/pompolic] has quit [Disconnected by services]13:22
-!- Pompolic_ is now known as Pompolic13:22
-!- Pompolic is now known as pompolic13:23
Alcyiusanyone want to watch Mr. Nobody?13:23
lkjhfrfun fact: there are over 5000 links to pdfs in the logs13:24
-!- CheckDavid [uid14990@gateway/web/irccloud.com/x-hnqpwtbifqfbbjzm] has joined ##hplusroadmap13:24
AlcyiusPsuedo time travel film about the last man to die in a future where everyone is immortal13:24
Alcyiushttps://rabb.it/r/jy5im1?i=xW9ERLCMhjtY13:24
AlcyiusLink if anyone's interested13:24
-!- Houshalter [~Houshalte@oh-71-50-56-224.dhcp.embarqhsd.net] has joined ##hplusroadmap13:26
-!- chris_99 [~chris_99@unaffiliated/chris-99/x-3062929] has joined ##hplusroadmap13:28
-!- Netsplit *.net <-> *.split quits: sandeepkr, jtimon, xentrac, helleshin, juul13:29
streetylkjhfr: what are the most common topics? or domains?13:39
-!- Aurelius_Laptop [~cpopell@c-73-129-20-70.hsd1.md.comcast.net] has joined ##hplusroadmap13:39
maakuAlcyius another fun movie is "the man from earth"13:41
AlcyiusHas anyone here read Stranger from a Strange Land13:42
AlcyiusMy mom keeps trying to get me to read it13:42
Diablo-D3its a very good book13:42
Diablo-D3I read it when I was 1213:43
lkjhfrstreety, most were posted by kanzure, so i'd guess it would be something about treating and enhancing humans13:44
streetymight be interesting to try and extract topics13:46
streetyIt seems we occasionally focus on particular areas for a while then change focus.13:47
maakuThat's a weird book for your mom to push.13:51
lkjhfrstreety, i'm not sure it would be about treating and enhancing humans, there is just so many different things13:52
streetyyes, the conversation does touch on a wide variety of topics13:52
lkjhfrpdftk could read the titles13:53
lkjhfrsomething would be needed to grep bare links to make download list for wget13:53
-!- Houshalter [~Houshalte@oh-71-50-56-224.dhcp.embarqhsd.net] has quit [Ping timeout: 276 seconds]13:57
cluckjstranger in a strange land?13:58
maakuMaybe your mom is less conservative than mine :)13:58
lkjhfri just launched wget13:59
lkjhfrthis will crash the whole internet, i am sure of it13:59
Alcyiusmaaku, my mom was a pro domme13:59
AlcyiusMy mom is about as not conservative as you can get13:59
maakuMost of Heinlein's books deal with, uh, non traditional families/relationships. Stranger in a strange land in particular.14:00
maakuHeh, cool.14:00
maakuOh had to Google what that meant. Your mom must be awesome.14:01
lkjhfrwget won't do, i need something that downloads in parallel14:01
cluckjlol14:01
caternhaha14:01
caternyou're like that guy from questionable content!!14:01
AlcyiusIn some ways yes14:01
AlcyiusBut she's also a con artist and not the best mother14:01
cluckjdecent, albeit old, taste in scifi14:02
cluckjthat's good at least14:02
AlcyiusAlso, the movie lineup this year looks great14:02
AlcyiusRemakes and sequels to a bunch of classics, some new properties that are looking pretty good14:02
maakuWget can be configured to do that14:03
-!- Houshalter [~Houshalte@oh-71-50-56-224.dhcp.embarqhsd.net] has joined ##hplusroadmap14:04
cluckjif you want weird families and relationships nothing can beat octavia butler's xenogenesis14:05
AlcyiusI actually14:05
AlcyiusThink I have a copy of that14:05
TMAAlcyius: Stranger from a Strange Land is remarkable piece of sci-fi even discounting the relationship matter; it's one of those books that is pleasant to read AND makes you think14:05
cluckjit's aka lilith's brood14:05
-!- xentrac [~kragen@adjuvant.canonical.org] has joined ##hplusroadmap14:05
-!- jtimon [~quassel@126.31.134.37.dynamic.jazztel.es] has joined ##hplusroadmap14:05
