Show Idle (>14 d.) Chans


← 2020-08-11 | 2020-08-13 →
02:15 gregorynyssa asciilifeform: what exactly happened during the Lisp Winter? always wondered.
02:18 gregorynyssa did the proliferation of UNIX play a role?
02:24 gregorynyssa and how do you view the purported ongoing AI Renaissance, catalyzed by DeepMind and TensorFlow?
02:26 gregorynyssa (I am familiar with Garbiel's "Good News, Bad News" essay and Quarter Century of UNIX by Salus.)
~ 8 hours 52 minutes ~
11:18 asciilifeform http://logs.nosuchlabs.com/log/asciilifeform/2020-08-12#1018670 << the answer to this q could easily fill a lengthy book, and occupy archaeologist for a lifetime. but i'ma try to give snapshot ( per asciilifeform's cosmography ) .
11:18 snsabot Logged on 2020-08-12 02:15:29 gregorynyssa: asciilifeform: what exactly happened during the Lisp Winter? always wondered.
11:22 asciilifeform gregorynyssa: if you picture tech development as a 'hill-climb' search -- there are 'peaks' that can only be reached by plowing through lotsa 'valley' by ~brute force. i.e. high-risk, ruinously-expensive experimentation.
11:24 asciilifeform in '60s-'70s, various academics in usa reported 'sexy' breakthroughs in automated problem-solving. some -- genuine, some -- arguably 'smoke & mirrors' ; some -- a combination of both.
11:24 snsabot Logged on 2020-08-11 17:21:04 asciilifeform: jurov: there were several folx from 'golden age' who seemed to be 'onto something' -- t. winograd, j. weizenbaum , m. minsky, d. lenat -- who moar or less vanished from the stage after their early 'something'. traditionally thought that 'they hit wall' or even 'was only hyped, no substance' but per asciilifeform's cosmography, they were simply victims of the 'lisp winter'.
11:24 snsabot (trilema) 2014-11-14 asciilifeform: the idea in the article can (and has) been rephrased like this. if you remove all the 'suggestively-named strings' like 'understand' - and replace with 'gensyms' - e.g., 100324 - does the resulting machine still do or even appear to do anything of interest?
11:27 asciilifeform by end of '70s became apparent that work w/ lisps would benefit from dedicated hardware.
11:30 asciilifeform the early lispm's were bulky & expensive on acct of being hand-sewn prototypes. the two firms that managed to do lispm biz for short time were lmi (dead by mid'80s) and symbolics (nasdaq:smbx, delisted in '90 iirc, existed formally until '97) , you can read about'em elsewhere in detail
11:30 snsabot (trilema) 2018-11-16 asciilifeform: ( if you were to start today with two tonnes of 74xxx, you would end up with something that looks quite a bit like cadr )
11:31 asciilifeform both faced very similar problem in 1980 -- how to go from 'two tonnes of 74xxx' to vlsi. and at ~same time~ fund refinement of os and application programs (at the time, useful software was expected to ~ship with a comp~, like you expect tyres to come w/ a new auto )
11:33 asciilifeform note that lmi & smbx were not the only firms involved -- 'thinking machines co.' (danny hillis) for instance had a audacious 65536-cpu monster, controlled from a symbolics box and programmed in an oddball multiprocessistic lisp. died with smbx (and spectacularly, they drank through the remaining money and partied like it were berlin '45, iirc discussed in earlier threads)
11:33 asciilifeform what follows applies to 'thinking machines' (and coupla others) equally as to lmi/smbx
11:36 asciilifeform in 'civilized world', if you have a 'high-risk high-expense long-term' project, there are exactly three ways to fund it. (1) a. carnegie and j. rockefeller think you have pretty face and gives you $B. (2) convince bureaucrats that your tech, if successful, will enable amazing Wunderwaffen with which surely ww3 will be won. (3) convince investors that they will get 300% in coupla yrs.
11:37 asciilifeform (1) we will disregard because it was part of largely bygone era. but included for completeness, it was how e.g. tesla was funded (westinghouse.)
11:38 asciilifeform (2) and (3) the pertinent items. it was how lmi & smbx were funded. the problem with these is that they are both ~lying contests~.
11:39 asciilifeform and if a ~better liar~ comes along, he will win, and you -- lose. the technical merits of the proj being ~irrelevant.
11:40 asciilifeform focusing on symbolics co. -- they were powered by heavy doses of 2+3 -- both extravagant promises of 'market growth' (while building costly factories to make plausible 'honest signal') and at same time gigatonnes of dough from reagan's 'star wars' et
11:40 asciilifeform c
~ 45 minutes ~
12:25 asciilifeform reagan's handouts were eventually taken over by moar effective liars. which is why e.g. 'missile defense' racket continues to this day.
12:27 asciilifeform at same time, various vendors (e.g. franz, sold lisp for sun ws) pushed 'our margarine is just as good as real butter' scamola. and fooled enuff users to burn away what little of commercial market there was for lispm
12:29 asciilifeform ( much later on, similar phenomena nailed what remained of commercial lisp market. )
12:30 asciilifeform http://logs.nosuchlabs.com/log/asciilifeform/2020-08-12#1018671 << 'long past sell-by date' shitware (incl. microshit's) simply , imho, filled vacuum left by the demise of serious computing as a field.
12:30 snsabot Logged on 2020-08-12 02:18:44 gregorynyssa: did the proliferation of UNIX play a role?
12:31 asciilifeform http://logs.nosuchlabs.com/log/asciilifeform/2020-08-12#1018672 << almost purely 'financial' bubble, with very little in the way of interesting technical artifacts to show for it.
12:31 snsabot Logged on 2020-08-12 02:24:01 gregorynyssa: and how do you view the purported ongoing AI Renaissance, catalyzed by DeepMind and TensorFlow?
12:31 gregorynyssa thanks for the overview. I knew about Symbolics vs. LMI but was lacking the big picture.
12:32 asciilifeform gregorynyssa: the much larger scale thing that happened in 1980s usa, is that there was a shift in the values of the elites. went from 'built sr-71 blackbird' to 'liquidate, merge, move 100% conveyor to asia, buy 9000x yacht'
12:32 gregorynyssa sad how many members of the old guard lost interest in LISP: Norvig, Weinreb, Gat, Stallman..
12:32 asciilifeform gregorynyssa: it's rather like to say that man who lost both legs in a wreck 'lost interest in running'
12:33 asciilifeform rms for instance lost access to lispm and convinced self that 'was for the best, now i'ma write softs for junkyard comps'
12:34 asciilifeform weinreb (rip) was shipwrecked when bolix closed doors
12:34 asciilifeform norvig left academia and whored self out to google co. as chief ideologist or what was it
12:35 asciilifeform as for gat, here's from horse's mouth.
12:37 asciilifeform as all of this went on, 'software industry' invested virtually 100% of avail. capital in ~staying maximally broken~
12:37 gregorynyssa http://logs.nosuchlabs.com/log/asciilifeform/2020-08-12#1018700 << this cultural and economic transition was described quite well in Utopia of Rules (2015) by David Graeber, and also by Galbraith's son in his foreword to the 2007 republishing of The New Industrial State.
12:37 snsabot Logged on 2020-08-12 12:32:01 asciilifeform: gregorynyssa: the much larger scale thing that happened in 1980s usa, is that there was a shift in the values of the elites. went from 'built sr-71 blackbird' to 'liquidate, merge, move 100% conveyor to asia, buy 9000x yacht'
12:37 asciilifeform e.g. the whole 'security' racket -- ultimately based on idiocies stemming from use of c/cpp, w/ overflows etc
12:38 asciilifeform gregorynyssa: aha, quite well-documented
12:49 asciilifeform !w poll
12:49 watchglass Polling 12 nodes...
