maximkazhenkov

• maximkazhenkov 10 Oct 2020 15:40 UTC

  1 point

  in reply to: Viliam’s comment on: What hard sci­ence fic­tion sto­ries also got the so­cial sci­ences right?

  There is no perfect match with Bostrom’s vuln­er­a­bil­ities, be­cause the book as­sumed there was a rel­a­tively safe strat­egy: hide. If no one knows you are there, no one will at­tack you, be­cause al­though the “nukes” are cheap, they would be de­stroy­ing po­ten­tially use­ful re­sources.

  Not rele­vant, if you suc­ceed in hid­ing you sim­ply fall off the vuln­er­a­bil­ity land­scape. We only need to con­sider what hap­pens when you’ve been ex­posed. Also, whose re­sources? It’s a cos­mic com­mons, so who cares if it gets de­stroyed.

  The point of the Dark For­est hy­poth­e­sis was pre­cisely that in a world with such asym­met­ric weapons, co­or­di­na­tion is not nec­es­sary. If you naively make your­self visi­ble to thou­sand po­ten­tial en­e­mies, it is statis­ti­cally al­most cer­tain that some­one will pull the trig­ger; for what­ever rea­son.

  That’s just Type-1 vuln­er­a­ble world. No need for the con­trived ar­gu­men­ta­tion the au­thor gave.

  There is a self­ish rea­son to pull the trig­ger; any alien civ­i­liza­tion is a po­ten­tial ex­tinc­tion threat.

  Not re­ally, clean­ing up ex­tinc­tion threats is a pub­lic good that gen­er­ally tends to fall prey to Tragedy of the Com­mons. Even if you made the num­bers work out some­how—which is very difficult and re­quires cer­tain con­di­tions that the au­thor has ex­plic­itly re­futed (like the im­pos­si­bil­ity to colonize other stars or to send out spam mes­sages) - it would still not be an ex­am­ple of Moloch. It would be an ex­am­ple of pan-galac­tic co­or­di­na­tion, albeit a per­verted one.

• maximkazhenkov 10 Oct 2020 15:18 UTC

  4 points

  in reply to: Brittany’s comment on: What hard sci­ence fic­tion sto­ries also got the so­cial sci­ences right?

  Very much dis­agree. My sense is that the book se­ries is pretty mea­gre on pre­sent­ing “thought­ful hard sci­ence” as well as game the­ory and hu­man so­ciol­ogy.

  To pick the most ob­vi­ous ex­am­ple—the ti­tle of the tril­ogy* - the three body prob­lem was mis­rep­re­sented in the books as “it’s hard to find the gen­eral an­a­lytic solu­tion” in­stead of “the end state is ex­tremely sen­si­tive to changes in the ini­tial con­di­tion”, and the char­ac­ters in the book (both hu­mans and Tri­so­lar­i­ans) spend eons try­ing to solve the prob­lem math­e­mat­i­cally.

  But even if an ex­act solu­tion was found—which does ex­ist for some chaotic sys­tems like lo­gis­tic maps—it would have been use­less since the ini­tial con­di­tion can­not be known perfectly. This isn’t a minor nit­pick like the myr­iad other sci­en­tific prob­lems with the Tri­so­lar­ian sys­tem that can be more eas­ily for­given for artis­tic li­cense; this is miss­ing what chaotic sys­tems are about. Why even in­voke the three-body prob­lem other than as at­tire?

  *not tech­ni­cally the ti­tle of the book se­ries, but fre­quently referred to as such

• maximkazhenkov 2 Oct 2020 12:08 UTC

  3 points

  in reply to: Viliam’s comment on: What hard sci­ence fic­tion sto­ries also got the so­cial sci­ences right?

  Where ex­actly do you see Moloch in the books? It’s quite the op­po­site if any­thing; the ma­ture civ­i­liza­tions of the uni­verse have co­or­di­nated around clean­ing up the cos­mos of nascent civ­i­liza­tions, some­how, with­out a clear co­or­di­na­tion mechanism. Or per­haps it’s a Type-1 vuln­er­a­ble world, but it doesn’t fit well with the au­thor’s ar­gu­men­ta­tion. I’m not sure, and I’m not sure the au­thor knows ei­ther.

  I’m still a lit­tle puz­zled by all the praises for the deep game the­o­retic in­sights the book se­ries sup­pos­edly con­tains though. Maybe game the­ory as at­tire?

