cousin_it(Vladimir Slepnev)

• cousin_it 5 Apr 2020 16:34 UTC

  2 points

  in reply to: Liron's comment on: How About a Re­mote Var­i­o­la­tion Study?

  Yes, if the po­ten­tial effect size is large, you can get away with im­pre­cise an­swers to some ques­tions. But if there are many ques­tions, at some point your “im­pre­ci­sion bud­get” will be spent. For ex­am­ple, will you be able to de­tect if your dos­ing leads to later hos­pi­tal­iza­tion in­stead of no hos­pi­tal­iza­tion? Or it weak­ens im­mu­nity in­stead of strength­en­ing it?

• cousin_it 4 Apr 2020 17:35 UTC

  7 points

  in reply to: Liron's comment on: How About a Re­mote Var­i­o­la­tion Study?

  Let’s say X% get hos­pi­tal­ized within 2 weeks. What’s the high­est value of X that would say var­i­o­la­tion is a good idea? Keep in mind that:

  • The de­mo­graph­ics of your sam­ple aren’t the same as the gen­eral pop­u­la­tion, hope­fully you didn’t in­clude many 60+ folks.

  • You don’t know how many botched the pro­to­col. Could botch in any di­rec­tion (dose too high, too low, or no dose at all).

  • You don’t know the hos­pi­tal­iza­tion rate af­ter con­tact­ing corona in nor­mal ways, which can also be low dose. Many peo­ple don’t get tested now and the epi­demic is spread­ing.

  • Etc.

• cousin_it 4 Apr 2020 13:38 UTC

  7 points

  on: How About a Re­mote Var­i­o­la­tion Study?

  1. Spain has sta­bi­lized at 7K new cases/​day, Italy at 5K new cases/​day. At this rate it will take many months to reach a sig­nifi­cant per­centage of the pop­u­la­tion. The same will prob­a­bly hap­pen in the US. Most peo­ple won’t get in­fected, so try­ing am­a­teur vac­ci­na­tion is more dan­ger­ous than do­ing noth­ing.

  2. How will you send doses to vol­un­teers? If I were a de­liv­ery com­pany, I would re­fuse to de­liver this and would call the cops.

  3. How will you mea­sure the re­sults? Peo­ple have trou­ble mea­sur­ing the death rate from corona, some­times they can’t even agree on the or­der of mag­ni­tude. It’s re­ally low and de­pends on de­mo­graphic fac­tors, en­vi­ron­ment, treat­ment and other things that aren’t well un­der­stood. If you want to mea­sure a change in that rate by look­ing at 10k re­mote vol­un­teers in rea­son­able time, I’d like to see your method­ol­ogy and er­ror bounds.

• cousin_it 28 Mar 2020 8:43 UTC

  12 points

  on: mind viruses about body viruses

  Coun­ter­point: most peo­ple who will read your post are already bet­ter than av­er­age at vet­ting-memes-be­fore-spread­ing. If you suc­ceed at mak­ing these folks even more cau­tious, ev­ery­one else in the world will still keep spread­ing un­vet­ted memes, so worse memes will win.

• cousin_it 27 Mar 2020 7:29 UTC

  6 points

  in reply to: jacobjacob's comment on: March 24th: Daily Coron­avirus Link Updates

  Wait, so your graph shows the num­ber of peo­ple hav­ing their 2-day “in­fec­tious pe­riod” at any given time, which could be much lower than the num­ber of peo­ple in­fected at a given time? That doesn’t seem to be ex­plained on the page.

  Any­way, I think the re­ally im­por­tant num­ber is how many peo­ple are hav­ing their “re­quired hos­pi­tal­iza­tion pe­riod” at any given time (which is longer than 2 days). Maybe you could show that too, since you’re already show­ing the “care ca­pac­ity” line?