12:49 watchglass 208.94.240.42:8333 : Could not connect!
12:49 watchglass 205.134.172.6:8333 : (172-6.core.ai.net) Alive: (0.082s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=643389
12:49 watchglass 205.134.172.26:8333 : Alive: (0.082s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=643389
12:49 watchglass 205.134.172.27:8333 : Alive: (0.084s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=643389 (Operator: asciilifeform)
12:49 watchglass 108.31.170.3:8333 : (pool-108-31-170-3.washdc.fios.verizon.net) Alive: (0.159s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=643389 (Operator: asciilifeform)
12:49 watchglass 205.134.172.4:8333 : (172-4.core.ai.net) Alive: (0.144s) V=70001 (/therealbitcoin.org:0.7.0.1/) Jumpers=0x1 (TRB-Compat.) Blocks=643389
12:49 watchglass 143.202.160.10:8333 : Alive: (0.168s) V=70001 (/therealbitcoin.org:0.7.0.1/) Jumpers=0x1 (TRB-Compat.) Blocks=643389
12:49 watchglass 192.151.158.26:8333 : Alive: (0.208s) V=70001 (/therealbitcoin.org:0.7.0.1/) Jumpers=0x1 (TRB-Compat.) Blocks=643389
12:49 watchglass 213.109.238.156:8333 : Alive: (0.338s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=643389
12:49 watchglass 188.121.168.69:8333 : (rev-188-121-168-69.radiolan.sk) Alive: (0.308s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=643389
12:49 watchglass 103.36.92.112:8333 : (terebe.ns01.net) Alive: (0.525s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=643389
12:50 watchglass 176.9.59.199:8333 : Busy? (No answer in 20 sec.) (Operator: jurov)
12:51 gregorynyssa concerning Naggum's bathtub-analogy -- while physics is not my forte, I am pretty sure the bathtub would be objectively faster.
12:53 asciilifeform would depend on the launcher, neh
12:54 asciilifeform but regardless of propulsion, 'splat'. which was, partly, his point, imho.
12:55 asciilifeform better analogy would have been e.g. early commercial airliners vs. current-day. in the smaller, non-pressurized machines, that had to fly 'around weather' etc., passenger certainly 'felt flight moar' than in current-day airbus etc
12:58 asciilifeform could, for instance, watch the ground 'scroll by' impressively, when flying at 1500 metres, rather than today's 15k
13:03 gregorynyssa the effect of false economy is sadly seen to be pervasive within C and C++ programs
13:04 asciilifeform gregorynyssa: it long ago went from 'mistake, like leaded petrol' to deliberate 'job-creating tech' fraud.
13:04 snsabot (agriculturalsupremacy) 2020-02-27 asciilifeform: black holes that can absorb ~infinite effort and still suck -- sell. 'job-creating technology'(tm)(r).
13:08 asciilifeform oblig. related piece re the moar general phenomenon of 'de-facto illegal to build car that lasts 50y' .
13:09 gregorynyssa I remember that essay by Orlov. free market doesn't provide sufficient incentives for good engineering or science.
13:09 asciilifeform the punishment for building mechanical products designed to last and satisfy user 100% , is bankruptcy. and similarly for building cpu+os where no buffer overflows, where software inspectable & well-documented, requiring no 'consulting contracts' , (in)security-ware, etc
13:10 gregorynyssa the LaRouche movement openly embraced the role of government in engineering and science.
13:11 asciilifeform i recall. iirc one of the folx who was briefly here ( jfw? ) grew up in that religion
13:11 gregorynyssa ahh.. interesitng.
13:12 asciilifeform and current-day governments, 'moar involved than ever' , at least if you count the paper moneys. but it ultimately buys yachts + warmed chairs, rather than sr71 or lispm.
13:14 asciilifeform consider, e.g. manhattan project -- reportedly cost ~50 bil $ in today's $. f35 -- over $trillion.
13:16 asciilifeform i dun claim to have a simple answer re why. there are 'over 9000' things you could point to, and chances are all will be correct.
13:18 asciilifeform any rational attempt at explanation imho ought to account for both 'east' and 'west'. ~same decline took place in usa and su (continuing into modern-day ru), both have their 'f35' and 0 serious breakthroughs for decades.
13:19 asciilifeform so in that light e.g. mp's 'usa declined because fathers won't belt their misbehaving children and women uppity' fails to account for su, where neither problem historically
13:20 asciilifeform seems like each particular crackpot who wakes up to civilizational decline fixates on 2-3 favourite boojums and impossible to persuade otherwise, or that anyffin else matters
13:21 gregorynyssa http://logs.nosuchlabs.com/log/asciilifeform/2020-08-12#1018740 << this passage by Dr Graeber sheds some light on this phenomenon.. my own theory which I developed before reading his work coincides with and expands upon the cultural aspects to which he is alluding.
13:21 snsabot Logged on 2020-08-12 13:12:06 asciilifeform: and current-day governments, 'moar involved than ever' , at least if you count the paper moneys. but it ultimately buys yachts + warmed chairs, rather than sr71 or lispm.
13:21 asciilifeform see also oblig.: 'social osteoporosis' hypothesis.
13:21 gregorynyssa https://pastebin.com/raw/Yk4KeUS1
13:21 * asciilifeform will read
13:22 asciilifeform gregorynyssa: did this fella not believe in paragraphs ?!
13:22 gregorynyssa the original book had paragraphs. this was taken from a processed .epub.
13:22 asciilifeform a
13:24 * gregorynyssa incidentally grew up reading classics; used to not having paragraph-breaks
13:29 gregorynyssa http://logs.nosuchlabs.com/log/asciilifeform/2020-08-12#1018744 << one of the problems with this view is that it is, in fact, quite common within America. it is the Reagan / Buckley / Cowboy / Southern position. hence, the phenomenon which it describes cannot be anything approaching ubiquitous.
13:29 snsabot Logged on 2020-08-12 13:19:30 asciilifeform: so in that light e.g. mp's 'usa declined because fathers won't belt their misbehaving children and women uppity' fails to account for su, where neither problem historically
13:37 asciilifeform gregorynyssa: the typical algo is, '1) pick yer favourite personally-annoying people 2) thatswhodunit! they killed civilization!1' roughly
13:38 asciilifeform the obvious retort, 'why did 'you' let'em do it?' is ~never said for sumreason.
13:38 gregorynyssa http://logs.nosuchlabs.com/log/asciilifeform/2020-08-12#1018748 << this posting comes from the quasi-Reagan precursor to the Alt Right. cf. F Roger Devlin et al.
13:38 snsabot Logged on 2020-08-12 13:21:15 asciilifeform: see also oblig.: 'social osteoporosis' hypothesis.
13:39 asciilifeform Steven B. Harris. what's the connection ? ( and who's devlin ? )
13:41 asciilifeform harris was a medical doctor and rather prolific usenet figure (quite a few tidbits from him in yarchive for instance)
13:42 gregorynyssa http://logs.nosuchlabs.com/log/asciilifeform/2020-08-12#1018758 << the only way to solve this puzzle is to find structural commonalities between ALL the American sub-communities. the truth is that BOTH left and right descend from the same undisclosed movement.
13:42 snsabot Logged on 2020-08-12 13:38:47 asciilifeform: the obvious retort, 'why did 'you' let'em do it?' is ~never said for sumreason.
13:42 asciilifeform gregorynyssa: or didja mean graeber ? (who i also haven't previously heard of)
13:42 gregorynyssa after slaying their shared enemy, they turned on each other, but until fairly recently, their battles still had the half-hearted nature of a family-feud.
