Vaniver

• Vaniver 10 Aug 2020 18:25 UTC

  15 points

  on: Fan­tasy-For­bid­ding Ex­pert Opinion

  If X had been proven good for your health with lit­tle room for doubt, it would have reached the ears of me or my peers, be­cause I can’t imag­ine some­thing be­ing definitely good for ev­ery­one and not be­ing adopted into stan­dard health prac­tices.

  You might be in­ter­ested in Inad­e­quate Equil­ibria, which is a book-length treat­ment of this sort of rea­son­ing and when it is or isn’t effec­tive. While I think it does rule out, say, that Chee­rios grant you im­mor­tal­ity, I don’t think it rules out things like “SAD can be effec­tively treated by more light” or “many peo­ple can get an ex­tra hour a day by in­creas­ing their sleep qual­ity with mela­tonin” or “you should sleep with a win­dow open to have lower CO2 ex­po­sure.”

  I imag­ine that there are things that doc­tors know about, but don’t ac­tu­ally try; The Last Psy­chi­a­trist’s ex­pe­rience with Ra­machan­dran’s Mir­ror seems rele­vant here. The Epley Ma­neu­ver is taught in med­i­cal schools and the Red­dit Tin­ni­tus Cure isn’t (as far as I know), and so even pre­sented with a pa­tient with tin­ni­tus af­ter hav­ing seen that video a doc­tor might not think of sug­gest­ing it. [Of course, peo­ple also com­ment that the Red­dit Tin­ni­tus Cure doesn’t work for them, so who knows what the ac­tual effect size is, or what other parts of the tech­nique are es­sen­tial and not ad­e­quately ex­plained; one com­ment claims that you need your palms to ac­tu­ally have a tight seal with your ears, for ex­am­ple.]

  I imag­ine there are even more things that doc­tors haven’t heard of yet, ei­ther be­cause they haven’t been dis­cov­ered or haven’t been ap­plied to that prob­lem yet. (Like, even of cur­rently ap­proved drugs, do we have all the sen­si­ble off-la­bel uses mapped out?)

• Vaniver 8 Aug 2020 0:38 UTC

  3 points

  in reply to: willbradshaw’s comment on: Prop­erty as Co­or­di­na­tion Minimization

  What’s your source for this claim?

  I first heard it… in a talk, I think? Which is where I picked up the nar­rower claim of “restau­rants” over the more broad claim that you can find in the en­try on Sy­baris from A Clas­si­cal Dic­tionary, which states:

  On the other hand, great en­courage­ment was held out to all who should dis­cover any new re­fine­ment in lux­ury, the prof­its aris­ing from which were se­cured to the in­ven­tor by patent for the space of a year.

  That source I found from the Wikipe­dia page on patents, which I why I trusted my mem­ory of the talk enough to in­clude it in the OP. The other source Wikipe­dia cites is more di­rect:

  And if any con­fec­tioner or cook in­vented any pe­cu­liar and ex­cel­lent dish, no other artist was al­lowed to make this for a year; but he alone who in­vented it was en­ti­tled to all the profit to be de­rived from the man­u­fac­ture of it for that time; in or­der that oth­ers might be in­duced to labour at ex­cel­ling in such pur­suits.

• Vaniver 7 Aug 2020 21:44 UTC

  1 point

  in reply to: waveman’s comment on: The Quan­ti­ta­tive Rea­son­ing Deficit

  Here’s a table sorted for math. The US is 37th out of 78 on the list, be­low Spain and above Is­rael; the tiers are “rich Asian city-state”, “small coun­try in Asia or Europe”, and then “large coun­try in Europe or less im­pres­sive small coun­try,” and the US is low-ranked in that third tier. (The differ­ence be­tween Ja­pan and the US is smaller than the differ­ence be­tween the US and Mex­ico.)

• Vaniver 7 Aug 2020 5:36 UTC

  3 points

  on: The Quan­ti­ta­tive Rea­son­ing Deficit

  This is a large part of why COVID-19 has hit the United States so hard.

  Is the United States sig­nifi­cantly less nu­mer­ate than other coun­tries? I agree quan­ti­ta­tive rea­son­ing is good and we want more of it, but I’m not sure it’s the ma­jor con­trib­u­tor to the thing you’re try­ing to ex­plain here.

• Vaniver 6 Aug 2020 21:19 UTC

  4 points

  in reply to: Dagon’s comment on: Prop­erty as Co­or­di­na­tion Minimization

  Aban­don all hope for a bet­ter past! I un-apolo­get­i­cally pri­ori­tize the fu­ture half of my light-cone.