• maximkazhenkov 2 Oct 2020 11:47 UTC

  −1 points

  in reply to: G Gordon Worley III’s comment on: Covid 10/​1: The Long Haul

  You’re also ex­posed to all sorts of risks if you’re “be­low some wealth thresh­old where they can­not act as they would rea­son­ably like to be­cause their al­ter­na­tive is home­less­ness, mal­nour­ish­ment, etc.” even be­fore Corona came around. The situ­a­tion hasn’t changed all that much.

  But, as Elon Musk fa­mously said: “If you don’t make stuff, there is no stuff”.

• maximkazhenkov 1 Oct 2020 12:28 UTC

  0 points

  in reply to: frontier64’s comment on: The God­dess of Every­thing Else

  Spread­ing across the stars with­out num­ber sounds more like a “KILL CONSUME MULTIPLY CONQUER” thing than it sounds like an “Every­thing Else” thing. I’m miss­ing some­thing of the point here.

  Spread­ing across the stars with­out numbers

  ETA: Is the point that over time Man evolved to be what he is to­day, we have a con­cep­tion of right and wrong, and we’re the first link in the chain that ac­tu­ally cares about mak­ing sure our morals prop­a­gate for­ward as we evolve? So now the force of evolu­tion has been co-opted into spread­ing hu­man moral­ity?

  No. I recom­mend read­ing Med­i­ta­tions on Moloch first, then ev­ery­thing be­comes clear.

• maximkazhenkov 30 Sep 2020 18:22 UTC

  2 points

  in reply to: Raemon’s comment on: What hard sci­ence fic­tion sto­ries also got the so­cial sci­ences right?

  That’s a pretty ex­treme over-drama­ti­za­tion. Corona isn’t even 1% as bad.

• maximkazhenkov 30 Sep 2020 18:17 UTC

  6 points

  on: What hard sci­ence fic­tion sto­ries also got the so­cial sci­ences right?

  The Mote in God’s Eye is a pretty good ex­am­ple of so­cial sci­ence fic­tion in ad­di­tion to be­ing a great sci­ence fic­tion novel in gen­eral.

• maximkazhenkov 29 Sep 2020 11:22 UTC

  2 points

  in reply to: Ben Pace’s comment on: What hard sci­ence fic­tion sto­ries also got the so­cial sci­ences right?

  If the Coron­avirus had a 30% fatal­ity rate peo­ple would care a lot about not get­ting in­fected in the real world, too.

• maximkazhenkov 27 Sep 2020 22:41 UTC

  1 point

  in reply to: lsusr’s comment on: About a lo­cal vari­a­tion of Rock-Paper-Scis­sors and how it self-negated its own prob­le­matic dynamic

  You mean Nash equil­ibrium strat­egy? Rock-Paper-Scis­sors is a zero-sum game, so Pareto op­ti­mal is a triv­ial no­tion here.

• maximkazhenkov 27 Sep 2020 22:39 UTC

  1 point

  in reply to: KyriakosCH’s comment on: About a lo­cal vari­a­tion of Rock-Paper-Scis­sors and how it self-negated its own prob­le­matic dynamic

  Re­gard­less of what the new player does, there is no rea­son to ever play scis­sors. I don’t see any in­ter­est­ing “4-choice dy­namic” here. Per­haps you should pick a differ­ent ex­am­ple with mul­ti­ple Nash equil­ibria.

• maximkazhenkov 22 Sep 2020 1:31 UTC

  1 point

  in reply to: Diffractor’s comment on: Needed: AI in­fo­haz­ard policy

  Another ad­van­tage AI se­crecy has over nu­clear se­crecy is that there’s a lot of noise and hype these days about ML both within and out­side the com­mu­nity, mak­ing hid­ing in plain sight much eas­ier.

• maximkazhenkov 22 Sep 2020 1:27 UTC

  1 point

  on: Needed: AI in­fo­haz­ard policy

  In the midgame, it is un­likely for any given group to make it all the way to safe AGI by it­self. There­fore, safe AGI is a broad col­lec­tive effort and we should ex­pect most re­sults to be pub­lished. In the endgame, it might be­come likely for a given group to make it all the way to safe AGI. In this case, in­cen­tives for se­crecy be­come stronger.

  I’m not sure what “safe” means in this con­text, but it seems to me that pub­lish­ing safe AGI is not a threat and it’s the un­safe but po­ten­tially very ca­pa­ble AGI re­search we should worry about?