13:43 asciilifeform gregorynyssa: my cosmography is rather different. the 'shared enemy' is alive & well, and succeeded in replacing 100% of movements with 'movements' constructed to sum to 0.
13:44 gregorynyssa F Roger Devlin was a secular, right-wing political writer who frequently resorted to Evolutionary Psychology to explain events.
13:45 asciilifeform gregorynyssa: in 1 form or another, this has gone on since darwin
13:46 gregorynyssa of course.. William James etc.. but there was a specific community of thinkers which arose in the 1970s, of which this posting by Dr Harris is reminiscent.
13:46 asciilifeform the folx butthurt from darwin, split into 2 camps -- 'it didn't happen, see here in bible' and 'it happened, but doesn't matter'
13:46 asciilifeform presently the latter is dominant
13:47 gregorynyssa I think Evolution has penetrated Ameican culture much more than people give it credit for.
13:47 gregorynyssa American *
13:48 asciilifeform considering that camps 1 and 2 are 99+% of modern-day usa, and the remainder -- crackpots in samizdat -- obviously troo
13:48 gregorynyssa if I may raise a counterexample to that..
13:49 asciilifeform hm?
13:49 gregorynyssa modern-day Americans strongly believe in the existence of Psychopaths, Sociopaths, and Narcissists, and use actively employ these concepts to explain events which happen around them. these concepts are imho. Evolutionary in nature.
13:50 gregorynyssa EDIT. "and actively employ these concepts..."
13:51 gregorynyssa someone who decisively denies Evolution would not believe in these concepts. indeed, Americans did not until ~ the Reagan era, when the country was, according to the accepted, sanitized narrative, getting more religious.
13:51 asciilifeform there's a ~bottomless pit of these models. and at the core, factual -- erry culture had a concept of folx who excel at fucking over others, breaking oaths, biting off moar than their share etc
13:52 asciilifeform occurs in some detectable form in just about all social animals.
13:53 gregorynyssa the concept of Psychopathy transcends classical ideas of good and evil persons. represents more than a difference in vocabulary.
13:54 asciilifeform simple game-theoretical concept -- 'defector'
13:54 thimbronion asciilifeform: isn't his point though that the popular aetiology for such behavior was in the past religious rather than pseudo-scientific?
13:55 asciilifeform all social creatures have some natural defense mechanisms against'em. but imperfect, unsurprise.
13:55 asciilifeform thimbronion: in the past, religious, daemonological, etc. but the phenomenon is quite real.
13:56 gregorynyssa I would argue that "evil" and "Psychopathy" are hardly the same "phenomenon," although both may involve hurting others.
13:56 asciilifeform 'evil' is the daeomonological attempt at describing the underlying concept.
13:56 gregorynyssa among other things, this disparity between "evil" and "Psychopathy" has profound effects on the way that children are raised and educated.
13:57 trinque hm, how much have you traveled in the US gregorynyssa?
13:57 * asciilifeform brb
13:57 trinque I have a hard time seeing the various regions as a coherent whole, and I suspect it was always an amalgam.
13:58 trinque I think asciilifeform has it in terms of the function of the political "division", which is entirely manufactured and a distraction.
13:59 trinque there's also deep cultural division, which is why the manufactured one works so well.
13:59 gregorynyssa http://logs.nosuchlabs.com/log/asciilifeform/2020-08-12#1018789 << hope I am not annoying you haha.. this is exactly my point, though. the public outgrew the daemonological view owing to the diffusion of Evolutionary ideas into the culture (not necessarily openly labelled as Evolutionary to avoid backlash by certain groups).
13:59 snsabot Logged on 2020-08-12 13:56:40 asciilifeform: 'evil' is the daeomonological attempt at describing the underlying concept.
13:59 shinohai My twitter experiment thus far has shown that cultural/political division is indeed powerful tool.
14:00 gregorynyssa trinque: born in America. parents were Communist defectors.
14:00 gregorynyssa * governmental defectors.
14:04 gregorynyssa http://logs.nosuchlabs.com/log/asciilifeform/2020-08-12#1018793 << I largely agree with the view of Fischer, Woodard, Garreau, et al. that America is an amalgram of N < 10 ethno-regional subcultures.
14:04 snsabot Logged on 2020-08-12 13:57:55 trinque: I have a hard time seeing the various regions as a coherent whole, and I suspect it was always an amalgam.
14:06 gregorynyssa Fischer argues N = 4. Garreau, N = 9. Woodard, N = 11 (but two of them are very small in area).
14:07 gregorynyssa let us suppose N = 4 at the time of WW2. my claim is that three of these subcultures converged upon certain socio-philosophical ideas, and /joined forces/ to fight the fourth.
14:09 trinque absent the external enemy, they've got nothing to do with each-other, plainly evident in how we've been doing since 91
14:10 gregorynyssa after the fourth subculture vanished / got absorbed, these three began bickering just as you are saying.. but the bickering was restrained among the older people, because they still had memories of the /joining of forces/ and the dreaded /fourth subculture/.
14:10 trinque could you put some labels on these numbers?
14:10 trinque who's the dreaded fourth.
14:11 trinque and yes, much of internal politics since then has been "pulling the stick" trying to conjure another enemy that'd cohere the derps.
14:11 trinque much like a cutter slices herself for a moment's mental integration
14:11 trinque absent the external pain, just parts bickering with themselves.
14:11 gregorynyssa however, the younger generation actually takes these bickerings seriously. they are not fully in the loop on the older alliance.
14:12 gregorynyssa hence, you see greater militancy nowadays on both sides.
14:12 gregorynyssa Alt Right; BLM.
14:12 trinque doesn't sound like we have exactly the same model here, just a few similar parts.
14:12 gregorynyssa the fourth subculture is the morally conservative Yankee Puritans. they dominated society until the 1960s.
14:13 gregorynyssa despite that they were numerically outnumbered several times by then.
14:13 trinque probably measuring by economic clout is a better tool than bodies.
14:13 gregorynyssa the other three subcultures: (1) academia/Feminist/SJW, (2) cowboy/Southern/Evangelical, (3) Catholic/Irish/Italian/Polish
14:16 gregorynyssa the "cowboy" subculture (non-South flyover states, basically) despite having their own culture, looked to Yankees for leadership (maybe half-heartedly), and were considered WASPs. however, with the fall of New England to (1) and (3), they didn't bother trying to restore Yankee culture, and instead merged with the South.
14:17 gregorynyssa this happened in the 60s/70s. before that, cowboys were historically enemies with the Dixies. cf. Bleeding Kansas.
14:18 gregorynyssa this cowboy-Dixie reconcilation was a big thing in the 1970s.
14:26 trinque going to have to go afk for a bit, bbl
14:28 gregorynyssa asciilifeform: fall of Yankee New England coincides with shift in mangerial culture.
14:28 gregorynyssa (the shift mentioned by Graeber et al.)
14:28 gregorynyssa trinque: np. same here. have a good night.
~ 1 hours 24 minutes ~
15:53 asciilifeform gregorynyssa: imho the 'feminists vs catholics' or 'irish vs polacks' etc of 20th c usa are simply symptomatic distractions, rather than prime movers.
15:54 asciilifeform there was a much larger-scale process that took place, that all of these were organized to smokescreen.
15:56 asciilifeform at 1 pt, industrial output and agr. production stepped up to the point where e.g. 8hr workday became a practical demand. but notice that w/ continued tech ramp-up, the 8 did not turn into 7,5,6,4.
15:56 asciilifeform instead turned into world wars. and yachts.
15:56 asciilifeform and 1e8 chair-warming 'jobs'. and 'blm' et al.
