Zvi

• Zvi 2 Apr 2020 16:32 UTC

  9 points

  in reply to: Lukas_Gloor's comment on: Atari early

  I be­lieve that the poker bots met the mark for two player games and that Omaha/​7-Stud are both not much of an is­sue and wouldn’t ac­tu­ally be re­quired in any way, but ac­tu­ally win­ning the WSOP re­quires mostly win­ning at 9-10 per­son ta­bles. I do re­al­ize that there are claims they’ve been able to han­dle that, but do­ing it in per­son is… trick­ier. Prob­a­bly need an­other level of im­prove­ment be­fore they have a rea­son­able shot at win­ning.

  (Note that WSOP is a good struc­ture but even so is still pretty ran­dom, so e.g. I would win it some % of the time if I en­tered, whereas if I played the type of match they used to test the poker bots, my chances would be about ep­silon.)

Tak­ing Ini­tial Viral Load Seriously

• Zvi 31 Mar 2020 19:40 UTC

  8 points

  in reply to: AnnaSalamon's comment on: Han­son vs Mow­show­itz LiveStream De­bate: “Should we ex­pose the youth to coro­n­avirus?” (Mar 29th)

  I think there are plau­si­ble sce­nar­ios where we reach, likely not full herd im­mu­nity of ‘ev­ery­one can stop wash­ing hands now, we’re fine’ but cer­tainly lev­els that sub­stan­tially im­pact R0 and can com­bine with ‘or­di­nary-life-com­pat­i­ble’ lev­els of avoidance (e.g. hand wash­ing and face not touch­ing, mask wear­ing, in­creased work­ing from home and de­liv­ery, sports with­out crowds, less air travel, etc) to com­bine for effec­tive herd im­mu­nity. The hope is that it will also in­clude test+trace and thus be well be­low what it would take oth­er­wise. I be­lieve su­per spread­ers are a big part of the cor­rect model, and that we’re prob­a­bly already at 1%+ in­fec­tion rates for the USA and 10%+ (likely 20%+) for NYC, and that the first 20% in­fected is go­ing to knock in­fec­tion rates down 50%+ due to se­lec­tion effects (which won’t be enough on its own in NYC or other big cities, but is a huge boost even there).

  Usual warn­ings that I know noth­ing, but that’s my best guess.

  Robin I think puts sub­stan­tial prob­a­bil­ity that we end up 50%+ in­fected. I have less, but still sub­stan­tial. I do think it’s a sce­nario worth hav­ing a miti­ga­tion plan for.

Let My Peo­ple Stay Home

What is a School?

• Zvi 8 Mar 2020 14:20 UTC

  34 points

  in reply to: Davidmanheim's comment on: Cred­i­bil­ity of the CDC on SARS-CoV-2

  If and to the de­gree and in the cir­cum­stances and ways that the CDC is trust­wor­thy, I de­sire to be­lieve that the CDC is trust­wor­thy.

  If and to the de­gree and in the cir­cum­stances and ways that the CDC is un­trust­wor­thy, I de­sire to be­lieve that the CDC is un­trust­wor­thy.

  Let me not be­come at­tached to be­liefs I may not want.

  If you tell me that my state­ment that some­one else is ly­ing to us about im­por­tant fac­tual in­for­ma­tion that we need to get right in or­der to keep us and our friends and loved ones safe is true but harm­ful, and I need to delete my state­ment, be­cause it is im­por­tant that peo­ple be­lieve the ly­ing liars who are ly­ing for our own good, and I should ex­er­cise prior re­straint be­fore I point out such things?

  I too am sur­prised by this ob­jec­tion com­ing from David. But I also want to point out that it seems like it is mostly David’s ob­jec­tion, and the vast ma­jor­ity here are sup­port­ive of the post.

  It also seems like David thinks the post con­tains er­rors, and he says he would not have been any­thing like this vo­cal oth­er­wise. Ob­vi­ously we should work out quickly whether or not the post does con­tain er­rors, and cor­rect any we find.

An Open Let­ter To The Con­gre­ga­tion Re­gard­ing The Up­com­ing Holiday

Coron­avirus is Here

• Zvi 11 Feb 2020 12:31 UTC

  3 points

  in reply to: AnnaSalamon's comment on: Mazes Se­quence Roundup: Fi­nal Thoughts and Paths Forward

  Yes, I be­lieve that is mostly right. I would like to get bet­ter gears on this, though.

