# TheMajor

Karma: 589
• I’m happy to hear that some of these changes have been un­ex­pect­edly pos­i­tive for you! Per­son­ally I already did a bunch of these things (shop for one week’s worth at a time, have days work­ing from home, or­der on­line). To offer a bit of a peek at the flip side: I work in math­e­mat­ics, which is uniquely suited for work­ing at home (there’s a mod­ern joke that to do math­e­mat­ics all you need is pen, pa­per, a bot­tle of wa­ter and a su­per­com­puter), yet 2 of the 11 col­leagues in my group have suffered burnouts since March from the added stress of hav­ing to look af­ter their house­holds. We are con­sid­er­ing go­ing back to 20% office ca­pac­ity soon in stag­gered shifts, which while nice still means we won’t have a chance to talk or work to­gether in prac­tice. Ob­vi­ously this is ex­actly the point, but I want to note that this is a far cry from nor­mal. My pro­duc­tivity at the mo­ment is at an all time low, and sev­eral of my other friends have already heard that if this situ­a­tion con­tinues for much longer they will be let go from their jobs. In this sense I think this is un­sus­tain­able, or at the very least a se­ri­ous hit to our global growth and pro­duc­tivity.

I have ab­solutely no quan­ti­ta­tive guess what the im­pact of su­per­spread­ers is, and it would be amaz­ing if we could stop them quickly. I think Zvi also pointed out that su­per­spread­ers get elimi­nated quickly in a pan­demic, one way or an­other (and ‘hav­ing the dis­ease, sur­viv­ing and then be­com­ing im­mune’ counts as the other).

I thought that the cur­rent spread of Covid-19 in warmer coun­tries (Brazil, In­dia) was ev­i­dence against the virus be­ing very sus­cep­ti­ble to tem­per­a­tures, but there are a lot of con­founders. If you know of any good sum­mary of the cur­rent knowl­edge on this please let me know, I am very in­ter­ested in this (and it would likely change my pre­dic­tions mas­sively).

• Thank you for your com­ment. I don’t think we dis­agree—Slo­vakia (and other Euro­pean coun­tries) have done ex­tremely well by act­ing quickly. I also fully sup­port wash­ing hands, wear­ing face masks and be­ing smart about so­cial con­tact. I think you are re­spond­ing to my sentence

[...] this sug­gests to me that ‘The Dance’ will look a lot more like ‘a full quaran­tine, but with a few re­stric­tions lifted’ than like ‘restor­ing so­cial con­tact, but wash your hands and wear a face mask’.

I’m sorry for the con­fu­sion. The situ­a­tion you de­scribe sounds more like my sce­nario 3 than my sce­nario 2, and I think I ex­plained poorly where I draw the line in my quoted state­ment above. Go­ing by wikipe­dia, the most re­cent round of re­lax­ations in Slo­vakia still sounds far from life as nor­mal to me. The max­i­mum oc­cu­pancy set at shops is up to 1 per­son per 15 square me­ters, which is 25% of what it is in nor­mal times. I imag­ine similar con­cerns ap­ply to offices, but I can’t find how many peo­ple are back to work­ing at the office. The open­ing of schools hap­pened only this Mon­day, so it re­ally is too early to tell what the im­pact of that will be. Zvi voices some con­cerns.

Lastly, my en­tire goal was to try and talk about rel­a­tive im­pact, and give per­spec­tive to the mag­ni­tude of the mea­sures we can ex­pect go­ing for­wards. I don’t see where I have mis­tak­enly given ab­solute state­ments on effec­tive­ness of mea­sures (in fact I only men­tioned face masks once, with­out mak­ing any state­ment for or against them in the OP), but if you point them out I would be happy to change them.

• I agree com­pletely. How­ever, I think the amount it has gone up is crit­i­cal here. A lot of the coun­ter­mea­sures and in­creased prepa­ra­tion are lin­ear coun­ter­mea­sures against an ex­po­nen­tial threat—maybe a re­gion that could pre­vi­ously only han­dle 1000 ICU pa­tients can now take care of 2000, but if R0 is sig­nifi­cantly above 1 (lets say 1.5) this only buys you about one and a half week. I think this topic de­serves its own en­tire post at some point, and I didn’t want to get bogged down in de­tails in the sec­tion on “What doesn’t change”, but if the true rule is “if un­der X cir­cum­stances in March it was smart to go into lock­down, it is Novem­ber smart to go into lock­down 2 weeks af­ter see­ing X” my con­clu­sions are still the same.