16:01 asciilifeform there is a large, brick (quite rarity in usa) house near where asciilifeform lives. owner -- died, no immediate heirs, somehow ended up in bureaucratic limbo, uninhabited, decaying lawn machinery, etc. for many yrs, i assumed it had belonged to a successful businessman, or high official. then found out from elderly townsfolks -- a schoolteacher had lived there.
16:02 asciilifeform ( for non-usa folx -- these get paid approx. what floor sweeper is paid. )
16:02 asciilifeform today you need 0.6 - 1 mil $ to buy ~cardboard and aluminum~ house on that street.
16:06 asciilifeform last ~75y of 'civilized world' was a steady march of 'work moar, get less' for virtually erryone. ( even mp whined that 'could no longer get troo silk top hat' to wear while he whipped his masochix. )
16:09 asciilifeform and just about all of the added 'work' sums to 0. the typical washington inmate 'works' effectively 12h day (if you count sitting in traffic) for his 100k$ , and the 'work' consists of warming chair. is imposed simply so that he doesn't start thinking hatethoughts.
16:10 edef hi! i'm edef, i used to read the loper-os blog when i was a teenager but have only recently rediscovered it
16:10 asciilifeform ... and so that the 'killed forest' of crapola he shits out, can be used to justify printing $trillions, so that coupla fuckwads can buy yachts. ( could ask, why not simply conscript the chair warmer to build yacht directly... )
16:11 asciilifeform welcome edef
16:11 asciilifeform edef: teenager huh. i suppose it's been around long enuff
16:11 edef yep!
16:11 verisimilitude I agree, asciilifeform.
16:11 asciilifeform edef: roughly what did you find of interest on my www ?
16:12 edef i was trying to cite your laws of personal computing and trying to figure out if there was a last name i should use to credit you when citing it
16:12 asciilifeform edef: scroll to bottom of page
16:12 edef and noticed there was an IRC link
16:12 edef спасибо
16:12 verisimilitude Say, participate in this as well, regarding the earlier discussion, edef: What's the difference between the United States and Australia?
16:12 asciilifeform edef: на здоровье
16:13 edef ^_^
16:13 asciilifeform edef: do you have a www ?
16:14 edef edef.eu states "this page intentionally left blank"
16:14 asciilifeform why not make not blank. i'll read..
16:15 verisimilitude The United States was composed of the ``religious extremists'' and criminals of a functioning society; with Australia, the ``religious extremists'' were excluded.
16:15 asciilifeform edef: the previous young fella who grew up reading asciilifeform's www, went mad and (afaik) died. i'd like to improve the record, with some survivers!
16:15 snsabot Logged on 2020-08-01 19:41:56 asciilifeform: meanwhile in vintage misc., nao browsable chronologically.
16:15 edef asciilifeform: i was writing on the bird website about signal's explicit refusal to honour the 2nd/3rd laws
16:16 verisimilitude I can only for now hope for similar results with mine, asciilifeform.
16:16 edef initial thread https://nitter.net/edefic/status/1293615935430053892 follow-up specifically citing your work https://nitter.net/edefic/status/1293639064558739456
16:17 edef asciilifeform: i was surprised to learn that Terry Davis had been reading the blog around when i was
16:18 asciilifeform davis got flattened by a train, btw, in case you didn't know
16:19 asciilifeform ( he -- also went mad, was even certified invalid w/ pension )
16:19 edef yeah, i heard the news
16:19 verisimilitude His mother recently died.
16:21 asciilifeform verisimilitude: how didja discover this ? were friends in meatspace ?
16:22 verisimilitude Anyway, I agree that those laws of sanity are valuable, and I intend to keep referencing them. They're good inspiration.
16:23 asciilifeform http://logs.nosuchlabs.com/log/asciilifeform/2020-08-12#1018857 << most importantly, 100% of these shitwares break the 7th law . which makes errything else about'em considerably less imho interesting.
16:23 snsabot Logged on 2020-08-12 16:15:55 edef: asciilifeform: i was writing on the bird website about signal's explicit refusal to honour the 2nd/3rd laws
16:23 verisimilitude Some still ``keep tabs'' on news related to Terry Davis; here's a game for TempleOS released yesterday: https://blazeitfgt.neocities.org
16:23 asciilifeform nifty
16:23 edef asciilifeform: yeah, agreed
16:24 edef asciilifeform: i'm fairly open to participating in the social contract voluntarily and with mutual agreement
16:25 edef asciilifeform: for example, i am not meaningfully opposed to Signal's timed disappearing messages — they establish a shared context where the ephemerality of the conversation is mutually understood
16:25 edef (and thus have no intention of modifying my own software to dishonour that)
16:25 asciilifeform edef: the problem is that you have no way of knowing ~what~ yer participating in.
16:26 asciilifeform when i sign a contract with e.g. datacenter, i can ~read~ the contract. when install a 500MB shitware , it can contain literally anything
16:27 edef correct, yes
16:27 edef Signal is not very large, but its source code is not very straightforwardly organised
16:28 edef however, it does adhere to the fourth law to a reasonable approximation — there is raw JavaScript i can patch, and i have done so
16:28 edef i use NixOS, so patching software i install is quite easy
16:29 asciilifeform edef: what's the attraction of that program anyway ? as i understand it is simply '9000th' clone of 'aol messenger'
16:29 edef (i assume you have played with Nix and/or NixOS — they are an interesting angle on some things i think might interest you)
16:29 edef asciilifeform: kind of, but with rather solid cryptography
16:30 edef in practice i prefer XMPP+OMEMO, because free client choice is respected in the XMPP universe
16:30 asciilifeform edef: asciilifeform re 'nixos' .
16:30 asciilifeform edef: in what sense 'solid' is this cryptography ?
16:30 edef > Not interested in running other people's pre-built binaries at all. (And that it presently cannot be fully avoided, is a problem to be dealt with, rather than an excuse to behave like a Microsoft victim.)
16:30 asciilifeform aha.
16:31 edef this is not quite true
16:31 edef the Nix way of distributing binaries is a "binary cache"
16:32 asciilifeform i'm aware
16:32 edef which is to say, it is considered a cache — from-source builds are treated as being pure functions (which works to a reasonable approximation)
16:32 edef it is fairly easy to turn off the officially-maintained cache
16:32 asciilifeform thing is, there's a reason all of my linuxen are based off 2016-era gentoo.
16:32 asciilifeform i don't want ANY trace of systemd, gtk3, and related filth, on my boxes.
16:32 asciilifeform (see linked article.)
16:33 edef ah. i don't personally mind systemd that much (it is the closest experience to illumos SMF i can get on Linux), but i do resent not having a meaningful choice there
16:33 edef i'm quite curious what you are using to appear information so quickly
16:33 asciilifeform hm?
16:34 edef you seem to be able to very quickly retrieve relevant information when i mention something you have written about in the past
16:35 asciilifeform edef: from memory. my www isn't so big.
16:36 edef my current best shot at finding IRC conversations is ripgrep through text files, i haven't got any full-text indexing on that
16:37 edef (i have written a logging IRC proxy with a friend, that i intend to extend further, but i am yet to deploy it)
16:37 asciilifeform edef: you can search more or less errything asciilifeform spoke about publicly since '13 in my irc log www.
16:37 asciilifeform ( simply hit 'search all chans' button. )
16:38 asciilifeform syntax e.g. f:asciilifeform yoursearchstring .
16:38 edef neat!
16:38 asciilifeform worx for other people's irc handles too.
16:39 asciilifeform edef: you will notice in e.g. today's #a log, there are links to discussions in e.g. 2014.
