orthonormal

• orthonormal 4 Apr 2020 5:45 UTC

  2 points

  in reply to: orthonormal's comment on: April Coron­avirus Open Thread

  Text of the shared image; don’t say I didn’t warn you about the qual­ity of the writ­ing, and [sic] for the whole thing. It does read like it could re­ally be from an el­derly phys­i­cal ther­a­pist.

  If you end up with pul­monary symp­toms of corona virus pneu­mo­nia… there can be lethal dam­age from effu­sion (mu­cous filling lungs) or cy­tokine storm (body over-re­acts with more effu­sion.

  This kills peo­ple… ESPECIALLY when the num­ber of pa­tients is greater than the num­ber of ICU beds or ven­tila­tors. You will be left to drown in your mu­cous. That mu­cous can also be in­fected by other germs dur­ing your strug­gle. That is hap­pen­ing in Italy where they have 5x more pa­tients than they have hos­pi­tal beds. And the USA has FEWER hos­pi­tal beds per pop­u­la­tion than does Italy.

  Many years ago, phys­i­cal ther­a­pists have suc­cess­fully treated this with POSTURAL DRAINAGE… where the pa­tient is tipped over a wedge to tilt the lungs and bronchial tubes up­side down… to al­low the mu­cous to flow out, where it can be coughed out.

  Google it. It is EASY to do for your­self and your fam­ily mem­bers. Sim­ply get in po­si­tion and let it flow, helping it along with breath­ing tech­niques that em­pha­size full, pro­longed ex­hale, while puffing your cheeks and you blow out long and steady.

  Start as soon as you feel lungs get­ting filled. Don’t wait un­til you are too sick to bother. 3-5 min­utes sev­eral times per day.

  I did this in­side a nurs­ing home in VT dur­ing the 1976 flu epi­demic for res­i­dent pa­tients. We did not lose any­one, while other nurs­ing homes lost dozens. It is an old PT tech­nique that has faded away since we have ven­tila­tors and re­lated ma­chines. BUT this time, we will NOT have nearly enough ven­tila­tors, not the ICU beds where they are pro­vided.

  One easy way to get into po­si­tion is to lie over an EXERCISE BALL.

  ...and there the image ends.

• orthonormal 4 Apr 2020 5:31 UTC

  2 points

  on: April Coron­avirus Open Thread

  I’ve seen an image on so­cial me­dia that sug­gests pos­tu­ral drainage, a phys­i­cal ther­apy prac­tice used mostly for cys­tic fibro­sis, as a way to cope with COVID-19 at a sub-hos­pi­tal­iza­tion stage; the shared image sug­gests that drain­ing the mu­cus can keep a pa­tient from need­ing a ven­tila­tor. (I’ll tran­scribe the ac­tual text at­tached to the image in a sub­thread, but it’s of pretty low qual­ity; I’ve writ­ten here what I think is the only in­ter­est­ing point.)

  Un­for­tu­nately, Googling “pos­tu­ral drainage coro­n­avirus” just gets me all the med­i­cal pages on pos­tu­ral drainage (be­cause they now have head­ers about coro­n­avirus).

  It’s a very cheap in­ter­ven­tion for pa­tients not on ven­tila­tors, the mechanism seems at least plau­si­ble, and it’s the sort of thing that med­i­cal pro­fes­sion­als might fail to con­sider. Is it worth tak­ing a closer look?

• orthonormal 3 Apr 2020 5:09 UTC

  6 points

  in reply to: habryka's comment on: Peter’s COVID Con­soli­dated Brief for 2 April

  Nice leg­work! It’s in­san­ity and in­com­pe­tence on the part of the ex­perts af­ter all.

• orthonormal 2 Apr 2020 20:45 UTC

  5 points

  in reply to: leggi's comment on: Tak­ing Ini­tial Viral Load Seriously

  Robin Han­son started it.

• orthonormal 2 Apr 2020 20:39 UTC

  5 points

  on: Atari early

  I think it shouldn’t mat­ter much which defi­ni­tion was used, but the World Series of Poker has one “Main Event” con­sist­ing of no-limit Texas Hold-Em, and sev­eral smaller events for the differ­ent styles of poker. I would have in­ter­preted the ques­tion as ask­ing about the Main Event only if it didn’t spec­ify.

• orthonormal 2 Apr 2020 20:23 UTC

  2 points

  on: Peter’s COVID Con­soli­dated Brief for 2 April

  [EDIT: the fol­low­ing is mis­taken and the claim in OP was cor­rect, though that wasn’t know­able from the pub­li­cly re­leased data. See habryka’s com­ment.]

  Many of the ex­pert pre­dic­tions were in­deed crazily op­ti­mistic and had tiny er­ror bars, but there’s a prob­lem with the story. FiveThir­tyEight mis­tak­enly re­ported (and they still haven’t up­dated this!) that the March 16-17 sur­vey asked ex­perts about the num­ber of cases re­ported on Covid Tracker on March 29, when in fact the study asked about March 23rd.