  Uh, does this also in­volve 2-box­ing in Trans­par­ent New­comb’s Prob­lem?

• Vaniver 6 Aug 2020 19:19 UTC

  14 points

  in reply to: Dagon’s comment on: Prop­erty as Co­or­di­na­tion Minimization

  I think pri­ori­tiz­ing wishes of the dead over the those of the liv­ing is egre­giously wrong.

  The nar­row­ing cir­cle in ac­tion!

• Vaniver 6 Aug 2020 19:06 UTC

  2 points

  in reply to: RamblinDash’s comment on: Prop­erty as Co­or­di­na­tion Minimization

  you can pretty eas­ily come to the con­clu­sion that the op­ti­mal dis­tri­bu­tion of prop­erty rights is for Jeff Be­zos and Elon Musk to own ev­ery­thing. But I don’t think that Jeff Be­zos’s life changes one whit if his wealth changed from $185b to $300b, or $100b, or $10b, or even $1b.

  So, let’s set aside the con­trac­tual ques­tion, where so­ciety lets Jeff Be­zos and Elon Musk keep their stuff be­cause so­ciety agreed to do it, and hold­ing up so­ciety’s end of the bar­gain is im­por­tant. In­stead let’s ask the ques­tion: what does so­ciety get out of Musk own­ing some­thing in­stead of some­one else own­ing some­thing?

  I ar­gue so­ciety re­ally doesn’t get much benefit from Musk eat­ing ad­di­tional cap­i­tal; like, if he buys re­ally fancy steaks or re­ally fancy yachts or what­ever, this is mostly benefit­ing him in­stead of us (the in­di­rect benefits, like be­ing able to vi­car­i­ously con­sume it on Twit­ter or Youtube or what­ever, are prob­a­bly pretty small and en­ter­tain­ers prob­a­bly have a com­par­a­tive ad­van­tage here).

  I do think so­ciety gets a sig­nifi­cant benefit from Musk own­ing ad­di­tional cap­i­tal, be­cause he turns it into busi­nesses that are plau­si­bly benefi­cial, many of which seem like the sort of vi­sion­ary long­shots that oth­er­wise wouldn’t hap­pen. Similarly for Be­zos, al­tho his fo­cus is pointed at a par­tic­u­lar com­pany. The world where Ama­zon has $100B of as­sets and Me­diocre­Corp has $100B of as­sets will be poorer than the world where Ama­zon has $175B of as­sets and Me­diocre­Corp has $25B of as­sets.

  
  

  I think a similar story goes through for many his­tor­i­cal titans. On the for-profit side, it’s easy to see the cre­ation of mas­sive for­tunes through in­creased effi­ciency; like, Ikea be­came mas­sive and Kam­prad hugely wealthy be­cause they had a bet­ter way of do­ing things than the com­pe­ti­tion. The more money Kam­prad was ‘al­lowed’ to di­rect, the more things he im­proved. On the non-profit side, it’s also easy to see ex­am­ples where they ap­plied the same effi­ciency and long-sight­ed­ness where se­lect­ing pro­grams was more difficult than ‘just writ­ing checks’. Carnegie’s de­ci­sion to fund libraries seems like a sig­nifi­cant ex­am­ple here, but prob­a­bly more cen­tral is Rock­efel­ler, him­self a fan of home­o­pathic medicine, set­ting up a med­i­cal re­search in­sti­tute that would ac­tu­ally figure out the truth in­stead of just push­ing his fa­vorite. Both of these were pi­o­neer­ing pro­jects in a way that seems easy to ‘fade into the back­ground’, in the same way that Ikea might seem like “just how fur­ni­ture is” to some­one young enough. [Both of them, of course, also have effi­ciency sto­ries be­hind how they made their wealth, but they’re dis­tant enough in the past that they might be hard to em­pathize with.]

• Vaniver 6 Aug 2020 18:38 UTC

  1 point

  in reply to: Wei_Dai’s comment on: Prop­erty as Co­or­di­na­tion Minimization

  The point is that 2 and 3 aren’t that differ­ent in terms of “cor­rup­tion”.