16:40 asciilifeform it is very handy to be able to do this , instead of repeating same thing 9000 times .
16:40 edef i noticed the linkification! does your IRC client have any display enhancements for that, or are you simply seeing the same text i am
16:40 asciilifeform client nope. only logger.
16:41 verisimilitude I don't see the appeal of ripgrep.
16:42 edef ripgrep outperforms most other things i have compared it to, and it handles gitignore for me
16:42 asciilifeform never even heard of it. can't fathom what it's got over grep xyz | grep pqr ...
16:42 verisimilitude It to me looks like ``Same hacksaw the from 1970s, now written in Rust!''.
16:42 edef i have fairly large source trees i want to grep through recursively
16:42 verisimilitude I loathe my current IRC setup for Freenode. This is the sole source of the errors in my writing here, know.
16:43 edef verisimilitude: it has some niceties, like gitignore support, and smart casing (ie `rg -S foo` is case-insensitive, but `rg -S Foo` is case-sensitive)
16:43 verisimilitude I don't consider grep worthy of such attention.
16:44 asciilifeform this sounds like specifically the sort of horror i carefully avoid keeping on my machines.
16:44 edef it has saved me significant time in grepping through unfamiliar codebases
16:44 asciilifeform 'smart' is deadly in software.
16:44 verisimilitude I solve that by avoiding most code written by others, edef.
16:44 edef i do not have that luxury, unfortunately
16:44 asciilifeform i prefer predictable, simple tools that existed for the most part since 1970s.
16:45 edef yeah, i can understand preferring that mode of existence
16:45 asciilifeform ( and well-documented on paper )
16:46 edef there are fairly few tools on my system with smarts i didn't pick myself, this one i explicitly prefer
16:47 edef despite this, i still habitually pass -i to disable all case-sensitivity, because years of grep use have accostumed me to paranoia about casing
16:47 asciilifeform edef: what kind of grep to use, is between you and satan. but i specifically believe that the proliferation of 'reinvented bicycles' that half-implement a standard tool that existed for generations, ultimately is large part of why 'software sucks'
16:48 edef on the whole, i have spent more time througout my life on missed search results than excessive results
16:48 edef mm, i can see that
16:48 edef i think of grep and rg serving different purposes
16:48 verisimilitude That, to me, seems to be result of spending too much time searching, then.
16:48 edef verisimilitude: i suppose so
16:49 edef verisimilitude: but part of the purpose of having an exocortex is to allow me to process large amounts of information easily
16:49 verisimilitude I seek to avoid programming without text at all; care to see the result of my work so far?
16:49 edef verisimilitude: if it comes in a raw form where i have no semantic search ability, then a good text search tool is interesting
16:49 verisimilitude Part of the purpose of UNIX is to waste as much of that as possible.
16:49 edef i am quite interested in your work and results
16:51 asciilifeform oblig. naggum re reinvention of bicycles.
16:51 asciilifeform 'It is not only that I want some stability and predictability, I want to make sure that we actually evolve. Computer Science is a field that shows some danger signs of not evolving.'
16:51 verisimilitude This is the main page for my MMC, a machine code development tool.
16:52 edef most of what i spend my time grepping for is precisely names
16:52 asciilifeform edef: you might think it is strange, but i hardly ever search -- in the grep sense -- inside code. usually i know where desired item is -- or how to find approach to it logically.
16:52 edef often a well-chosen search string gets me a lot
16:53 verisimilitude Leaving a comment for the actual website requires sending me an email, but I'll accept one over IRC PMs, as well.
16:53 edef asciilifeform: right, but i am often faced with a large codebase i do not have the time to fully acquaint myself with
16:53 edef asciilifeform: often i feel that understanding a codebase well enough to place myself in the mindset of it would contaminate my thinking with that of its authors
16:54 verisimilitude I search for symbols, because I don't know how to have SLIME bring me to the point of the definition.
16:54 asciilifeform right but what are you even doing touching a pile of garbage if 'don't want to contaminate thinking'
16:55 asciilifeform asciilifeform, for instance, makes living maintaining a large cpp codebase. and whether liked or not, had to become familiar. and at this point hardly ever need to grep etc
16:55 edef i need not agree with all the tenets of a piece of software to find it useful to use it as a reference
16:55 edef or to use it as a tool for accomplishing work
16:56 * asciilifeform finds that there is vanishingly little public soft that is worth 'as reference' at all, for anything
16:56 edef i don't particularly enjoy how my toaster works, but i also have no real interest in building toasters, although i am fairly sure i could do so with less than a week's time
16:56 edef asciilifeform: i have a fairly strong interest in Borg, and the associated suite of software
16:57 asciilifeform for instance when asciilifeform was writing ffa, found existing bignum libs to be ~worse than useless~ as references.
16:57 edef i think it gets a lot of things wrong, but also exposes a fascinating possibility-space that is worth exploring with more hindsight
16:57 asciilifeform edef: link to 'borg' ? i have never seen it
16:58 edef i don't believe in many of the implementation choices, but speaking to people who have operated such a system has taught me many things about which simplifications one can make, and which ones one cannot afford to make
16:58 edef asciilifeform: borg is Google's distributed scheduler, i can link the papers, but honestly this tweet is my favourite description of it
16:59 verisimilitude Oh, I see a mention of ngnghm, which also came to my mind when thinking of those sanity laws.
16:59 asciilifeform a, a cloudism
16:59 * asciilifeform avoids having anything to do with the subj
16:59 edef the Google Cloud offering is merely what they share with filthy externals
16:59 verisimilitude I figured this was that company's system, but refused to mention it by name here.
17:00 edef the internal system it operates on adheres to more of the Laws
17:00 asciilifeform edef: so it's a process scheduler ?
17:00 asciilifeform edef: let me ask, how much does it weigh ?
17:00 edef it's kind of a process scheduler, but it is aware of failure domains and data locality
17:01 edef the codebases involved are humongous, and i think all this could be done much more parsimoniously with the hindsight available now
17:01 asciilifeform edef: how big, do you know roughly ?
17:02 edef that number is secret, but the average individual Google machine has enough RAM and cores to need kernel patches for the latter
17:02 asciilifeform cuz, for instance, the original bell labs unix, with ~all~ utils, was approx 10 thousand lines .
17:02 edef oh, the codebase
17:02 edef i genuinely don't know
17:02 edef and it's hard to delineate
17:03 edef would you include the paxos libraries it relies on?
17:03 asciilifeform include everything.
17:03 asciilifeform edef: do you see where this is going ?
17:03 edef would we measure the transitive closure of borg alone, or do we include the entire monorepo?
17:04 asciilifeform everything that 1) it needs to build 2) isn't part of a paper standard (e.g. posix)
17:04 edef oh, no, i agree these systems are no doubt full of inessential complexity
17:04 asciilifeform not simply 'accidentally filled up with complexity.' they are products of people who literally don't give a fuck about complexity; in fact it is a kind of religion for them.
17:05 edef i'm not quite sure of that, i think there is a split between people who genuinely don't care, and people who consider it as the cost of doing business
17:05 verisimilitude There's a league of difference between ``We can only do things this way with these complications.'' and ``We can get the same end-result with less.''.
17:05 asciilifeform their productivity is measured in 'lines written'
17:06 edef asciilifeform: google does not adhere to that one, and is fairly with you on the cost of code
17:06 edef re: standards, there are standards for this within google that seem quite well-adhered to — i find that public standards often fall victim to Conway's law
17:06 asciilifeform edef: then why do google products contain tens of millions of loc ? while entire e.g. pdp-11 computer shipped with less than 20k.