-!- helleshin [~talinck@66-161-138-110.ubr1.dyn.lebanon-oh.fuse.net] has joined ##hplusroadmap14:06
-!- sandeepkr [~sandeepkr@111.235.64.4] has joined ##hplusroadmap14:06
lkjhfrturns out there is less than 5000, around 4700 - there were many duplicates14:07
lkjhfraround 1500 are hosted on diyhplus and heybryan14:08
maakuTMA true of most classic Heinlein14:09
maakuWouldn't paperbot have the rest?14:10
TMAmaaku: yeah. one of the reason my friend said: "I would read a telephone directory if it were written by Heinlein"14:14
maakuHahaha. And true.14:15
cluckjlol14:15
-!- PatrickRobotham [uid18270@gateway/web/irccloud.com/x-nhgnumlamfdkwtoc] has joined ##hplusroadmap14:17
kanzurei believe i have read telephone directories that were _not_ written by heinlein. what of it?14:23
lkjhfrwho is paperbot?14:26
-!- Netsplit *.net <-> *.split quits: dpk, xrr14:27
-!- Netsplit over, joins: dpk, xrr14:28
lkjhfr2015-11-30.log:12:07 < chris_99> i think paperbot has been dead for a while?14:30
lkjhfrhttp://diyhpl.us/~bryan/papers2/paperbot/wiley-error.txt14:30
lkjhfrpaperbot has 940 pdfs, and those are included in those 1500 from diyhplus14:34
kanzurehttps://github.com/kanzure/paperbot14:35
lkjhfrah, so this means that paperbot pulled pdfs from links that didn't lead directly to pdfs14:41
lkjhfrit's still 4700 pdfs then14:42
-!- sh|tmp [~sh@89.101.222.213] has joined ##hplusroadmap14:47
-!- sh [~sh@89.101.222.213] has quit [Disconnected by services]14:47
-!- sh|tmp is now known as sh14:47
-!- Burninate [~Burn@pool-74-96-98-64.washdc.fios.verizon.net] has quit [Ping timeout: 240 seconds]14:47
-!- berndj [~berndj@azna.co.za] has quit [Ping timeout: 250 seconds]14:48
-!- Burninate [~Burn@pool-74-96-98-64.washdc.fios.verizon.net] has joined ##hplusroadmap14:48
-!- pompolic [~A@unaffiliated/pompolic] has quit [Ping timeout: 250 seconds]14:50
-!- FourFire [~fourfire@81.4.122.176] has quit [Ping timeout: 250 seconds]14:51
maakukanzure Melbourne telephone directory circa '60s would be a good one to memorize. Its where the CIA pulls their alias names from14:51
-!- |node [uid125132@gateway/web/irccloud.com/x-mqqwbdthlksjwene] has joined ##hplusroadmap14:51
-!- pompolic [~A@unaffiliated/pompolic] has joined ##hplusroadmap14:52
-!- FourFire [~fourfire@81.4.122.176] has joined ##hplusroadmap14:53
-!- Netsplit *.net <-> *.split quits: c0rw1n, nickjohnson, Diablo-D3, m0b, Aurelius_Home, AmbulatoryC0rtex14:53
-!- Netsplit over, joins: Diablo-D314:53
-!- abetusk [~abe@2601:184:4002:2290:104b:1dc8:23db:c6b1] has quit [Ping timeout: 260 seconds]14:54
-!- Netsplit over, joins: c0rw1n14:54
-!- AmbulatoryC0rtex [~Ambulator@173-31-155-69.client.mchsi.com] has joined ##hplusroadmap14:57
-!- m0b [sid24725@gateway/web/irccloud.com/x-itfegccbdizbmyvg] has joined ##hplusroadmap14:57
-!- nickjohnson [sid789@gateway/web/irccloud.com/x-wowierwclihagapo] has joined ##hplusroadmap14:57
-!- Aurelius_Home [~cpopell@c-73-200-185-48.hsd1.va.comcast.net] has joined ##hplusroadmap14:57
drethelinmaaku if that's generally known why wouldn't they change it15:04
maakuI'm sure its different now but they wouldn't burn all the old identities. Was at least the case through the 90's15:05
-!- abetusk [~abe@2601:184:4002:2290:58c4:228c:fcf7:9d03] has joined ##hplusroadmap15:07
drethelindo you really think people are maintaining the same identity for 55 years?15:07
-!- abetusk is now known as Guest4364415:07
-!- esmerelda [~andares@2607:fb90:1037:96ac:1bb0:9f55:8fbe:c705] has joined ##hplusroadmap15:10
-!- esmerelda [~andares@2607:fb90:1037:96ac:1bb0:9f55:8fbe:c705] has quit [Changing host]15:10
-!- esmerelda [~andares@unaffiliated/jacco] has joined ##hplusroadmap15:10
-!- esmerelda is now known as mabel15:10
lkjhfri am downloading ~1 pdf per second with ~80% success rate15:14