• cousin_it 26 Mar 2020 18:09 UTC

  2 points

  in reply to: habryka's comment on: March 24th: Daily Coron­avirus Link Updates

  It still looks weird to me. For ex­am­ple, in Switzer­land with no miti­ga­tion it es­ti­mates 1% of peo­ple in­fected now and 3% at the peak on Apr 14, which is 2.5 weeks from now. Since each in­fec­tion lasts a cou­ple weeks or more, and there have been few deaths and re­cov­er­ies so far, that means <5% of the pop­u­la­tion will have been in­fected by that point. And then it says ac­tive in­fec­tions will start fal­ling. Why?

• cousin_it 26 Mar 2020 16:26 UTC

  3 points

  on: March 24th: Daily Coron­avirus Link Updates

  Does any­one know why the dash­board says in­fec­tions will peak at 3% if no miti­ga­tion is done?

• cousin_it 23 Mar 2020 11:26 UTC

  4 points

  on: Oc­cam’s Guillotine

  I think there are two is­sues here: 1) what are the right be­liefs to have about life 2) what’s the right emo­tional at­ti­tude to life. You paint a pic­ture of truth as a harsh de­stroyer of illu­sions, but why not de­scribe it as a source of won­der /​ beauty /​ power /​ progress in­stead?

• cousin_it 15 Mar 2020 8:04 UTC

  13 points

  on: Ra­tion­al­ists, Post-Ra­tion­al­ists, And Ra­tion­al­ist-Adjacents

  In terms of con­ver­sa­tion style, I’d define a “ra­tio­nal­ist” as some­one who’s against non-fac­tual ob­jec­tions to fac­tual claims: “you’re not an ex­pert”, “you’re mo­ti­vated to say this”, “your claim has bad con­se­quences” and so on. An in­ter­me­di­ate stage would be “grudg­ing ra­tio­nal­ist”: some­one who can re­frain from us­ing such ob­jec­tions if asked, but still listens to them, and re­lapses to us­ing them when among non-ra­tio­nal­ists.

• cousin_it 14 Mar 2020 22:13 UTC

  4 points

  in reply to: Zack_M_Davis's comment on: The ab­sur­dity of un-refer­ence­able entities

  I think Jes­sica is right on this point. Within a sys­tem like ZFC, you can’t define the sys­tem’s own defin­abil­ity pred­i­cate, so the sen­tence “there are num­bers un­defin­able in ZFC” can’t even be said, let alone proved. (Which is just as well, since ZFC has a countable model, and even a model whose ev­ery mem­ber is defin­able.) The same ap­plies to the sys­tem of ev­ery­thing you be­lieve about math, as long as it’s con­sis­tent and at least as strong as ZFC.

• cousin_it 13 Mar 2020 15:52 UTC

  2 points

  on: Puz­zles for Physicalists

  I think coun­ter­fac­tu­als only make sense when talk­ing about a part of a sys­tem from the per­spec­tive of an­other part. Maybe prob­a­bil­ities as well. Similar to how in quan­tum me­chan­ics, a sys­tem of two qubits can be in a pure state, but from the per­spec­tive of the first qubit, the sec­ond is in a mixed state.

  In this view, causal­ity/​coun­ter­fac­tu­als don’t have to be phys­i­cally fun­da­men­tal. For ex­am­ple, you can have a Game of Life world where “all causal claims re­duce to claims about state” as you say: “if X then Y” where X and Y are suc­ces­sive states. Yet it makes perfect sense for an AI in that world to use prob­a­bil­ities or coun­ter­fac­tu­als over an­other, de­mar­cated part of the world.

  There is of course a ten­sion be­tween that and log­i­cal de­ci­sion the­o­ries, but maybe that’s ok?

• cousin_it 11 Mar 2020 8:12 UTC

  LW: 2 AF: 1

  AF

  in reply to: johnswentworth's comment on: Name of Prob­lem?

  I’m ac­tu­ally not sure it’s a reg­u­lar gram­mar. Con­sider this pro­gram:

  f(n) := n+f(n-1)
  

  Which gives the tree

  n+(n-1)+((n-1)-1)+…
  

  The path from any 1 to the root con­tains a bunch of minuses, then at least as many pluses. That’s not reg­u­lar.