17:06 edef i don't want to pretend to be an advocate for Google
17:06 asciilifeform don't tell me 'modern os is more useful' because it aint
17:07 edef i don't work there, i don't speak for them, i have merely studied their ways
17:07 asciilifeform i studied enough to understand that i don't want, if can avoid, to touch anything whatsoever that they touched.
17:07 asciilifeform it is like nuclear waste.
17:09 * asciilifeform never has, and never will, have any use for a million-line process scheduler, or a ten-million-line text editor, or any similar atrocity
17:09 asciilifeform when i so much as go near such thing, it is because someone is PAYING. serious money.
17:09 asciilifeform and i go near it for strictly so long as it takes to complete the job.
17:09 edef i can assure you that going near these things pays quite well
17:10 edef i have a complicated relationship to "complexity"
17:10 asciilifeform edef: you will also notice that asciilifeform does not discuss the war crimes that he participates in for daily bread here, or elsewhere. because they are uninteresting, and in fact bad for health, it is bad enuff that i have to work with'em, why inflict on other people.
17:11 asciilifeform h. melville did not write about his experience being an accountant, tesla did not write about digging ditches (for years dug ditches.)
17:13 verisimilitude So, any thoughts on my MMC so far, edef?
17:13 asciilifeform when the taximeter aint running, asciilifeform does not read or write about google's shitware, or microshit's api, etc. and does not go to where people do.
17:13 edef i think it is easy to write a small tool that feels complete and simple, but forces significant complexity on the user
17:14 asciilifeform edef: most classical example of this is c programing language.
17:14 edef i would consider most concurrency frameworks to be an instance of this — they present simple abstractions that are devilishly hard to use
17:14 edef C is an excellent example also
17:14 verisimilitude Oh, for a moment I thought my MMC was accused of such.
17:15 * asciilifeform must bbl, but will come back. edef don't hesitate to speak, i'ma answer in the log later. (is large part of why public logs are valuable. can have conversations that span years, if necessary.)
17:15 edef i think that it's easy to mistake lack of inessential complexity for hiding essential complexity
17:23 verisimilitude I can't quite tell what's being discussed, edef.
~ 20 minutes ~
17:43 edef verisimilitude: there are certain tools that consider the complexity of the domain they operate in as irrelevant
17:43 edef verisimilitude: by this ignorance they do not abolish that complexity, they merely pass the buck
17:45 edef eg, a C compiler is "simple" in the sense that the description of its semantics is only simple if you do not describe it in context
17:45 edef ie in ngnghm's lingo, it is simple in the human sense — but it is not at all simple in the Houyhnhnm way
17:56 edef specifically, i consider the purpose of computing to be to allow me to move complexity out of my head and into a computer
17:57 edef if some complexity can be eliminated as being redundant or inessential in the process, then that is wonderful and a success
17:58 edef but if we say, choose to delimit the domain at "search text files", then things like smart casing move a piece of complexity within the system out of my head and into the computer
17:58 edef that piece of complexity still *exists* either way — i still somewhat reflexively type -i, because that has been burnt into my brain at this point
18:00 edef ideally i would have cheap, fuzzy search, or even something aware of synonyms and words similar in meaning, which removes the complexity of guessing what exact spelling and wording was used
18:00 edef exact search is a useful tool, and allows me to emulate that process in my brain, but rest assured, a machine will be much better at it
18:01 verisimilitude Oh yes, worse is better and all of that. I was confused primarily because I'd asked for opinions on my work. I know a fellow working on fuzzy string searching right now, if curious.
18:01 verisimilitude I've a different system I want to work on which would handle that, edef.
18:01 verisimilitude I call it ``Ellision''.
18:01 edef i should figure out sometime if i can make good fuzzy matching work for my code search tooling
18:02 verisimilitude I cover that in some detail here.
18:02 edef the underlying abstractions for ripgrep are actually quite cool, it has a regex engine based on FSAs, and the author also wrote a bunch of lovely tooling for finite state *transducers*
18:02 edef on which you can do efficient fuzzy matching quite reasonably
18:03 edef ripgrep author's blog post about finite state transducers and his implementation, with a focus on text indexing
18:04 edef verisimilitude: i am enjoying the bit about storing words, i have considered similar things
18:04 edef FSTs are kind of neat because they can exploit natural redundancies like that
18:05 edef the sealion bit really gels well with that particular aspect of FSTs
18:08 edef verisimilitude: i enjoyed that post, thank you
18:09 edef i have considered similar concepts in terms of playing with representation, and am enjoying seeing others write about it (i have never really had a blog, conversations are a much more natural medium to me)
18:12 asciilifeform edef: outta curiosity, what do you do for a living ? is someone at least paying for all the sweat with 1e7-line code bases etc ?
18:14 edef i run a company that aggregates compute capacity from various datacenters into a somewhat coherent whole, with a focus on locality in scheduling
18:14 edef we work with CDNs and cable companies and such
18:17 asciilifeform edef: maybe i misunderstand, but what means 'aggregate capacity' ? would seem to me like each particular dc has whatever fiber it has locally, no more and no less
18:17 edef that is precisely the problem!
18:17 edef like, if you are interacting with a service, you would like it to be close to you in latency
18:17 asciilifeform or is this a 'load balancing' setup, where you reroute connections for client to whichever dc has spare cycles
18:17 edef correct, yes
18:18 asciilifeform a ok
18:18 edef ideally one close to the user
18:18 edef for example, spending 100ms crossing the atlantic is rarely desirable, and certainly doesn't make for a delightful experience
18:18 asciilifeform edef: how do you stay in biz against e.g. 'amazon' ?
18:18 edef (i am ~20ms from my IRC client, and it is already driving me nuts)
18:19 edef amazon and google are the thousand-pound gorillas of the market, but they are much less agile — they build very big datacenters and focus on economies of scale within those
18:19 edef the idea being that if you want efficient or affordable orchestration, you move your programs into their datacenters
18:20 asciilifeform right, but anyone with a coupla dollars can buy cloudism from amazon, and stand up proggy in 9 countries. can do similar at yours ?
18:20 asciilifeform or do i misunderstand what means 'agile'
18:20 edef yeah, that is what we're working towards. we are currently not open to the general public yet
18:20 edef we have a simpler proposition: there are already enough servers, they merely lack a cohesive way to interact with them
18:21 edef even setting up things in 9 countries on amazon web services involves doing nine things, it involves being consciously aware of this complexity and dealing with it actively
18:21 edef i believe there are more useful things to spend one's time on
18:22 edef i would rather say "here is a program, make it accessible to the public, and run it wherever it would run best"
18:22 asciilifeform edef: so your service automatically moves process between geographies based on demand ?
18:23 asciilifeform what if there's a 100tb db in it ? it also moves ?
18:23 edef generally we duplicate processes more than moving them
18:24 asciilifeform edef: how does one sync a 100tb tb b/w 9 countries without imposing delay on user ?
18:24 edef databases are generally not suitable for that, although there are data storage mechanisms that do pervade the system
18:24 edef you cannot get 100TB to the user, so actually moving the 100TB buys you nothing, no?
18:24 edef you merely want your queries answered in a reasonable time
18:25 edef ie you want distributed caches or you want to keep data specific to a particular user close to that user
18:25 edef the principle of locality holds, it is why caches work at all
18:25 edef i think of the whole system as fairly akin to a distributed cache
18:25 asciilifeform right but say yer running a www search engine. user obviously doesn't get sent 100tb, but his query can potentially ask for any given 1kb spread out inside that 100t
18:25 feedbot http://mvdstandard.net/2020/08/ussa-military-helicopter-shot-over-virginia-with-one-wounded/ << The Montevideo Standard -- USSA: Military Helicopter Shot Over Virginia With One Wounded
18:25 asciilifeform how do you actually query the 100t ?