• Zvi 11 Feb 2020 12:29 UTC

  5 points

  in reply to: Raemon's comment on: Mazes Se­quence Roundup: Fi­nal Thoughts and Paths Forward

  Trust and loy­alty seem to me to clearly be virtues if placed wisely and in mod­er­a­tion. Like all virtues, you go too far and bad things hap­pen. In­dus­trial scal­ing gives you new ways to back­fire, but there’s cer­tainly very non-in­dus­trial ways to go way over­board on ei­ther or both. Cults can be very small.

• Zvi 11 Feb 2020 12:25 UTC

  4 points

  in reply to: iceman's comment on: Mazes Se­quence Roundup: Fi­nal Thoughts and Paths Forward

  The plan would be to do a close listen to de­ter­mine which songs are most im­por­tant; ideally one would listen to the full albums (Pie­ces of You, Spirit, This Way, 0304 and Good­bye Alice in Won­der­land, in that or­der), and in gen­eral I’ve up­dated more and more to­wards ‘re­spect the artist once you like mul­ti­ple songs of theirs, and listen to full albums in or­der’ but that’s a lot of mu­sic.

  I think this is the min­i­mum story of the jour­ney I want to talk about.

  Who Will Save Your Soul?

  Pie­ces of You

  Lit­tle Sister

  I’m Sensitive

  What’s Sim­ple is True

  Down So Long

  In­no­cence Maintained

  Life Uncommon

  Je­sus Loves You

  Serve the Ego

  This Way

  Love Me, Just Leave Me Alone

  Intuition

  Sweet Temptation

  Yes U Can

  Good­bye Alice in Wonderland

  Words Get in the Way

  Stephenville, TX

• Zvi 7 Feb 2020 1:41 UTC

  4 points

  in reply to: Raemon's comment on: Mazes Se­quence Roundup: Fi­nal Thoughts and Paths Forward

  Agree that would be cool if it was good enough that it got used. Un­known how good it would have to be be­fore it would get used.

• Zvi 7 Feb 2020 1:40 UTC

  17 points

  in reply to: korin43's comment on: Mazes Se­quence Roundup: Fi­nal Thoughts and Paths Forward

  It was never my in­tent to say that the solu­tion to mazes is less com­pe­ti­tion be­tween cor­po­ra­tions or be­tween or­ga­ni­za­tions. If you look at the solu­tions that were pro­posed, none of them were about de­creas­ing com­pe­ti­tion.

  The idea I was go­ing for in terms of mazes, as I tried to ex­plain here (it is en­tirely pos­si­ble I botched this ex­pla­na­tion on top of the ini­tial con­fu­sion), was that su­per-perfect com­pe­ti­tion be­tween peo­ple to get ahead within an or­ga­ni­za­tion (or larger sys­tem that like­wise has not enough slots for too many peo­ple) is the prob­lem here.

  I also main­tain, as a dis­tinct claim, that if we were to see true perfect or su­per-perfect com­pe­ti­tion in the over­all world, that would have some very bad effects, and that the rea­son we don’t see this is be­cause perfect com­pe­ti­tion is a re­ally weird set of as­sump­tions that are not that close to ap­ply­ing in those situ­a­tions, and I do think ex­plor­ing these things more is in­ter­est­ing in its own right but isn’t what I’m try­ing to cen­trally do. Re­lated to that I would ar­gue that no, suffi­ciently com­pet­i­tive mar­kets don’t do the thing you think they do, they do some­thing else that can in some ways and situ­a­tions be won­der­ful or even op­ti­mal, but that de­pends on what you care about and a lot of de­tail, and you can also get a big dis­aster.

  Bar­ring a mas­sive edit or ad­di­tional post cre­ation I don’t know how to do bet­ter than that in terms of re­spond­ing.

Mazes Se­quence Roundup: Fi­nal Thoughts and Paths Forward

Pro­tect­ing Large Pro­jects Against Mazedom

• Zvi 1 Feb 2020 19:43 UTC

  5 points

  in reply to: johnswentworth's comment on: Create a Full Alter­na­tive Stack

  I was in no way try­ing to dis­guise that the prob­lem of peo­ple fak­ing al­ign­ment with the stack in or­der to ex­tract re­sources is the biggest prob­lem with the pro­ject, if some­one were to ac­tu­ally try to im­ple­ment it. If I get feed­back that this wasn’t clear enough I will edit to make it more clear. And cer­tainly one does not sim­ply throw money at the prob­lem.

  So that far, fair enough.