  My im­pres­sion is that this is mostly be­cause of ex­ter­nal com­pet­i­tive pres­sures; my im­pres­sion is that when the Hous­ing Bureau is the pri­mary source of hous­ing, it is mys­te­ri­ously the case that bet­ter con­nected peo­ple get bet­ter hous­ing. When you can buy your own house or en­ter the af­ford­able hous­ing lot­tery, most of the rich choose to buy their own house. (It might still be the case that the poli­ti­cally well-con­nected poor end up with dis­pro­por­tionately many af­ford­able hous­ing slots com­pared to the un­con­nected poor, but that’s less of a cor­rupt­ing force be­cause the stakes are smaller over­all.)

  Like, a sys­tem where peo­ple are free to co­or­di­nate at what­ever level makes lo­cal sense seems like it’s ob­vi­ously su­pe­rior, and there are ways in which hav­ing cor­po­ra­tions al­lows you to hit bet­ter points in the ‘ag­gre­gated in­di­vi­d­ual benefit minus co­or­di­na­tion cost’ space.

  Con­sider if there is in fact a bunch of nega­tive ex­ter­nal­ities that to­gether out­weigh the benefits of build­ing an­other floor. Without this meet­ing how would all those af­fected peo­ple re­al­is­ti­cally co­or­di­nate (sup­pos­ing none of them in­di­vi­d­u­ally has enough in­cen­tive) to stop you?

  The ba­sic ques­tion here for me is some­thing like “rule of law” vs. “rule of men”; for ex­am­ple, Wash­ing­ton DC has the Height Act that pro­hibits build­ings above a cer­tain height (ac­tu­ally re­lated to the street width, but in gen­eral it’s about 11 sto­ries tall). This gives DC its par­tic­u­lar char­ac­ter, and en­sures the ma­jor gov­ern­ment build­ings re­main im­pres­sive com­pared to their sur­round­ings. When em­bark­ing on a con­struc­tion pro­ject in DC, there’s no ques­tion about how high the gov­ern­ment will let you build; it sim­ply won’t be above the height cap.

  Similarly, a rule that banned back­yard cot­tages in gen­eral, or third floors in gen­eral, might make sense, as would a law that caused prop­erty taxes to be pro­por­tional to de­mand on pub­lic ser­vices (like traf­fic and sewer and trash) or to be pe­ri­od­i­cally re­assessed (so that im­prove­ments in the prop­erty lead to in­creased taxes) in­stead of sim­ply re­assessed at sale. Similarly it could make sense to tax on­go­ing con­struc­tion pro­por­tional to the length of the con­struc­tion. That way the ex­ter­nal­ities would be priced in, ei­ther with a clear policy re­stric­tion or a tax based on the es­ti­mated cost.

  In­stead, there’s a sys­tem which has in­creased un­cer­tainty and co­or­di­na­tion cost. Does it make sense to can­vass your neigh­bors be­fore mak­ing a change, in or­der to re­duce their op­po­si­tion? Well, what if you’re look­ing to buy a prop­erty in or­der to im­prove it? Now pric­ing a lot be­comes much more un­cer­tain, as it also in­volves es­ti­mat­ing the de­vel­op­ment-friendli­ness of all neigh­bor­hoods in ques­tion. This also makes the rules ap­ply un­evenly be­tween peo­ple; quite pos­si­bly more at­trac­tive peo­ple have an eas­ier time con­vinc­ing the com­mi­tee and their neigh­bors to let them build than uglier peo­ple, for ex­am­ple.

• Vaniver 6 Aug 2020 0:56 UTC

  2 points

  in reply to: Dagon’s comment on: Prop­erty as Co­or­di­na­tion Minimization

  So it’s a lit­tle hard to say, be­cause most of the his­tor­i­cal ev­i­dence we have of them has them is situ­a­tions where they’re the only sta­ble prop­erty rights, and so likely they were over-uti­lized. It also seems in­flex­ible in im­por­tant ways, and so I’m more of a fan of the mod­ern Amer­i­can sys­tem of trusts.

  But the cen­tral premise—that rather than will­ing your prop­erty to peo­ple, you could will it to a pur­pose—seems pretty great, once you’ve in­cor­po­rated the les­son of lost pur­poses, and so can write a will that will fail grace­fully with time.

• Vaniver 5 Aug 2020 20:52 UTC

  2 points

  in reply to: RamblinDash’s comment on: Prop­erty as Co­or­di­na­tion Minimization

  Tho it should be noted that given the way union /​ strike law is in the US, isn’t it also the case that work­ers can close a fac­tory unilat­er­ally? [Like, even if the own­ers could find other work­ers, they’re of­ten pre­vented from be­ing able to use those other work­ers in­stead.] And it is also the case that lo­cal gov­ern­ments can close a fac­tory unilat­er­ally (as hap­pened re­cently for health rea­sons in many places).