  The cor­rect num­ber on the 23rd was 42,152. This was of course in line with the ex­po­nen­tial ex­trap­o­la­tion, and it was worse than the worst-case es­ti­mates of 13 out of 18 re­searchers, but at least their es­ti­mates show only typ­i­cal lev­els of in­san­ity and in­com­pe­tence.

• orthonormal 31 Mar 2020 20:06 UTC

  LW: 2 AF: 1

  AF

  on: Think­ing About Filtered Ev­i­dence Is (Very!) Hard

  If the listener is run­ning a com­putable log­i­cal un­cer­tainty al­gorithm, then for a difficult propo­si­tion it hasn’t made much sense of, the listener might say “70% likely it’s a the­o­rem and X will say it, 20% likely it’s not a the­o­rem and X won’t say it, 5% PA is in­con­sis­tent and X will say both, 5% X isn’t nam­ing all and only the­o­rems of PA”.

  Con­di­tioned on PA be­ing con­sis­tent and on X nam­ing all and only the­o­rems of PA, and on the listener’s log­i­cal un­cer­tainty be­ing well-cal­ibrated, you’d ex­pect that in 78% of such cases X even­tu­ally names it.

  But you can’t use the listener’s cur­rent prob­a­bil­ities on [X say­ing it] to sort out the­o­rems from non-the­o­rems in a way that breaks com­putabil­ity!

  What am I miss­ing?

• orthonormal 30 Mar 2020 18:56 UTC

  5 points

  on: Solip­sism is Underrated

  I don’t know whether you missed this or just didn’t spell it out, but the rea­son that the like­li­hood ra­tio is so much in fa­vor of ma­te­ri­al­ism over solip­sism is that if solip­sism were true, you could ex­pe­rience liter­ally any­thing, and ap­par­ently or­dered uni­verses con­sis­tent with sim­ple phys­i­cal laws are a van­ish­ing sub­set of the pos­si­ble ex­pe­riences.

  You’d need a solip­sist the­ory that strongly pre­dicts or­dered uni­verses with­out adding in too much com­plex­ity, just in or­der to be in the con­ver­sa­tion with the like­li­hood of qualia be­ing re­ducible to ma­te­ri­al­ism.

  “Qualia be­ing ir­re­ducible” is, to be as char­i­ta­ble to you as pos­si­ble, in the refer­ence class of philo­soph­i­cal po­si­tions that some peo­ple have seen as unas­sailable and oth­ers have seen as flawed. You don’t get to as­sign in­cred­ibly high prob­a­bil­ity within this refer­ence class.

  (To be un­char­i­ta­ble, it is an in­tu­ition for which you can­not provide even what looks like an air­tight philo­soph­i­cal ar­gu­ment, just louder re­it­er­a­tions of your in­tu­ition.)

The Real Standard

• orthonormal 29 Mar 2020 6:13 UTC

  23 points

  in reply to: Pattern's comment on: “No ev­i­dence” as a Valley of Bad Rationality

  In gen­eral, it’s good to check your in­tu­itions against ev­i­dence where pos­si­ble (so, seek out ex­per­i­ments and treat ex­per­i­men­tally val­i­dated hy­pothe­ses as much stronger than in­tu­itions).

  The valley be­ing de­scribed here is the idea that you should just dis­card your in­tu­itions in fa­vor of the null hy­poth­e­sis, not just when ex­per­i­ments have failed to re­ject the null hy­poth­e­sis (though even here, they could just be un­der­pow­ered!), but when ex­per­i­ments haven’t been done at all!

  It’s a gen­er­al­ized form of an iso­lated de­mand for rigor, where what­ever gets defined as a null hy­poth­e­sis gets a free pass, but any­thing else has to prove it­self to a high stan­dard. And that leads to re­ally poor perfor­mance in do­mains where ev­i­dence is hard to come by (quickly enough), rel­a­tive to trust­ing in­tu­itive pri­ors and weak ev­i­dence when that’s all that’s available.

• orthonormal 28 Mar 2020 17:44 UTC

  4 points

  in reply to: ChristianKl's comment on: mind viruses about body viruses

  But we already filter more than the refer­ence class of smart In­ter­net peo­ple, that’s the point. cousin_it ar­gues, and I agree, that this com­mu­nity may already be on the ex­treme of “filters too care­fully in the face of the need for ur­gent up­dates”. We did well by tak­ing COVID-19 se­ri­ously be­fore it was proven, and we could have done still bet­ter on that front.

• orthonormal 27 Mar 2020 17:30 UTC

  2 points

  in reply to: Atmosper's comment on: Just Lose Hope Already

  Some­times. More of­ten the hero just tries AGAIN, BUT HARDER.