-!- berndj [~berndj@azna.co.za] has joined ##hplusroadmap15:17
lkjhfrno, thats 20% error response rate, there are also domain lookup errors and authentication failures15:22
lkjhfr50% success rate15:22
-!- CheckDavid [uid14990@gateway/web/irccloud.com/x-hnqpwtbifqfbbjzm] has quit [Quit: Connection closed for inactivity]15:29
kanzureold identities are definitely maintained. useful to have lots of built-up history.15:34
kanzuregreat now i am suspicious about all the old people i know in melbourne15:34
-!- Aurelius_Laptop [~cpopell@c-73-129-20-70.hsd1.md.comcast.net] has quit [Ping timeout: 256 seconds]15:48
lkjhfrhttp://dunno.sdf-eu.org/titles.txt16:25
lkjhfrthis is just pathetic16:25
lkjhfrand this is from 1gb of pdfs16:25
lkjhfra better idea would be to use sets of words commonly used in materials of given category to categorize it16:28
lkjhfrdo statistics of these words and check which category they match best16:29
-!- Guest43644 is now known as abetusk16:36
-!- abetusk [~abe@2601:184:4002:2290:58c4:228c:fcf7:9d03] has quit [Quit: Leaving]16:37
-!- chris_99 [~chris_99@unaffiliated/chris-99/x-3062929] has quit [Quit: Leaving]16:43
-!- Madplatypus [uid19957@gateway/web/irccloud.com/x-zkladkvhktltwwwh] has quit [Quit: Connection closed for inactivity]16:47
lkjhfr"We must incrementally build up the capabilities of intelligent systems, having complete systems at each step of the way and thus automatically ensure that the pieces and their interfaces are valid."16:51
-!- erasmus [~erasmus@unaffiliated/erasmus] has joined ##hplusroadmap16:55
lkjhfrthere is so much different stuff16:57
lkjhfrall kinds of stuff16:57
doclFourFire: the idea that you would need more than a few decades to make a dyson sphere is a persistent myth. you don't actually even need planetary disassembly as a prereq -- you could surround the sun with thin sail-like solar collectors using the mass of a large asteroid.17:01
doclIf you stuck to 1.0 AU, a sphere with the density of a thousand sheets of graphene would be the mass of Pallas.17:03
Diablo-D3well theres another thing17:06
Diablo-D3you'd spend the first decade manufacturing manufacturing.17:06
Diablo-D3then you'd build the dyson sphere on this star17:06
Diablo-D3within another decade17:06
Diablo-D3and then already have your gear headed towards the next star17:06
lkjhfran asteroid would be enough to collect the energy, not to store it17:07
Diablo-D3to only use your dyson sphere building army once17:08
Diablo-D3is a waste17:08
Diablo-D3you'd want to build the army here17:08
Diablo-D3build the first dyson sphere here17:08
Diablo-D3have the army build two or three more armies17:08
Diablo-D3and then send them off in opposite directions17:08
Diablo-D3building dyson spheres, 3-4 every decade17:09
Diablo-D3and start building larger systems to take stars apart that dont have enough matter around them to build spheres17:09
lkjhfrwhy haven't i known this channel for so long17:09
-!- Madplatypus [uid19957@gateway/web/irccloud.com/x-ddhovkgjowhlqdlo] has joined ##hplusroadmap17:09
lkjhfri have this one question i already wanted to ask a day or two ago17:10
lkjhfrhas there any research been made towards von neumann molecular machines that wouldn't be restricted by water freezing and boiling temperatures that would be suitable for outer space?17:11
Diablo-D3research? yes17:12
Diablo-D3results? not so much17:12
Diablo-D3nanoscale engineering like that is still in the future17:12
Diablo-D3and Im not sure if I can realistically say near future, or just future future17:12
Diablo-D3so Im just goign to go with future future17:12