  So it’s not ob­vi­ous to me that the own­ers are uniquely priv­ileged in this re­gard; for any deal that re­quires the con­tinued con­sent of all par­ties, any of them could back out even though it af­fects oth­ers greatly.

• Vaniver 5 Aug 2020 20:46 UTC

  2 points

  in reply to: Dagon’s comment on: Prop­erty as Co­or­di­na­tion Minimization

  I have a much harder time sup­port­ing “yay dy­nas­ties”.

  I don’t know, waqfs seem pretty great to me.

• Vaniver 5 Aug 2020 20:44 UTC

  2 points

  in reply to: RamblinDash’s comment on: Prop­erty as Co­or­di­na­tion Minimization

  Above, you seem to be ar­gu­ing that prop­erty can en­com­pass all sorts of rights to de­ter­mine things.

  Right; it’s a so­cial agree­ment, and so it could be al­tered if the rele­vant par­ties de­cide to al­ter it.

  I think you’re right to ob­ject to me call­ing it “too lit­tle prop­erty”; like, if the thing is a two-di­men­sional ob­ject, rather than say­ing “too lit­tle area” I should be more pre­cise and say it’s “too short”. That is, a ve­toc­racy is what you get when you have too much dis­tributed own­er­ship and too lit­tle con­cen­trated own­er­ship.

  You can’t get more to­tal rights.

  Seems right, al­though im­por­tantly you can max­i­mize the sum of in­di­vi­d­ual benefit minus co­or­di­na­tion costs; I think my over­all sense is that’s how you de­ter­mine what the cor­rect level of rights is, but that’s a longer ar­gu­ment (where I would mostly be lean­ing on Hayek, I sus­pect). Of course this gets you into the prob­lems in­her­ent in ag­gre­gat­ing benefit across peo­ple, and other thorny ter­ri­tory; I’m not say­ing it’s easy, just that there’s a tar­get.

  The fun­da­men­tal point of the so­cial­ist cri­tique of pri­vate prop­erty is that it as­signs more rights to rel­a­tively dis­in­ter­ested ab­sen­tee cap­i­tal-own­ers at the ex­pense of the rights that could oth­er­wise be as­signed to the per­son who is most af­fected.

  Eh, I’m not very sym­pa­thetic to this. I’ve rented some­thing like a dozen apart­ments in a dozen years; to the best of my knowl­edge, all of those places are still stand­ing and rented out to some­one else now. It seems odd to claim that when I was there for a year, my in­ter­est in the place out­weighed the land­lord’s in­ter­est, be­cause it re­quires for­get­ting about the in­ter­ests of ev­ery other renter that the land­lord con­tracted with.

• Vaniver 5 Aug 2020 18:52 UTC

  2 points

  in reply to: clearthis’s comment on: Prop­erty as Co­or­di­na­tion Minimization

  As the end of the para­graph sug­gests, prop­erty can also be im­ma­te­rial, but I agree that sen­tence should be tight­ened up a bit.

  As far as I un­der­stand, prop­erty comes with vary­ing de­grees of ex­clud­abil­ity and are some­times not ex­clud­able at all (e.g. pub­lic prop­erty, com­mon prop­erty).

  A lens on pub­lic prop­erty is that it’s where the pub­lic uses its right to ex­clude oth­ers from tak­ing own­er­ship of the thing. As Sher­lock Holmes is in the pub­lic do­main, I can’t say “Sher­lock Holmes is my IP!”, whereas I could say that about char­ac­ters I in­vent that aren’t in the pub­lic do­main. And the pub­lic do­main doesn’t just ex­tend to things that are cur­rently known; there are whole swaths of in­tel­lec­tual effort where so­ciety has de­cided dis­cov­er­ies can­not be patented.

• Vaniver 5 Aug 2020 2:12 UTC

  LW: 9 AF: 5

  AF

  in reply to: rohinmshah’s comment on: Three men­tal images from think­ing about AGI de­bate & corrigibility

  The claims “If X drifts away from cor­rigi­bil­ity along di­men­sion {N}, it will get pul­led back” are clearly struc­turally similar, and the broad basin of cor­rigi­bil­ity ar­gu­ment is meant to be an ar­gu­ment that ar­gues for all of them.

  To be clear, I think there are two very differ­ent ar­gu­ments here:

  1) If we have an AGI that is cor­rigible, it will not ran­domly drift to be not cor­rigible, be­cause it will proac­tively no­tice and cor­rect po­ten­tial er­rors or loss of cor­rigi­bil­ity.