• orthonormal 27 Mar 2020 2:05 UTC

  2 points

  in reply to: Isnasene's comment on: Ad­ding Up To Normality

  Huz­zah, con­ver­gence! I ap­pre­ci­ate the points you’ve made.

• orthonormal 26 Mar 2020 21:29 UTC

  5 points

  in reply to: Isnasene's comment on: Ad­ding Up To Normality

  Don’t know if you saw, but I up­dated the post yes­ter­day be­cause of your (and khafra’s) points.

  Also, your caveat is a good re­frame of the main mechanism be­hind the post.

  I do still dis­agree with you some­what, be­cause I think that peo­ple go­ing through a crisis of faith are prone to flailing around and tak­ing naive ac­tions that they would have re­con­sid­ered af­ter a week or month of ac­tu­ally think­ing through the im­pli­ca­tions of their new be­lief. Try­ing to max­i­mize util­ity while mak­ing a ma­jor up­date is safe for ideal Bayesian rea­son­ers, but it fails badly for ac­tual hu­mans.

  In the ab­sence of an ex­ter­nal crisis, tak­ing rel­a­tively safe ac­tions (and few ir­re­versible ac­tions) is cor­rect in the short term, and the sta­tus quo is go­ing to be rea­son­ably safe for most peo­ple if you’ve been liv­ing it for years. If you can back off from newly-sus­pected-wrong ac­tivi­ties for the time be­ing with­out do­ing so ir­re­versibly, then yes that’s bet­ter.

• orthonormal 26 Mar 2020 2:40 UTC

  2 points

  in reply to: shminux's comment on: Ad­ding Up To Normality

  Well, if you have a space pro­gram and you’re deal­ing with crys­tal spheres...

• orthonormal 25 Mar 2020 17:51 UTC

  2 points

  in reply to: lincolnquirk's comment on: Ad­ding Up To Normality

  I think khafra and Is­nasene make good points about not ap­ply­ing this in cases where the plane shows signs of ac­tu­ally drop­ping and you’re up­dat­ing on that. (In this case, the signs would be watch­ing peo­ple you re­spect tell you to start prep­ping im­me­di­ately- act on the warn­ing lights in the cock­pit rather than wait­ing for the en­g­ines to fail.)

• orthonormal 25 Mar 2020 17:46 UTC

  2 points

  in reply to: Isnasene's comment on: Ad­ding Up To Normality

  I agree that care­fully land­ing the plane is bet­ter than main­tain­ing the course if catas­trophic out­comes sud­denly seem more plau­si­ble than be­fore.

  Ob­vi­ously it ap­plies if you’re the lead on a new tech­nolog­i­cal pro­ject and sud­denly re­al­ize a plau­si­ble catas­trophic risk from it.

  I don’t think it ap­plies very strongly in your ex­am­ple about an­i­mal welfare, un­less the pro­tag­o­nist has un­usu­ally high lev­er­age on a big de­ci­sion about to be made. The cost of con­tin­u­ing to stay in the old job for a few weeks while think­ing things over (es­pe­cially if leav­ing and then com­ing back would be in­fea­si­ble) is plau­si­bly worth the value of in­for­ma­tion thus gained.

• orthonormal 25 Mar 2020 17:40 UTC

  4 points

  in reply to: khafra's comment on: Ad­ding Up To Normality

  I’d mod­ify that, since panic can make you falsely put your­self in weird refer­ence classes in the short run. It’s more re­li­able IMO to ask whether any­thing has shifted mas­sively in the ex­ter­nal world at the same time as it’s shifted in your model.

  How about promise your­self to keep steer­ing the plane mostly as nor­mal while you think about lift, as long as the plane seems to be fly­ing nor­mally?

Ad­ding Up To Normality

• orthonormal 24 Mar 2020 19:53 UTC

  6 points

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

  I wish I’d re­mem­bered to in­clude this in the origi­nal post (and it feels wrong to slip it in now), but Scott Aaron­son neatly par­alleled my dis­tinc­tion be­tween ra­tio­nal­ists and post-ra­tio­nal­ists when dis­cussing in­ter­pre­ta­tions of quan­tum me­chan­ics:

  But the ba­sic split be­tween Many-Wor­lds and Copen­hagen (or bet­ter: be­tween Many-Wor­lds and “shut-up-and-calcu­late” /​ “QM needs no in­ter­pre­ta­tion” /​ etc.), I re­gard as com­ing from two fun­da­men­tally differ­ent con­cep­tions of what a sci­en­tific the­ory is sup­posed to do for you. Is it sup­posed to posit an ob­jec­tive state for the uni­verse, or be only a tool that you use to or­ga­nize your ex­pe­riences?

  Scott tries his best to give a not-an­swer and be done with it, which is in keep­ing with my cat­e­go­riza­tion of him as a promi­nent ra­tio­nal­ist-ad­ja­cent.