doclI've been more focused on macroscopic von neumann systems. not that we won't ever have nano versions of that concept, but in theory we can make large scale systems using known tech.17:14
doclone variant of this would be printing everything in place using an "omnivorous" process that can work with a variety of materials.17:15
doclhttp://www.molecularassembler.com/KSRM/3.14.htm17:15
doclthe idea is essentially a large scale mass spectrometer with 3d printing capabilities.17:16
doclit wouldn't be very efficient though -- replication times were estimated at 3 years. everything gets superheated to a plasma in order to separate the elements out, which results in a lot more waste heat than the chemical processes we usually use.17:17
lkjhfrto build stuff from asteroid most useful would be some kind of asteroid eating bacteria17:19
lkjhfrbut it couldn't be earth bacteria, because these have very narrow working temperature range17:20
doclyou could put a bag around the asteroid so it holds water, bake it a little so it releases moisture, then colonize it with something. maintaining the right temperature would just be a matter of mirrors (also, water itself is something of a thermal battery since it makes ice or steam by absorbing or releasing a lot of energy)17:21
doclthere's also a process for extracting nickle using carbon monoxide, the Mond Process. https://en.wikipedia.org/wiki/Mond_process17:26
doclit can be used on iron as well. so you could get purified nickle-iron relatively easy. https://en.wikipedia.org/wiki/Carbonyl_metallurgy17:28
lkjhfrmedia often report news regarding water in space, and i just don't understand how important it is, would it really be impossible to have organism that would work without water? is the hydrogen essential somehow?17:28
doclthere's water in c-type asteroids. yes you need it for all life as we know it, but it's not that rare and is recycleable.17:31
doclc-type asteroids are essentially asphalt. so they have some rocks, and a lot of hydrocarbons. rocks are full of oxygen. so if you cook them, you get water as part of the result.17:32
lkjhfrbut the only life we know is one that we happened to live besides17:33
doclthere is a sci-fi series on syfy called The Expanse, which has a lot of political drama about water. the people living in the asteroid belt are somehow dependent on shipments of ice from Saturn. it's one of the less plausible elements of the show, since belters would necessarily live in enclosures with the ability to recycle water.17:35
doclthere could hypothetically be life based on other elements. but hydrogen has some unique properties that make it unlikely that it would be totally non-hydrogen-based. look up hydrogen bonds. those are the bonds between hydrogen-containing molecules, for example what makes ice crystals hexagonal.17:36
doclI think I've seen some proposals of NH3 (ammonia) based life, or methane based. those are extremely common forms of hydrogen in the gas giant planet moons. the gas giants themselves are of course largely made of hydrogen (as is the sun, obviously)17:38
-!- PatrickRobotham [uid18270@gateway/web/irccloud.com/x-nhgnumlamfdkwtoc] has quit [Quit: Connection closed for inactivity]17:40
doclmost plastics are hydrogen based. there are some exceptions though, such as fluorocarbons (like PTFE, aka teflon). pretty sure they use a hydrocarbon precursor to make that stuff though.17:40
lkjhfrfound it: https://en.wikipedia.org/wiki/Hypothetical_types_of_biochemistry17:42
doclnote that most of those are alternate hydrogen compounds, not non-hydrogen17:44
doclanother thing to remember is that hydrogen is very lightweight per atom. so transporting it around is not such a big deal compared to oxygen.17:46