  2) If we have an AGI that is partly cor­rigible, it will help us ‘finish up’ the defi­ni­tion of cor­rigi­bil­ity /​ edit it­self to be more cor­rigible, be­cause we want it to be more cor­rigible and it’s try­ing to do what we want.

  The first is “cor­rigi­bil­ity is a sta­ble at­trac­tor”, and I think there’s struc­tural similar­ity be­tween ar­gu­ments that differ­ent de­vi­a­tions will be cor­rected. The sec­ond is the “broad basin of cor­rigi­bil­ity”, where for any barely ac­cept­able ini­tial defi­ni­tion of “do what we want”, it will figure out that “help us find the right defi­ni­tion of cor­rigi­bil­ity and im­ple­ment it” will score highly on its ini­tial met­ric of “do what we want.”

  Like, it’s not the ar­gu­ment that cor­rigi­bil­ity is a sta­ble at­trac­tor; it’s an ar­gu­ment that cor­rigi­bil­ity is a sta­ble at­trac­tor with no nearby at­trac­tors. (At least in the di­men­sions that it’s ‘broad’ in.)

  I find it less plau­si­ble that miss­ing pieces in our defi­ni­tion of “do what we want” will be fixed in struc­turally similar ways, and I think there are prob­a­bly a lot of traps where a plau­si­ble sketch defi­ni­tion doesn’t au­to­mat­i­cally re­pair it­self. One can lean here on “barely ac­cept­able”, but I don’t find that very satis­fy­ing. [In par­tic­u­lar, it would be nice if we had a defi­ni­tion of cor­rigi­bil­ity where could look at it and say “yep, that’s the real deal or grows up to be the real deal,” tho that likely re­quires know­ing what the “real deal” is; the “broad basin” ar­gu­ment seems to me to be mean­ingful only in that it claims “some­thing that grows into the real deal is easy to find in­stead of hard to find,” and when I re­word that claim as “there aren’t any dead ends near the real deal” it seems less plau­si­ble.

  1. Why aren’t the di­men­sions sym­met­ric?

  In phys­i­cal space, gen­er­ally things are sym­met­ric be­tween swap­ping the di­men­sions around; in al­gorithm-space, that isn’t true. (Like, per­mute the weights in a layer and you get differ­ent func­tional be­hav­ior.) Thus while it’s sort of wacky in a phys­i­cal en­vi­ron­ment to say “oh yeah, df/​dx, df/​dy, and dy/​dz are all in­de­pen­dently sam­pled from a dis­tri­bu­tion” it’s less wacky to say that of neu­ral net­work weights (or the ap­pro­pri­ate medium-sized ana­log).

Prop­erty as Co­or­di­na­tion Minimization

• Vaniver 4 Aug 2020 19:10 UTC

  LW: 6 AF: 3

  AF

  on: Three men­tal images from think­ing about AGI de­bate & corrigibility

  And by the way, how do these AGIs come up with the best ar­gu­ment for their side any­way? Don’t they need to be do­ing good de­liber­a­tion in­ter­nally? If so, can’t we just have one of them de­liber­ate on the top-level ques­tion di­rectly? Or if not, do the de­baters spawn sub-de­baters re­cur­sively, or some­thing?

  This is an ar­bi­trary im­ple­men­ta­tion de­tail, and one of the mer­its of de­bate is that it lets the com­puter figure this out in­stead of re­quiring that the AGI de­signer figure this out.

• Vaniver 4 Aug 2020 19:07 UTC

  LW: 11 AF: 4

  AF

  in reply to: rohinmshah’s comment on: Three men­tal images from think­ing about AGI de­bate & corrigibility

  e.g. I could ar­gue against “1 + 1 = 2” by say­ing that it’s an in­finite con­junc­tion of “1 + 1 != 3″ AND “1 + 1 != 4” AND … and so it can’t pos­si­bly be true.

  Uh, when I learned ad­di­tion (in the foun­da­tion-of-math­e­mat­ics sense) the fact that 2 was the only pos­si­ble re­sult of 1+1 was a big part of what made it ad­di­tion /​ made ad­di­tion use­ful.

  There’s a huge struc­tural similar­ity be­tween the proof that ‘1 + 1 != 3’ and ‘1+1 != 4’; like, both are generic in­stances of the class ‘1 + 1 != n \forall n != 2’. We can in­crease the num­ber of num­bers with­out de­creas­ing the plau­si­bil­ity of this claim (like, con­sider it in Z/​4, then Z/​8, then Z/​16, then...).