I might write that full post some­time on this and more back-and-forth, if peo­ple are in­ter­ested. I made se­ri­ous con­ces­sions to brevity above.

• Thank you for your com­ment! I’d like to briefly clar­ify that I don’t con­sider the “lock­down ver­sus kill the peo­ple” ques­tion a dilemma—I’m not try­ing to pro­pose policy, just at­tempt­ing to make re­al­is­tic pre­dic­tions for the next few months.

I think go­ing for the ap­proach you sug­gest (‘op­tion 2’) is un­re­al­is­tic in most west­ern coun­tries, and that the coun­tries where this strat­egy worked had unique cir­cum­stances that worked in their favour.

As gbear605 points out, it is a lot more difficult to re­vert to this strat­egy when the dis­ease is en­demic, com­pared to when the num­ber of ac­tive cases is in the sev­eral hun­dreds. In the post I link to an es­ti­mate of 1.1 mil­lion ac­tive cases of Covid-19 in the US to­day. One of the big­ger weak links in the non-quaran­tine ap­proach is that, when peo­ple test pos­i­tive for cold/​flu symp­toms, there needs to be some plan or sup­port in place to thor­oughly iso­late them for up to two weeks. At lower num­bers this can plau­si­bly han­dled by hos­pi­tals or nurs­ing homes (which has been a big part of South Korea’s method), but at up to a mil­lion cases this is un­re­al­is­tic. In effect this would be akin to a lock­down with some ex­tra steps.

Another ad­van­tage that South Korea, Taiwan and Sin­ga­pore (as far as I know the biggest suc­cess sto­ries of this method) had is that they are ge­o­graph­i­cally iso­lated. I imag­ine this cuts down on the threat of re-in­fec­tion from out­side the coun­try, and also why it is so much more difficult to pre­vent Covid-19 from be­com­ing en­demic in the US.

Wide­spread screen­ing for ele­vated tem­per­a­tures and cold/​flu symp­toms sounds like a very clever (and cost-effec­tive!) method to com­bat the virus, if we also have a plan ready for peo­ple who test pos­i­tive. But I don’t think it can be the suc­cess story in west­ern coun­tries that it is in some Asian coun­tries. We ei­ther missed our win­dow and let Covid-19 grow too much, or we never had a win­dow to be­gin with.

As for your later two points, I be­lieve I ad­dressed those in the origi­nal post. I think bring­ing R0 down to around 1 and keep­ing it there is pro­hibitively tax­ing, in mul­ti­ple ways, and that even slight re­lax­ation of our re­sponse will already have se­vere con­se­quences. We should be think­ing on the R0 scale, which runs from about 0.7 to about 2.5-3. Also I ex­plic­itly men­tion that I am fine with leav­ing the level at which we achieve herd im­mu­nity open for de­bate (I give 20% as a some­what lower bound, based on es­ti­mates that this ex­po­sure level has already been reached in NYC).

# My pre­dic­tion for Covid-19

31 May 2020 23:25 UTC
17 points
• Ex­cel­lent post, thank you for writ­ing this! I’ve been mean­ing to write up some­thing similar, about my opinion on the pub­lic re­sponse to COVID-19 and a few fu­ture pre­dic­tions. I think this com­ment sec­tion is a perfect place for this, and I’ll write my own thoughts as a re­sponse to yours. My ex­pe­riences are not based on the situ­a­tion in the USA.

As a dis­claimer, my pre­dic­tions are among the most pes­simistic of all the ones I’m see­ing peo­ple dis­cuss. Con­sider it the bot­tom pit of the Over­ton win­dow. Hav­ing ex­pec­ta­tions that COVID-19 will be worse than peo­ple are think­ing so far can and does have nega­tive effects on your men­tal health, so be care­ful (and stop read­ing). I’d love to be proven wrong.

Risks Fol­low Power Laws

Amaz­ingly writ­ten, I did not give this enough thought. I should stop do­ing my own gro­cery shop­ping, and have it de­liv­ered in­stead.

Sacri­fices To The Gods Are De­manded Everywhere

I agree fully. Your steel­man hits quite close to the mark in my opinion. I want to add that if ev­ery­body has in­ter­nal­ised that ‘pleas­ing the Gods’ is the way to re­spond (to any­thing, re­ally), then this ac­tu­ally be­comes the cor­rect way for au­thor­i­ties to write guidelines – any­thing else will just add an­other layer of trans­la­tion.