  And the state­ment “In the midgame, it is un­likely for any given group to make it all the way to safe AGI by it­self” seems a lot more du­bi­ous given re­cent de­vel­op­ments with GPT-3, at least ac­cord­ing to most of the Less­wrong com­mu­nity.

• maximkazhenkov 22 Sep 2020 0:22 UTC

  1 point

  in reply to: NathanielMRouth’s comment on: Luna First, But Not To Live There

  In­dus­tri­al­iz­ing space is de­sir­able be­cause in­dus­tri­al­iz­ing Earth has had a num­ber of nega­tive side effects on the bio­sphere, so mov­ing pro­duc­tion out­side the bio­sphere would be a pos­i­tive de­vel­op­ment. My ar­gu­ment is that the op­tion of stay­ing home is clearly eco­nom­i­cally prefer­able for now, and will be un­less we see ma­jor cost re­duc­tions in space tech­nol­ogy.

  I thought your ar­gu­ment is that we should in­dus­tri­al­ize space be­cause it’s eco­nom­i­cally vi­able?

  Put­ting that aside, en­vi­ron­men­tal­ism is just about the last rea­son for space ac­tivi­ties. Space travel has had a neg­ligible im­pact on the en­vi­ron­ment thus far only be­cause there has been so lit­tle space travel. But on a per-kilo­gram pay­load ba­sis, even as­sum­ing the clean­est metholox/​hy­drolox fuel com­po­si­tion pro­duced purely from so­lar power, the NO2 from hot ex­haust plumes and ozone-eat­ing free rad­i­cals from reen­try heat alone are enough make any en­vi­ron­men­tal­ist screech in hor­ror. You’d have to go to the far end of level 3 tech to be­gin mak­ing this ar­gu­ment, and even then it still isn’t an eco­nomic in­cen­tive. You can’t se­ri­ously dis­miss space tourism as a driver for space travel and then pro­pose en­vi­ron­men­tal­ism as an al­ter­na­tive.

  Whether SpaceX and other launch ve­hi­cle or­ga­ni­za­tions can reach the Level 2 thresh­old you de­scribe re­mains to be seen, and LVs are only part of the pric­etag. Ma­te­ri­als, equip­ment, and la­bor rep­re­sent a large seg­ment of space mis­sion cost, and un­less we can also drive those down by similar de­grees do the eco­nomics of coloniza­tion start mak­ing sense.

  Space is hard, sure, but how does that help your point ex­actly? Coloniza­tion doesn’t have to (and won’t) make eco­nomic sense. In­dus­tri­al­iza­tion does.

  Note, too, that ΔV is non-triv­ial, even when we start get­ting to high spe­cific-im­pulse tech­nolo­gies.

  Not re­ally. This isn’t rele­vant for the Moon vs Mars de­bate, but even for the outer planets I would ar­gue

  • Short travel time isn’t nec­es­sary for coloniz­ing or in­dus­tri­al­iz­ing outer planets

  • Nu­clear fu­sion can re­al­is­ti­cally go up to 500,000s Isp, dwarfing any rea­son­able re­quire­ment for travel in­side the so­lar system

  Also, all the anal­y­sis with hy­per­bolic or­bits are kind of un­nec­es­sary as the so­lar grav­ity well be­comes triv­ial for short trans­fers. You could just as well as­sume the tar­get planets to be fixed points and get the Δv re­quire­ment from dis­tance di­vided by de­sired travel time (x2 for de­cel­er­a­tion).

• maximkazhenkov 21 Sep 2020 23:34 UTC

  1 point

  in reply to: NathanielMRouth’s comment on: Luna First, But Not To Live There

  Govern­ment: Ask Kennedy

  Pri­vate sec­tor: Ask Musk

• maximkazhenkov 18 Sep 2020 5:37 UTC

  1 point

  in reply to: betulaster’s comment on: The Ax­iolog­i­cal Treadmill

  I’m still con­fused about your cri­tique, so let me ask you di­rectly: In the sce­nario out­lined by the OP, do you ex­pect hu­mans to even­tu­ally evolve to stop feel­ing pain from elec­tri­cal shocks?