  But if in­stead I make a claim of the form “I am the only per­son who uses the name ‘Vaniver’”, we don’t have the same sort of struc­tural similar­ity, and we do have to check the names of ev­ery­one else, and the more peo­ple there are, the less plau­si­ble the claim be­comes.

  Similarly, if we make an ar­gu­ment that some­thing is an at­trac­tor in N-di­men­sional space, that does ac­tu­ally grow less plau­si­ble the more di­men­sions there are, since there are more ways for the thing to have a deriva­tive that points away from the ‘at­trac­tor,’ if we think the di­men­sions aren’t all sym­met­ric. (If there’s only grav­ity, for ex­am­ple, we seem in a bet­ter po­si­tion to end up with at­trac­tors than if there’s a ran­dom force field, even in 4d, 8d, 16d, etc.; similarly if there’s a ran­dom po­ten­tial func­tion whose deriva­tive is used to com­pute the forces.)

• Vaniver 4 Aug 2020 18:50 UTC

  LW: 4 AF: 2

  AF

  in reply to: TurnTrout’s comment on: Three men­tal images from think­ing about AGI de­bate & corrigibility

  I think there ar­gu­ment might be mis­lead­ing in that lo­cal sta­bil­ity isn’t that rare in prac­tice

  Surely this de­pends on the num­ber of di­men­sions, with lo­cal sta­bil­ity be­ing rarer the more di­men­sions you have. [Hence the ar­gu­ment that, in the in­finite-di­men­sional limit, ev­ery­thing that would have been an “lo­cal min­i­mum” is in­stead a sad­dle point.]

• Vaniver 4 Aug 2020 18:48 UTC

  LW: 6 AF: 3

  AF

  on: Three men­tal images from think­ing about AGI de­bate & corrigibility

  To say “cor­rigi­bil­ity is a broad basin of at­trac­tion”, you need ALL of the fol­low­ing to be true:

  At some point, ei­ther in an in-per­son con­ver­sa­tion or a post, Paul clar­ified that ob­vi­ously it will be ‘nar­row’ in some di­men­sions and ‘broad’ in oth­ers. I think it’s not ob­vi­ous how the ge­o­met­ric in­tu­ition goes here, and this ques­tion mostly hinges on “if you have some parts of cor­rigi­bil­ity, do you get the other parts?”, to which I think “no” and Paul seems to think “yes.” [He might think some limited ver­sion of that, like “it makes it eas­ier to get the other parts.” which I still don’t buy yet.]

• Vaniver 4 Aug 2020 18:35 UTC

  7 points

  in reply to: AllAmericanBreakfast’s comment on: Unify­ing the Si­mu­lacra Definitions

  Can some­body en­lighten me as to how the “we live in the Ma­trix, and it’s in­escapable” per­spec­tive might be rea­son­able, just on an ev­ery­day lived-ex­pe­rience level?

  A lot of this de­pends on “what mat­ters” to you; how much is lunch about the food, ver­sus who you’re eat­ing it with, ver­sus what you’re talk­ing about? It seems quite easy for some­one to be in a realm where ev­ery­thing that mat­ters to them re­lates to so­cial re­al­ity in­stead of phys­i­cal re­al­ity. (This is mostly be­cause the as­pects of phys­i­cal re­al­ity that do mat­ter to them have had their var­i­ance re­duced sig­nifi­cantly, or have been de­clared not to mat­ter.)

  Much of the pan­demic re­sponse, for ex­am­ple, makes sense when you think of peo­ple act­ing in so­cial re­al­ity, and lit­tle sense when you think of peo­ple as act­ing in phys­i­cal re­al­ity.

  And, par­tic­u­larly if you’re am­bi­tious, you might care a lot about what your era re­wards. Mili­tary suc­cess? New dis­cov­er­ies? Build­ing new sys­tems? Care­ful and pa­tient ac­cu­mu­la­tion of cap­i­tal? Timely de­ci­sions? In­fluence ac­cu­mu­la­tion and de­ploy­ment? Clever ar­gu­men­ta­tion?

  On that front, I feel con­fused about whether our era is good or bad. It still seems like we’re still in one of the bet­ter eras for am­bi­tion be­ing achieved through new dis­cov­er­ies /​ build­ing new sys­tems, but also com­plaints seem valid that star­tups are too much about mar­ket­ing and get­ting money from in­vestors in­stead of build­ing a great product, and more broadly that we only have progress in bits in­stead of atoms.