doclthere's lots of oxygen in every asteroid, so if you have a big tank of earth-sourced hydrogen your chemical rocket or manufacturing process that needs water can still benefit a lot. 1 atom of oxygen weighs 16 times that of hydrogen, so water is 8 times as much oxygen by mass.17:48
doclI mentioned hydrogen bonds... one of the cool things about those is how much effect they have on phase change properties of chemicals. that is, for example when ice melts or water boils, the energy it soaks up or (releases when the processes are reversed). this holds to a lesser degree with other hydrogen compounds like ammonia.17:51
doclthat's why they use ammonia (NH3) in refrigerators, and why water evaporation is used for cooling nuclear plants.17:53
docldespite all that, do we really need water/hydrogen for space based manufacturing? I think probably not, if we are okay with the process being an energy hog. like I said, heat it to plasma, then you can separate it out without chemical bonds getting in the way. the final particle stream can then be directed to a cold target where it will stick.17:58
-!- esmerelda [~andares@2607:fb90:fda:85b8:709e:639:e359:ecd8] has joined ##hplusroadmap18:02
-!- esmerelda [~andares@2607:fb90:fda:85b8:709e:639:e359:ecd8] has quit [Changing host]18:02
-!- esmerelda [~andares@unaffiliated/jacco] has joined ##hplusroadmap18:02
-!- Jawmare [~Jawmare@unaffiliated/jawmare] has joined ##hplusroadmap18:05
-!- mabel [~andares@unaffiliated/jacco] has quit [Ping timeout: 260 seconds]18:07
doclI wonder what the energy cost is to sort 1 ton of random rock into its constituent elements using a mass spectrometer type mechanism?18:26
doclhttps://en.wikipedia.org/wiki/Abundance_of_elements_in_Earth%27s_crust18:28
docl300 ppm of rubidium, for example, is 300 grams. it sells for $25/gram, so that's $17500 worth.18:29
streetythat won't be evenly distributed throughout the crust though18:32
streetythe enthalpy of formation would be a good place to start for the energy cost of sorting by mass spec18:32
superkuhAnd in that other 1 ton of stuff you have lots of "toxic" stuff you have to pay >>17k for proper disposal of.18:33
superkuhKind of the problem with rare earth mining.18:33
doclIf you have purified mercury, thorium, and so on, can you not sell them?18:38
superkuhNo one is buying thorium. Even the US DOE is trying to get rid of it's giant stockpile of refined thorium.18:41
superkuhThe one or two operating rare earth mines in the US have big problems getting rid of it. It costs them a lot.18:42
doclmaybe just mix it into another ton of silica sand in the natural proportions to turn it back into rock?18:43
superkuhRegulations do not work like that.18:45
doclcoal is something like $20/ton and yields around 20 GJ/ton. unless we are talking about a stronger chemical bond than CO2, the energy cost for separating out the elements can't be that much higher (without a rather major level of inefficiency being factored in).18:46
xentracsuperkuh: oh, that's interesting!18:46
doclso if we have to pay double to turn the toxic component back to rock, it potentially wouldn't be that big of a deal.18:46
xentracdocl: in addition to the raw energy cost, there's also an entropy cost due to the elements being diluted, but it's very small in absolute terms18:47
doclyou can also get silica sand for $50/ton, so if you were trying to produce oxygen and silicon, with an energy cost near that of $20 worth of coal, it's possible in theory. note that metallurgical grade silicon sells for $2/kg, so you could turn out $1000 worth even if it isn't ultra-pure.18:53
Diablo-D3so....18:53
Diablo-D3you're making glass?18:53
doclnah, silicon is a metal18:54
Diablo-D3yeah but you said silicon and oxygen18:54
doclglass is silicon dioxide and such18:54