Govern­ments Most Places Are Ly­ing Liars With No Abil­ity To Plan or Phys­i­cally Rea­son. They Can’t Even Stop In­terfer­ing and Killing People

Yes.

Silence is Golden

I am com­pletely un­sure on the rel­a­tive im­por­tance of this, com­pared to other risk fac­tors. How much of the spread are you will­ing to at­tribute to this? Have there been stud­ies on this?

Sur­faces Are Mostly Harmless

I have been keep­ing all pack­ages that ar­rived in a sealed plas­tic bag for 24 hours be­fore open­ing (like a Good Boy), I guess that was stupid. Read­ing this para­graph changed my mind.

Food Is Mostly Harmless

Yeah, this all makes sense.

Out­door Ac­tivity Is Rel­a­tively Harmless

As does this.

Masks Are Effec­tive, And Even Cloth Ones Are Al­most Enough

I to­tally dis­agree with this. Where are the num­bers com­ing from? I con­sider cloth masks closer to a Sacri­fice to the Gods than a real coun­ter­mea­sure.

Six Feet Is An Ar­bi­trary Num­ber, Peo­ple Aren’t Treat­ing It That Way, And That’s Terrible

You have this one com­pletely back­wards. Peo­ple re­ally re­ally are not able to make risk as­sess­ments and benefit analy­ses. This world you de­scribe, where the rule is taken as a best point es­ti­mate, and peo­ple in­ter­pret the ‘6 feet’ as any­thing else than a Boolean yes/​no ques­tion, was never go­ing to hap­pen. I con­sider the cur­rent pub­lic aware­ness of the im­por­tance of keep­ing dis­tance an amaz­ing im­prove­ment on the sta­tus quo be­fore this rule went pub­lic. You men­tion this (as the open­ing sen­tence, in fact). I dis­agree with any sug­ges­tion that we could plau­si­bly do bet­ter col­lec­tively.

Herd Im­mu­nity Comes Well Be­fore 75% In­fected and Par­tial Herd Im­mu­nity Is Su­per Valuable

This is a very good point, but can you ex­plain the num­bers you give be­low? I have no in­tu­ition on what typ­i­cal spread in per­sonal in­fec­tion risk is (sure, the range is large, but we also need the fre­quen­cies with which these high risks are taken).

Yes, We Know Peo­ple Who Have Been In­fected Are Immune

I agree fully.

Our Lack of Ex­per­i­men­ta­tion Is Still Com­pletely Insane

Agreed.

We Should Be Spend­ing Vastly More on Vac­cines, Test­ing and Other Med­i­cal Solutions

I agree here, ex­cept with the last line. With all the re­gres­sion to the mean the cur­rent med­i­cal sys­tem is ex­pe­rienc­ing, it is to­tally un­clear to me how we would plau­si­bly cre­ate a vac­cine on a short (“a few months”) timescales.

R0 Un­der Amer­i­can-Style Lock­down Con­di­tions De­faults To Just Un­der One, Which New York Es­caped Via Par­tial Herd Immunity

Another great ob­ser­va­tion, and I’ll re­ply in more de­tail in a sep­a­rate post..

The De­fault In­fec­tion Fatal­ity Rate (IFR) Is 0.5%-1%, Depend­ing on Conditions

There has been dis­cus­sion of this on LessWrong be­fore, and I agree.

Typ­i­cally in Amer­ica, 33% of Deaths and 90% of In­fec­tions Are Missed

This sounds like the cor­rect or­der of mag­ni­tude for both, and also agrees with my es­ti­mates for the un­der-re­port­ing in the offi­cial num­bers where I live.

Peo­ple Don’t Mod­ify Be­hav­ior In Re­sponse To Re-Open­ings All That Much, When Given a Choice

This ti­tle is a very im­por­tant (pos­i­tive) re­mark, be­cause it sug­gests (con­trary to what I per­son­ally be­lieve) that it might be pos­si­ble to re­sume a lot of ac­tivi­ties al­most as nor­mal while main­tain­ing R0<1. I agree that the situ­a­tion for schools is uniquely bad, and un­for­tu­nately I can tell you from per­sonal ex­pe­rience this is one of the places com­pletely in­fected with mak­ing Sacri­fices to the Gods. This is a dis­aster wait­ing to hap­pen.

It’s Out of Our Hands

When was this win­dow for peo­ple in au­thor­ity? Look­ing back I hon­estly think we were never ca­pa­ble of dodg­ing this dis­aster (our so­ciety isn’t ad­e­quate at the level of sav­ing peo­ple dur­ing a pan­demic).