18:26 edef some amount of network interaction
18:26 edef the abstraction we provide is in terms of Linux containers, and a few data storage elements
18:27 edef in this respect it is somewhat similar to the extant public cloud offerings — i think Google Cloud Run is the closest approximation
18:27 asciilifeform sounds rather like amazon's db, was gonna say
18:27 edef Google Spanner is probably the closest approximation of what i would like a data store to look like
18:28 asciilifeform edef: outta curiosity, are you familiar with 'erlang' and its 'mnesia' db ?
18:28 edef i am! i think a lot of things in this space are very erlang-like
18:28 asciilifeform solved (to reasonable approximation) this problem. in 1990s.
18:28 edef yes, i think it is a very worthwhile system
18:28 asciilifeform it isn't 100% clear to asciilifeform why this bicycle has to be reinvented five times a year since, by 'over 9000' firms
18:29 edef however, many systems have been written in languages that aren't Erlang that i would like to distribute in this fashion
18:29 edef and it is less costly to reinvent this particular bicycle than all of the applications one might want to run on it (alongside retraining hundreds of thousands, if not millions of programmers)
18:30 asciilifeform you'll have to retrain them either way -- if you give'em a language where the distributed message abstraction doesn't exist as organic part of the lang, nao you have to teach'em whatever crutch you used to reimplement it, neh
18:30 edef erlang isn't very suitable for a variety of the workloads we're interested in, for example, some are related to processing high-bandwidth network flows
18:31 edef asciilifeform: it turns out that people will pay me for that particular adventure, but not for teaching them erlang
18:31 asciilifeform ericsson implemented telco exch. for a whole sweden in erlang. on largely off-shelf irons, as i understand.
18:31 asciilifeform so i dun see why intrinsically unsuited for large bw
18:31 edef but yes, i would enjoy a world where i could earn my bread improving the languages directly, and i consider this a stepping stone to doing so
18:31 edef asciilifeform: it's not intrinsically unsuited, but modern packet processing frameworks target C++ and such
18:32 asciilifeform iirc they had cpp in it for tight inner loops in places
18:32 edef asciilifeform: and i would struggle to handle 80Gbit on an Erlang-based system without writing a bunch of C++
18:32 edef most of the essential complexity of that particular class of tasks is not well-addressed by Erlang, unfortunately
18:33 asciilifeform edef: afaik it was the only system to date where transparent process migration (and real-time failover) actually 100% worked.
18:33 edef but yes, all of this is an affordance to use existing languages and tools to build for a computer system that is much different
18:34 edef hmm, i can think of others, although it is one of the few public examples. google certainly has very pleasant automated checkpointing internally (but of course, this code is secret, like borg)
18:34 asciilifeform edef: are folx buying ? or thing runs off investment atm ?
18:35 asciilifeform your service, that is
18:35 edef we spent a few years struggling to run a consulting business on the side, lacking significant revenue, and finally took investment late last year
18:35 edef it wasn't my first choice, but i am a sucker for a decent tradeoff
18:36 asciilifeform i wouldn't categorically say 'impossible to compete w/ amazon' but understand that they get iron 10x cheaper than you likely ever will. through sheer economy of scale. it's rather like trying to compete with your national electric co.
18:36 edef it is certainly more comfortable to be able to write code in an ergonomic chair, with a nice screen
18:36 asciilifeform ( by running petrol generators in your back yard )
18:36 asciilifeform wait edef , i thought you ran this company
18:36 asciilifeform rather than simply 'it sounded like ok job'
18:37 edef i'm the CTO, i have a cofounder who handles CEOing and all the business end of things
18:38 * asciilifeform fwiw played this role 3 times. each time lost shirt.
18:39 edef i've been working on the underlying ideas for ~6y, with some inspiration from working for a PaaS company along the way, and trying to found one at some point
18:40 edef back then, i spent most of my actual time doing ops work for a living, and it all felt like it could be automated much better
18:40 edef i had already experienced illumos zones, and Joyent's SmartDC built on top of it (now Joyent Triton), and figured even that could be much simpler
18:41 edef the rest of the industry took a while to catch up to that even being an interesting thing to do
18:47 edef asciilifeform: re: economy of scale, i do think aggregation is precisely where the value is
18:47 edef asciilifeform: amazon is through its scale able to bargain for hardware quite effectively
18:49 edef asciilifeform: everyone else is too small a player to bargain effectively, or have the money to spend on efficient orchestration
18:52 asciilifeform edef: that was my point. it isn't clear to me how a small firm could effectively compete in cloudism w/ it
18:52 edef many companies that have expended capital on datacenters are leaving them unutilised because they don't have the means to exploit them, and instead have towering AWS bills they are unhappy with
18:53 asciilifeform they have not only 10x cheaper per cpu cycle irons, but private fibers etc
18:53 edef because their developers want to use the orchestration tools available
18:53 asciilifeform can your firm afford to pull atlantic fiber ?
18:53 edef one of my claims is that you do not need to own that fiber
18:53 edef we don't even own a single datacenter
18:53 edef or a single physical server, for that matter
18:54 asciilifeform so what's to keep the people who do own these things, from copying your algo and selling ~same thing you sell for 1/10 the cost ?
18:54 edef many companies are forced to have datacenters anyway, some have tax breaks for building infrastructure, etc
18:54 asciilifeform ( and dun tell me 'patent' . ever litigate a patent ? )
18:54 edef antitrust law, mostly
18:54 asciilifeform sigh
18:54 edef and in general, them not willing to give away the secret sauce
18:55 edef Google sells a solution for running in your own datacenter, and it is purposely watered down as not to compete with their own offerings
18:55 edef Amazon is following a similar path
18:56 edef it turns out that people have their reasons for running their own datacenters — telcos and cable companies cannot outsource having a datacenter to google, and won't
18:57 edef i have no doubt that Google would be quite effective at running datacenters like that, but they are instead selling watered-down software to the incumbents, because they can't take over that market
18:58 asciilifeform edef: it is possible that you'll 'win lottery' and get bought. but i dun see how otherwise could profit, google/amazon will simply lift your tech and spend a coupla 100 k$ on patent fiddler to make the cribbing 'legit'.
18:58 edef oh, no, they have the tech
18:58 edef and/or could build it cheaply
18:58 edef i'm not counting on patents or anything
18:58 asciilifeform so that was my q, what ~are~ you counting on ?
18:58 edef i'm counting on their unwillingness to sell their secret sauce
18:58 edef which has held for decades
18:59 asciilifeform why would they have to sell ? they'll rent it out, just like your co, neh
18:59 edef they aren't though
18:59 edef for use in one's own datacenter, they are renting out a watered-down version
18:59 asciilifeform if your firm shows any symptoms of making a profit, they'll consider it
19:00 asciilifeform recall what happened to the original authors of 'google calendar'.
19:00 edef making them consider it is a win to me
19:00 asciilifeform only if they buy your co, neh
19:00 edef is that true?
19:00 asciilifeform otherwise your item simply becomes a google product, and you don't see a cent
19:01 edef i'm sitting in a comfortable chair and eating pleasant food, that's valuable to me right now
19:01 edef i don't have to dominate a market for all eternity for it to be worthwhile to me
19:01 asciilifeform aite. so long as you understand that you work for food. otherwise set up for very painful surprise.