doclpull the oxygen out and you get the metal18:54
doclso you get two products. maybe just release the oxygen, or maybe trap it in a tank and sell it.18:55
-!- AmbulatoryC0rtex [~Ambulator@173-31-155-69.client.mchsi.com] has quit [Quit: Leaving]18:56
docl(silica sand is silicon dioxide as well)18:56
doclgiven the above, our tech for refining materials must really suck. like, we must be getting only 2% efficiency when all is said and done.19:01
doclaltogether, there are 184 grams of rare earth elements in a given ton of earth (184 ppm). http://geology.com/usgs/ree-geology/19:06
AdrianGpretty good amount19:07
AdrianGprobably much higher proportion in trash dumps.19:07
superkuhSome of the Thorium molten salt reactor talks have a good explanation of the economics of this and why it is not feasible in the current regulatory environment.19:08
AdrianGthorium?19:09
maakuAn element that can be fissioned, is way more abundant than Uranium, requires less processing, makes less radioactive waste, but unfortunately doesn't make weapons grade material as a side product19:11
superkuhhttps://t.co/DpXsRQWal9 - Video of the SpaceX first stage rocket landing back on the drone ship. It landed... but one of the legs didn't lock. So it tipped after landing then exploded.19:11
-!- PatrickRobotham [uid18270@gateway/web/irccloud.com/x-ttozgumurzmxoamf] has joined ##hplusroadmap19:13
-!- wrldpc1 [~ben@p276100-ipngn200702kyoto.kyoto.ocn.ne.jp] has quit [Quit: wrldpc1]19:27
docla lot of rees seem to sell for less than a dollar per gram on the commodities markets, some more (scandium $15/g for example). if the average is around $1/g, 184 grams is only $184. that is a lot less valuable than the silicon or rubidium.19:27
-!- wrldpc1 [~ben@p276100-ipngn200702kyoto.kyoto.ocn.ne.jp] has joined ##hplusroadmap19:40
AdrianGmaaku: i know what thorium is. why cant molten reactors be approved?19:41
maakuI don't think the molten reactors specifically are the problem19:42
doclcosts for disposal of nuclear waste run around 200,000 euros per cubic meter. thorium is denser than water, so 12 grams is less than 12 cubic centimeters. 12 cubic centimeters of waste costs 24 eurocents to dispose of.19:42
doclhttps://en.wikipedia.org/wiki/Economics_of_nuclear_power_plants#Waste_disposal_costs19:42
AdrianGnuclear "waste"...what nonsense.19:43
AdrianGwe just need fast neutron nukes.19:44
doclfast neutrons scare people for some reason.19:48
-!- sandeepkr [~sandeepkr@111.235.64.4] has quit [Ping timeout: 260 seconds]19:51
doclone possible reason mass spec type omnivorous refining would be opposed is that it makes it too easy to enrich fissile uranium. there are around 10 ppb fissile uranium in granite, since it's 1.6 ppm uranium which is 0.7% u-235.20:02
doclassuming you need to burn 1 ton of coal per ton of refining, at $20/ton, that would cost $2000 for 10 grams worth. luckily, it is probably not possible to get anywhere near that efficient.20:03
docl*20,00020:03
-!- sandeepkr [~sandeepkr@111.235.64.4] has joined ##hplusroadmap20:04
-!- Houshalter2 [~Houshalte@oh-71-50-56-224.dhcp.embarqhsd.net] has joined ##hplusroadmap20:07
-!- Houshalter [~Houshalte@oh-71-50-56-224.dhcp.embarqhsd.net] has quit [Ping timeout: 265 seconds]20:08
AdrianGdocl: fast neutrons scare people because you can make bombs with those reactors.20:12
doclyup20:15
doclnow that you've brought up the topic of nuclear physics, I'm actually a little worried that talk of general-purpose element refining carries x-risk.20:18
doclmaybe it's the type of thing that should only be developed secretly by well-vetted teams :/20:19
-!- ArturShaik [~ArturShai@212.97.2.228] has joined ##hplusroadmap20:21
AdrianGtoo late, its out of the bag.20:33