• Both out­side view rea­son­ing and cor­rigi­bil­ity use the out­come of our own util­ity calcu­la­tion/​men­tal effort as in­put for mak­ing a de­ci­sion, in­stead of out­put. Per­haps this should be in­ter­preted as tak­ing some gods-eye-view of the agent and their sur­round­ings. When I in­voke the out­side view, I re­ally am ask­ing “in the past, in situ­a­tions where my brain said X would hap­pen, what re­ally hap­pened?”. Look­ing at it like this I think not in­vok­ing the out­side view is a weird form of du­al­ity, where we (will­ingly) ig­nore the fact that his­tor­i­cally my brain has dis­pro­por­tionately sug­gested X in situ­a­tions where Y ac­tu­ally hap­pened. Of course in a world with ideal rea­son­ers (or at least, where I am an ideal rea­soner) the out­side view will agree with the out­put of my men­tal progress.

To me this feels differ­ent (though still similar or pos­si­bly re­lated, but not the same) to the cor­rigi­bil­ity ex­am­ples. Here the differ­ence be­tween cor­rigible or in­cor­rigible is not a mat­ter of ex­pected fu­ture out­comes, but is de­cided by un­cer­tainty about the de­sir­a­bil­ity of the out­comes (in par­tic­u­lar, the AI hav­ing false con­fi­dence that some bad fu­ture is ac­tu­ally good). We want our un­trained AI to think “My real goal, no mat­ter what I’m cur­rently ex­plic­itly pro­grammed to do, is to satisfy what the re­searchers around me want, which in­cludes com­ply­ing if they want to change my code.” To me this sounds differ­ent than the out­side view, where I ‘merely’ had to ac­cept that for an ideal rea­soner the out­side view will pro­duce the same con­clu­sion as my in­side view, so any differ­ences be­tween them are in­ter­est­ing facts about my own men­tal mod­els and can be used to im­prove my abil­ity to rea­son.

That be­ing said, I am not sure the differ­ence be­tween un­cer­tainty around fu­ture events and un­cer­tainty about de­sir­a­bil­ity of fu­ture states is some­thing fun­da­men­tal. Maybe the con­cept of probu­til­ity bridges this gap—I am posit­ing that cor­rigi­bil­ity and out­side view rea­son on differ­ent lev­els, but as long as agents ap­ply­ing the out­side view in a suffi­ciently thor­ough way are cor­rigible (or the other way around) the differ­ence may not be phys­i­cal.

• I’ve tried to read through the linked page, and swapped to aca­demic read­ing’ (check­ing the pic­tures, and some­times the first and last line of para­graphs) halfway through. I think this is not vi­able.

There is a host of “the­o­ries of the uni­verse” with a similar struc­ture on a meta-level, con­sist­ing of some kind of emer­gent com­plex­ity. It is im­por­tant to keep in mind the strength of a the­ory lies in what it for­bids, not in what it per­mits. To date most the­o­ries of the uni­verse fail this test hard, by be­ing so vague and non­spe­cific that any sci­en­tific con­cept can be pat­tern-matched to some as­pect of it. Judg­ing by what I’ve read so far this is no ex­cep­tion (and in fact, I sus­pect that the rea­son Wolfram refer­ences so many big sci­en­tific the­o­ries is be­cause large con­cepts are eas­ier to pat­tern-match, whereas spe­cific pre­dic­tions are not as open to in­ter­pre­ta­tion). Why will his pat­terns pro­duce Ein­stein’s equa­tions (note that they cur­rently do no such thing, he states we first need to “find the right uni­verse”), and not New­ton’s, or Ein­stein’s with dou­ble the speed of light?

As always with these non­spe­cific the­o­ries’ it is very difficult to nail down one spe­cific weak­ness. But cur­rently all I’m see­ing are red flags. I pre­dict se­ri­ous me­dia at­ten­tion and pos­si­bly some rele­vant dis­cov­er­ies in physics (some of the para­graphs sounded bet­ter than all other crack­pot the­o­ries I’ve seen), but the ma­jor­ity of it seems wrong/​worth­less.

• I don’t have a good model to give me any pre­dic­tions on what rea­son­able num­bers of asymp­tomatic cases would be, or how trun­ca­tion in­fluences these num­bers. Could you ex­plain why the in­fer­ence is idiotic, and per­haps give a more rea­son­able one?