19:02 edef i think that overall the market is large enough to allow for interesting smaller players, and there is plenty of appetite for buying orchestration products
19:02 asciilifeform be sure to have new job ready before the investors bail. asciilifeform once ended up with 'we'll pay you Someday'(tm) co. very similar to the one you're describing
19:03 edef i pay myself a salary i consider sufficient, and i'm counting on the lower bound on the long-term value of this being a nice line in my CV
19:03 asciilifeform and i rec to demand the food up-front, rather than 'shares %%' of a 'jam tomorrow' hypothetical food..
19:04 edef i wouldn't sign up for "hired gun for equity"
19:04 edef someone has tried to rope me into that before
19:04 asciilifeform edef: i can only presume yer an adult, and know the typical fate of firms like these.
19:05 edef i'm in my mid-twenties, but yes, i'm aware of the startup game
19:05 asciilifeform coupla % get bought, in in coupla thou. actually turns a profit, the rest end up as office furniture auctions
19:05 asciilifeform *1 in
19:05 edef something like 90% of startups fail, and i'm comfortable with those odds
19:06 edef i could make bucketloads of money working at a bigcorp where i have fewer innovations to make (or that i am allowed to even make)
19:06 edef but i'm comfortable trading a certain amount of money for a certain amount of control over my working environment
19:07 asciilifeform in asciilifeform's experience, if you 'make innovation', the usual folx will steal it (and ~not~ usually in a way where you can 'make nice line in cv' for it, either) and you end up lucky if got to keep yer shirt.
19:07 edef nobody can steal the pleasure of having done it first from me
19:08 asciilifeform sure.
19:08 asciilifeform realize also that they only buy that <1% of small co.s so that to encourage others to make what for'em to steal.
19:08 asciilifeform it's very similar to how hollywood works, where young actors are paid almost 0, but encouraged by 'stars'
19:09 edef i didn't grow up particularly wealthy, so i'm not very money-hungry about the whole thing
19:09 asciilifeform ( the term of art is 'tournament market' )
19:09 edef i'm already making a decent integer multiple of what i need to live on
19:10 asciilifeform ( see e.g. )
19:10 snsabot (trilema) 2014-06-18 asciilifeform: 'One of various ways of organizing work that economists have identified, a tournament market "offers participants the chance of winning a big prize--an independent research career, tenure, a named chair, scientific renown, awards--through competition," writes Richard Freeman and co-authors. Tournament markets amplify "small differences in productivity into large differences in recognition and reward," Freeman an
19:10 asciilifeform edef: even ordinary programmer in most of planet typically makes integer multiple of 'need to live on'.
19:10 edef sure
19:11 edef i view this at some level as subsidising research i was already interested in doing personally, and having a workforce for it
19:11 asciilifeform but ordinary programmer works 8h days, and not 14 + weekends as is commonplace in startupism
19:11 edef i have a fairly irregular sleep and productivity schedule by nature
19:12 edef i've tried working 8h days but it doesn't work for me
19:12 asciilifeform edef: if it worx for you -- wainot. asciilifeform's advice is simply to remember that yer playing a lottery, and that such work is usually good for coupla yrs , after which 'suddenly' furniture auction. or partner takes the dough and runs, also happens rather often.
19:12 edef some days i have 12h+ of productive time, most days i get a few productive hours
19:12 edef and some days it's just chronic pain and brain troubles all day
19:13 asciilifeform edef: carpal tunnel ?
19:13 edef for now i am quite happy with nice furniture and never having to visit a physical office
19:13 edef scoliosis, mostly, although the carpal tunnel flares do happen
19:13 edef my body has a somewhat unfortunate inflammation cycle, in general most parts of me hurt a decent chunk of the time
19:14 edef not working from home isn't very viable for me, but this setup is quite suited to my style of working
19:14 asciilifeform this as i understand is a commonplace plague in usa, for instance. folx play startup lottery and lose shirt, but the carpal disease stays for life..
19:15 edef yeah, i have a decently ergonomic setup, i'm hoping to improve it further over time
19:15 asciilifeform edef: i also only from home. simply because i refuse to sit in traffic for 2-3h/day of unpaid sweat.
19:16 edef ++
19:16 edef i have a fairly bad time with having people walk behind me in any way, and deal poorly with interruptions, so being able to manage my inputs is quite pleasant
19:25 mats edef: i feel obligated to plug talon voice, https://www.youtube.com/watch?v=ddFI63dgpaI&list=PL2wTcyeSmhsbPZYt65mRiKSeuq6WU-rXZ&index=2
19:27 * mats also suffers from chronic pain
19:29 edef interesting, might watch when i am less tired
19:29 edef i assume you've seen the emacs voice recognition talk, i think presented at pycon
19:31 edef PYCON US 2013 - Using Python to Code by Voice — i found the input language interesting, and am quite interested in developing short specific vocabulary of that kind of computer interactions of all kinds
19:31 edef s/of computer/for computer/
19:32 asciilifeform wb mats
19:33 asciilifeform edef: i regret to say that i cannot comment in detail re your cloudism -- asciilifeform avoids cloudism and the entire culture associated with it
19:34 asciilifeform occasionally was unable to escape contact with it, in industry. but it is specifically opposite of what normally interested in.
19:35 asciilifeform ( oblig. see also. )
19:46 asciilifeform shinohai: * btcinfobot has quit (Remote host closed the connection)
19:50 shinohai ^ doing fix to make it autojoin chans sorry for noise
19:50 shinohai $vwap
19:50 btcinfobot The 24-Hour VWAP for BTC is $ 11586.14 USD
19:53 shinohai (bot also had no mechanism to self-identify, now fixed)
20:00 asciilifeform a ok
20:04 shinohai Well Eulora seems to be coming along well http://ossasepia.com/2020/08/12/what-the-fuck-planeshift/
~ 32 minutes ~
20:36 asciilifeform shinohai: i confess, i have not followed the subj
20:37 asciilifeform not outta spite, either. i didn't follow even before declared 'enemy of the people'
20:37 shinohai I had your logotron up and clicked on #o to see if anything interesting was there. Practically ghost town now.
20:38 asciilifeform i have nfi. maybe they're all on the beach.
20:47 asciilifeform shinohai: their www also apparently last updated in june
20:56 edef asciilifeform: yeah, idk, cloud stuff is full of nonsense, but most of the work i am interested in has been taken over by it
~ 28 minutes ~
21:25 mats edef: ive seen it. talon has the lowest word error rates in software of its kind, supports novel things like a scripting language that enables productivity in e.g. a variety of programming languages (go/py/js/sql/c#/etc) and environments (chrome/vim/emacs/jetbrains/vscode/tmux/term/etc) an eyetracking device for mouse and zoom control, SotA recognition engines like fb's wav2letter/google's thing and the less advanced
21:25 mats dragon
21:29 mats using a computer isnt something i actively dread anymore, which is nice
21:31 mats ah no emacs support actually, im sure itd be easy to add though. heres a script sample if interested (its gdb): https://github.com/knausj85/knausj_talon/blob/master/apps/linux/gdb.talon
21:39 edef that looks neat
21:39 edef re: the pycon talk, i liked the specific aspect of using a constructed language
~ 37 minutes ~
22:17 mats i dunno about animal sounds, but in talon you can add to a vocab list to do whatever tasks, pops/clicks for mousing, and change the alphabet (default is 'air bat cap drum ..', but can change to nato or your own thing if you like)
22:21 mats i suspect one in two people that drive a keyboard/mouse daily will eventually benefit from such alternatives, which is why im mentioning talon here for the second time. 'optikey' (windows only) is also pretty useful with an eyetracker
22:22 mats well, maybe one in four, its really the last resort after stuff like, in no particular order, rehab, behavioral modification, surgery
~ 51 minutes ~
23:14 edef i think i'd enjoy not being forced to type
← 2020-08-11 | 2020-08-13 →