AdrianGisrael built hundreds of nuclear bombs in a single 5-6 story building?20:33
AdrianGim quite amazed actually iran was using centrifuges. sounds almost ancient.20:34
maakuIraq was using calutrons20:37
maakuGo figure20:38
maakuAdrianG what do you mean? Israel has no bombs!20:38
AdrianGright, and im sure vanunu wasnt working on anything like that either20:48
AdrianGmust've been tried for taking selfies at work20:48
-!- superkuh [~superkuh@unaffiliated/superkuh] has quit [Read error: Connection reset by peer]20:49
-!- superkuh [~superkuh@unaffiliated/superkuh] has joined ##hplusroadmap20:50
-!- superkuh [~superkuh@unaffiliated/superkuh] has quit [Client Quit]20:52
-!- superkuh [~superkuh@unaffiliated/superkuh] has joined ##hplusroadmap20:55
-!- Gurkenglas [Gurkenglas@dslb-188-103-077-131.188.103.pools.vodafone-ip.de] has quit [Ping timeout: 260 seconds]21:04
-!- sandeepkr [~sandeepkr@111.235.64.4] has quit [Ping timeout: 272 seconds]21:14
-!- sandeepkr [~sandeepkr@111.235.64.4] has joined ##hplusroadmap21:20
-!- |node [uid125132@gateway/web/irccloud.com/x-mqqwbdthlksjwene] has quit [Quit: Connection closed for inactivity]21:21
-!- Burninate [~Burn@pool-74-96-98-64.washdc.fios.verizon.net] has quit [Ping timeout: 265 seconds]22:14
kanzure.title https://news.ycombinator.com/item?id=1092177322:16
yoleauxWebTorrent – BitTorrent over WebRTC | Hacker News22:16
-!- jtimon [~quassel@126.31.134.37.dynamic.jazztel.es] has quit [Ping timeout: 260 seconds]22:20
kanzuregrr why isn't simonster on irc this is like the one time i want to message him22:23
kanzure"Deep neural networks rival the representation of primate inferior temporal cortex for core visual object representation" http://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1003963 (2014)22:25
-!- yashgaroth [~yashgarot@2602:306:35fa:d500:f5e0:f867:a11d:8d52] has quit [Quit: Leaving]22:37
-!- juul [~juul@juul.io] has joined ##hplusroadmap22:42
kanzure"the main person that Satoshi wanted to hand this off to, if I understand correctly, was Hal Finnel. but he got ALS. and died."22:44
kanzurewelp so much for that plan22:44
-!- Burninate [~Burn@pool-108-31-204-167.washdc.fios.verizon.net] has joined ##hplusroadmap22:45
AdrianGvery unfortunate.22:48
AdrianGwe'd be better off with Hal.22:48
AdrianGhttp://gizmodo.com/an-ai-program-in-japan-just-passed-a-college-entrance-e-174275828623:10
pompolic>gawker23:18
lkjhfri think i know why that water and hydrogen is part of all life: it's needed for competition, it's allowing each cell to eat another one and to rebuild itself23:20
lkjhfrit's not worth looking for23:20
kanzuregiven sufficiently large random memory initialized with random opcodes, sequences of self-replication tend to be found with high probability, independent of details like single-protonated atoms or whatever.23:23
kanzuresee http://diyhpl.us/~bryan/papers2/ai/The%20evolution%20of%20self-replicating%20computer%20organisms%20-%20A.%20N.%20Pargellis.pdf23:24
pompolichow high?23:26
* pompolic reads23:26
-!- Gurkenglas [Gurkenglas@dslb-188-103-077-131.188.103.pools.vodafone-ip.de] has joined ##hplusroadmap23:30
maakuAdrianG: we'd be better off with the system creator not "handing off" to anyone23:37
-!- ArturShaik [~ArturShai@212.97.2.228] has quit [Ping timeout: 272 seconds]23:39
kanzuremaaku: that sounds an awful lot like a vote for the dead guy :-)23:45
-!- abetusk [~abe@2601:184:4002:2290:87e:6d5f:3cc8:8441] has joined ##hplusroadmap23:47
-!- ArturShaik [~ArturShai@37.218.164.38] has joined ##hplusroadmap23:53
-!- ArturShaik [~ArturShai@37.218.164.38] has quit [Ping timeout: 250 seconds]23:57
--- Log closed Mon Jan 18 00:00:08 2016

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