• maximkazhenkov 17 Sep 2020 4:00 UTC

  1 point

  in reply to: Itsnotme’s comment on: The Ax­iolog­i­cal Treadmill

  Evolu­tion can’t dic­tate what’s harm­ful and what’s not; big­ger pea­cock tails can be sex­u­ally se­lected for un­til it is too costly for sur­vival, and an equil­ibrium sets in. In our sce­nario, since pain-in­duc­ing stim­uli are gen­er­ally bad for sur­vival, there is no se­lec­tion pres­sure to in­crease the pain thresh­old for elec­tri­cal shocks af­ter a cer­tain equil­ibrium point. Be­cause we start out with a ner­vous sys­tem that as­so­ci­ates elec­tri­cal shocks with pain, this pain be­comes a pes­simistic er­ror af­ter the equil­ibrium point and never gets fixed, i.e. hu­mans still suffer un­der elec­tri­cal shocks, just not so bad they’d rather kill them­selves.

  Suffer­ing is not rare in na­ture be­cause ac­tu­ally harm­ful things are com­mon and suffer­ing is an ad­e­quate re­sponse to them.

  Why then is it pos­si­ble to suffer pain worse than death? Why do peo­ple and an­i­mals suffer just as in­tensely be­yond their re­pro­duc­tive age?

• maximkazhenkov 17 Sep 2020 0:53 UTC

  1 point

  in reply to: betulaster’s comment on: The Ax­iolog­i­cal Treadmill

  Yes, that’s what pes­simistic er­rors are about. I’m not sure what ex­actly you’re cri­tiquing though?

• maximkazhenkov 16 Sep 2020 0:50 UTC

  3 points

  on: The Ax­iolog­i­cal Treadmill

  The ob­vi­ous rea­son that Moloch is the en­emy is that it de­stroys ev­ery­thing we value in the name of com­pe­ti­tion and sur­vival. But this is miss­ing the big­ger pic­ture.

  No, it isn’t. What do I care which val­ues evolu­tion origi­nally in­tended to al­ign us with? What do I care which di­rec­tion dys­genic pres­sure will push our val­ues in the fu­ture? Those aren’t my val­ues, and that’s all I need to know.

  After all, if you for­get to shock your­self, or choose not to, then you are im­me­di­ately kil­led. So the peo­ple in this coun­try will slowly evolve re­ward and mo­ti­va­tional sys­tems such that, from the in­side, it feels like they want to shock them­selves, in the same way (though maybe not to the same de­gree) that they want to eat.

  No, there is no se­lec­tion pres­sure to shock your­self more than the re­quired amount, any­thing be­yond that is still detri­men­tal to your re­pro­duc­tive fit­ness. Once we’ve evolved to barely tol­er­ate the pain of elec­tric shocks so as to not kill our­selves, the se­lec­tion to­wards more pain tol­er­ance stops, and peo­ple will still suffer a great deal be­cause there is no in­cen­tive for evolu­tion to fix pes­simistic er­rors. You could per­haps en­g­ineer sce­nar­ios where hu­mans will gen­uinely evolve to like a dystopia, but it cer­tainly doesn’t ap­ply to most cases, else suffer­ing should already be a rare oc­cur­rence in na­ture.

• maximkazhenkov 15 Sep 2020 10:35 UTC

  1 point

  in reply to: Donald Hobson’s comment on: Are there non-AI pro­jects fo­cused on defeat­ing Moloch globally?

  Well this re­quire­ment doesn’t ap­pear to be par­tic­u­larly stringent com­pared to the abil­ity to sup­press over­pop­u­la­tion and other dys­genic pres­sures that would be nec­es­sary for such a global so­cial sys­tem. It would have to be to­tal­i­tar­ian any­way (though not nec­es­sar­ily cen­tral­ized).

  It is also a use­ful ques­tion to ask whether there are al­ter­na­tive ex­is­ten­tial op­por­tu­ni­ties if su­per-in­tel­li­gent AI doesn’t turn out to be a thing. For me that’s the most in­trigu­ing as­pect of the FAI prob­lem; there are plenty of ex­is­ten­tial risks to go around but FAI as an ex­is­ten­tial op­por­tu­nity is unique.

• maximkazhenkov 15 Sep 2020 7:22 UTC

  1 point

  in reply to: Donald Hobson’s comment on: Are there non-AI pro­jects fo­cused on defeat­ing Moloch globally?

  • Maybe one-shot Pri­soner’s Dilemma is rare and Moloch doesn’t turn out to be a big is­sue af­ter all

  • On the other hand, per­haps the FAI solu­tion is just sweep­ing all the hard prob­lems un­der the AI-al­ign­ment rug and isn’t any more vi­able than en­g­ineer­ing a global so­cial sys­tem that is sta­ble over mil­lions of years (pos­si­bly us­ing hu­man ge­netic en­g­ineer­ing)