• Is there rea­son to be­lieve the raw num­bers are more ac­cu­rate es­ti­mate of the rate than the model pre­dic­tion? Also, what are the type-1 and type-2 er­rors of the tests used on the Di­a­mond Princess? I heard some early re­ports that both of these might be sig­nifi­cant, but then never heard any­thing about them again.

I checked that link above and fol­lowed their refer­ences to find other datasets, but two of them are in Ja­panese, one only deals with self-se­lected pa­tients who showed symp­toms, and the last two have small sam­ple size (12 pa­tients, two pa­pers cover the same event).

Up­date: I have found https://​​www.eu­ro­surveillance.org/​​con­tent/​​10.2807/​​1560-7917.ES.2020.25.3.2000045, which bench­marks the real-time re­verse tran­scrip­tion polymerase chain re­ac­tion (RT-PCT) tests. They state zero false pos­i­tives in a trial with 297 non-COVID-19 sam­ples, al­though they do retest 4 sam­ples that showed “weak ini­tial re­ac­tivity”. Since the non real-time ver­sion of RT-PCT is sup­posed to be even more re­li­able, this means false pos­i­tives are pre­sum­ably not a big deal (even at a pes­simistic 4/​​297 false pos­i­tive this still means only 41 false pos­i­tives out of 3063 tests done on the Di­a­mond Princess).

• I think it is still very un­clear what the situ­a­tion a medium (~a few months) to long (~a year) amount of time from now will be like. I would love to see more dis­cus­sion on this. On a re­lated note, I think men­tal health and self-help are go­ing to very im­por­tant very soon, and while I am in a fine place per­son­ally I would still like to know a lot more about this, in­clud­ing how to help oth­ers. This strays a lit­tle bit from the other COVID dis­cus­sion top­ics, but I do think LessWrong might have a com­par­a­tive ad­van­tage here (es­pe­cially com­pared to the crap­shoot baseline that is the in­ter­net).

• Have you got a source for that ‘about half the cases are asymp­tomatic’? I was un­der the im­pres­sion that far more cases show symp­toms even­tu­ally, and that the stud­ies show­ing half of the in­fec­tions are asymp­tomatic add the dis­claimer ‘so far’, which means very lit­tle if the spread is grow­ing ex­po­nen­tially with a dou­bling time of sev­eral days.

• Thanks for shar­ing! I’m not a doc­tor, so I found this a tough read. This doc­u­ment is clearly a pro­posal (at­tempt­ing to con­vince the reader) in­stead of a sum­mary, but it still con­tains a lot of use­ful in­for­ma­tion. Nev­er­the­less, there were some parts I found es­pe­cially con­fus­ing.

On page 2 they men­tion there are cur­rently 22 stud­ies, of which one has com­pleted, on the effect of hy­drox­y­chloro­quine (HCQ) treat­ment on COVID-19 pa­tients. Fur­ther down in the piece (in par­tic­u­lar in the sec­tion “What about the stud­ies that show no benefit from HCQ?” on page 11) they dis­miss some stud­ies show­ing lit­tle or no effect. Is there a place to find more dis­cus­sion on which stud­ies are be­ing dis­counted, and for what rea­son? They link one study only, cit­ing that “only 400mg daily for 5 days was used”, al­though the sug­gested treat­ment in this doc­u­ment is “HCQ: 6.5-15mg/​kg PO in di­vided load­ing dose fol­lowed by 400-800mg/​day in di­vided doses for 4-9 days” (which en­com­passes 400mg daily for 5 days).

The recom­mended treat­ment is a com­bi­na­tion treat­ment with four differ­ent com­po­nents—an ini­tial oral hy­drox­y­chloro­quine ad­minis­tra­tion and a daily treat­ment of hy­drox­y­chloro­quine and two other medicines (zinc and Az­ithromycin). Fur­ther­more the doc­u­ment states that this treat­ment is ex­pected to work a lot bet­ter in early stages of the dis­ease (this part is also un­clear to me—again on page 11 they state that “[some stud­ies] waited to ini­ti­ate treat­ment un­til the dis­ease was too far pro­gressed to be effec­tive” as grounds for dis­mis­sal). Does this mean this treat­ment is ex­pected to have next to no effect in late stages? I’m wor­ried about Bon­ferfoni-es­que situ­a­tions here; are 21 in­com­plete and 1 com­plete study strong enough to mo­ti­vate this com­pli­cated treat­ment, es­pe­cially if we al­low our­selves to dis­count some pa­pers with con­flict­ing con­clu­sions as well as re­strict the time pe­riod over which the treat­ment is sup­posed to be effec­tive?