  How­ever, this also in­cor­po­rates a num­ber of as­sump­tions, and a gen­eral view of how things func­tion, that I do not share.

  First, the idea that al­ign­ment is a sin­gu­lar prob­lem, or that it ei­ther does or does not have a solu­tion. That seems very wrong to me. Align­ment has vary­ing known solu­tions de­pend­ing on the situ­a­tion and which prices you are wiling to pay and how much you care, and varies based on what al­ign­ment you are try­ing to ver­ify. You can also at­tempt struc­ture the deal such that peo­ple that are non-al­igned (e.g. with the maze na­ture, or even not very into be­ing op­posed to it) do not want what you are offer­ing.

  I don’t think there are cheap solu­tions. And yes, even­tu­ally you will fail and have to start over, but I do think this is tractable for long enough to make a big differ­ence.

  Se­cond, the idea that if there was a solu­tion then it would be im­ple­mented be­cause it out­com­petes oth­ers just doesn’t match my model on mul­ti­ple lev­els. I don’t think it would be worth pay­ing the kind of prices the stack would be will­ing to pay, in or­der to al­ign a generic cor­po­ra­tion. It’s not even clear that this level of anti-maze would be an ad­van­tage in that spot, given the gen­eral re­ac­tion to such a thing on many lev­els and the need for deep in­ter­ac­tion with mazes. And it’s of­ten the case that there are big wins, and peo­ple just don’t know about them, or they know about them but for some rea­son don’t take them. I’ve stopped find­ing such things weird.

  You can also do it back­wards-only if you’re too scared of this—award it to peo­ple who you already are con­fi­dent in now, and don’t ex­tend it later to avoid cor­rup­tion. It would be a good start on many goals.

  In any case, yes, I have thought a lot about the prac­ti­cal prob­lems, most of which such peo­ple already face much worse in other forms, and have many many thoughts about them, and the prob­lem is hard. But not ‘give up this doesn’t ac­tu­ally help’ kinds of hard.

  Not go­ing to go deeper than that here. If I de­cide to ex­pand on the prob­lem I’ll do it with more posts (which are not cur­rently planned).

• Zvi 1 Feb 2020 18:27 UTC

  14 points

  in reply to: Vaniver's comment on: Create a Full Alter­na­tive Stack

  The point is to com­pro­mise one’s in­ter­ac­tions with so­ciety in the sense that you want to change what they are. But in this frame, the idea is that your in­ter­ac­tions were pre­vi­ously be­ing com­pro­mised by the worry that some day you may need to ex­tract money from so­ciety /​ mazes, and this seeks to pre­vent that.

  Con­sider the Thiel fel­low­ship. Yes, it helps peo­ple get their start, but their or­ders are to go out into the world and start a nor­mal busi­ness and raise money the nor­mal way. It’s bet­ter than let­ting those peo­ple go to col­lege, so yay fel­low­ship, but it’s to­tally not this thing. It was a way to let kids who knew that col­lege was a trap skip col­lege. Or at least, that’s my un­der­stand­ing.

  Thiel liter­ally pro­posed fund­ing me in the full stack way at a meet­ing—not per­son­ally for life, but for a pro­posed com­pany, which was go­ing to be biotech-re­lated so it was much closer to nor­mal pro­ce­dure. He got the logic. But when he came back to his so­cial situ­a­tion he couldn’t fol­low through. Biotech has to work this way for com­pa­nies be­cause of hold-up prob­lems and de­pen­den­cies, you agree on the later rounds in ad­vance with crite­ria for un­lock­ing them. It’s not the full full stack, but it’s the core idea that you need to be se­cure from con­cerns that would bury the real op­er­a­tion if you had to worry about them.

  Creat­ing a new en­tire com­mu­nity in a new lo­ca­tion makes perfect sense, and is one good way to con­sider im­ple­men­ta­tion.

Create a Full Alter­na­tive Stack

• Zvi 30 Jan 2020 20:49 UTC

  9 points

  in reply to: Wei_Dai's comment on: Have epistemic con­di­tions always been this bad?

  Dr Schrecker tells us that as a pro­fes­sional his­to­rian she “feels com­fortable” with the ev­i­dence that it is true.

  I love this line. It could not have come more straight out of Mo­ral Mazes.

• Zvi 29 Jan 2020 21:07 UTC

  2 points

  in reply to: Bucky's comment on: Po­ten­tial Ways to Fight Mazes

  You are not wrong. Fixed in origi­nal. Mods please reim­port.