  So it’s prob­a­bly some other kind of gram­mar, and I don’t know if it has de­cid­able equiv­alence.

• cousin_it 10 Mar 2020 22:44 UTC

  LW: 10 AF: 5

  AF

  in reply to: johnswentworth's comment on: Name of Prob­lem?

  Ok, if we dis­al­low cy­cles of out­er­most func­tion calls, then it seems the trees are in­deed in­finite only in one di­rec­tion. Here’s a half-baked idea then: 1) in­ter­pret ev­ery path from node to root as a finite word 2) in­ter­pret the tree as a gram­mar for rec­og­niz­ing these words 3) figure out if equiv­alence of two such gram­mars is de­cid­able. For ex­am­ple, if each tree cor­re­sponds to a reg­u­lar gram­mar, then you’re in luck be­cause equiv­alence of reg­u­lar gram­mars is de­cid­able. Does that make sense?

• cousin_it 10 Mar 2020 7:57 UTC

  LW: 4 AF: 2

  AF

  in reply to: johnswentworth's comment on: Name of Prob­lem?

  Then isn’t it pos­si­ble to also have in­finite ex­pan­sions “in the mid­dle”, not only “in­side” and “out­side”? Some­thing like this:

  f(n) := f(g(n))
  g(n) := g(n+1)
  

  Maybe there’s even some way to have in­finite tow­ers of in­finite ex­pan­sions. I’m hav­ing trou­ble wrap­ping my head around this.

• cousin_it 9 Mar 2020 21:53 UTC

  LW: 3 AF: 2

  AF

  in reply to: johnswentworth's comment on: Name of Prob­lem?

  I don’t un­der­stand why the sec­ond looks like that, can you ex­plain?

• cousin_it 9 Mar 2020 21:45 UTC

  6 points

  in reply to: ErickBall's comment on: Coron­avirus Open Thread

  In their baseline sce­nario (for a small Swiss city with good in­fec­tion con­trol)

  Sad laugh. I’m in Switzer­land, we have ex­po­nen­tial growth and there’s no in­fec­tion con­trol to speak of. They just told peo­ple with non-se­vere symp­toms to not bother get­ting tested. Schools are open. Haven’t seen even one per­son wear­ing a mask.

• cousin_it 9 Mar 2020 21:40 UTC

  LW: 4 AF: 2

  AF

  on: Name of Prob­lem?

  Not sure I un­der­stand the ques­tion. Con­sider these two pro­grams:

  1. f(n) := f(n)

  2. f(n) := f(n+1)

  Which ex­pres­sion trees do they cor­re­spond to? Are these trees equiv­a­lent?

• cousin_it 7 Mar 2020 22:37 UTC

  37 points

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

  Yes, it’s im­por­tant that the pub­lic should trust au­thor­i­ties. Push­ing au­thor­i­ties to be more truth­ful helps, not harms, that goal.

• cousin_it 6 Mar 2020 10:12 UTC

  17 points

  on: Coron­avirus: Jus­tified Prac­ti­cal Ad­vice Thread

  I just thought of this idea: carry a box of dis­in­fec­tant wipes ev­ery­where, and when­ever you want to touch a sur­face that’s touched by many peo­ple (handrail, door han­dle, ele­va­tor but­ton, touch­screen kiosk...) give it a quick wipe-down first. That pro­tects you, helps pro­tect other peo­ple, shows them how to do the same, and also dis­in­fects your hands many times a day. Any prob­lems with this?

• cousin_it 3 Mar 2020 19:40 UTC

  7 points

  in reply to: steve2152's comment on: Open & Wel­come Thread—Fe­bru­ary 2020

  Look at Table 3, most peo­ple in the study re­ceived some kind of treat­ment, in par­tic­u­lar 40% re­ceived oxy­gen. You can’t figure out the un­treated fatal­ity rate from this.