• I am very in­ter­ested in dis­cus­sion on hy­drox­y­chloro­quine, but do not have a Face­book. Is there some other way to read the megath­read?

• Very nice! Two mis­takes though:

• Tech­ni­cally the in­tro­duc­tory part on deriva­tives on is in­cor­rect, in two differ­ent ways.

• Firstly the deriva­tive of a map is a map , that as­signs to ev­ery point x a lin­ear map send­ing di­rec­tion y to a real value (namely the par­tial deriva­tive of f at x in di­rec­tion y). Thank­fully the space of lin­ear maps from to is iso­met­ri­cally iso­mor­phic to through the in­ner product, re­cov­er­ing the ex­pres­sion you gave. Similarly the deriva­tive of a map is a map .

• Se­condly tech­ni­cally the do­main of any deriva­tive like the one above is not the vec­tor space we are work­ing with, but the set of di­rec­tions at point x. This no­tion is for­mal­ised in Man­i­fold the­ory and called the tan­gent space. Thank­fully for any finite-di­me­nional vec­tor space the tan­gent space at any point is canon­i­cally iso­mor­phic to the vec­tor space it­self (any vec­tor is a di­rec­tion, that’s what they were in­vented for). In in­finite di­men­sions this still holds just fine ex­cept for the small de­tail that the no­tions of man­i­fold and tan­gent space don’t ex­ist there. The same dis­tinc­tion is nec­es­sary in the range. So truly, for­mally, the deriva­tive of a map is a map , with and similarly , with the con­di­tion that is sim­ply on the first co­or­di­nate. This co­in­cides with the map above: for ev­ery we get a lin­ear map

• The above may seem very con­fus­ing for , since I claim that the deriva­tive in that case is a map in­stead of sim­ply a real-val­ued func­tion. This is re­solved by not­ing that each lin­ear map from to can be rep­re­sented with a num­ber, similar to the top bul­let point above (the in­ner product on is just mul­ti­pli­ca­tion). I think lec­tur­ers are quite jus­tified in not ex­plor­ing the de­tails of this when first in­tro­duc­ing deriva­tives or par­tial deriva­tives, but un­for­tu­nately in pos­si­bly in­finite-di­men­sional ab­stract vec­tor spaces the dis­tinc­tions are nec­es­sary, if only to avoid type er­rors.

• In the defi­ni­tion of the par­tial deriva­tive of M at f with re­spect to g (so with a range in­side a vec­tor space Y) we do not take the norm or ab­solute value of that ex­pres­sion, it should be the straight up limit . The claim that the limit ex­ists does de­pend on the topol­ogy of Y and there­fore on the norm, though.

Also there are a lot of dis­con­tin­u­ous lin­ear maps out there. A text­book ex­am­ple is con­sid­er­ing the vec­tor space of polyno­mi­als in­ter­preted as func­tions on the closed in­ter­val , equipped with supre­mum norm. The deriva­tive map is not con­tin­u­ous, and you can ver­ify this di­rectly by search­ing for a se­quence of func­tions that con­verges to 0 whose image does not con­verge to 0.

• Per­son­ally I did the ex­act op­po­site, and found that very re­fresh­ing. When­ever I ran into a snip­pet of ap­plied func­tional anal­y­sis with­out know­ing the for­mal back­ground it just con­fused me.

• I have a bit of a nega­tive an­swer, in the sense that I don’t know what to recom­mend. A lot of peo­ple in my life (col­leagues, friends, fam­ily) want to stay in touch (and I with them), but each per­son seems to have their own prefer­ences about soft­ware, or at least suffi­ciently many sub­groups of them do that no sin­gle tool will suffice. At this point I hon­estly think that it’s not so much the qual­ity of the tool, but more who are already us­ing it, which de­ter­mines which soft­ware is best. If you and your friends all used to meet on Minecraft ev­ery Tues­day, then that is prob­a­bly the ideal way to keep do­ing things.

As of yet I’m us­ing Skype, What­sApp (+What­sApp Web), per­sonal email, two work emails, two Dis­cord clients (one in browser, one as an app, with sep­a­rate ac­counts), my phone, two Slack workspaces, weekly Zoom group meet­ings, Google Talk, MS Teams and the oc­ca­sional Jitsi call. This is crazy, but all of them are suffi­ciently low traf­fic that I don’t re­ally mind.

All of them work fine, I do have mild per­sonal prefer­ences (for Slack, Dis­cord and my phone over all the oth­ers) but like I said above it’s far more rele­vant to stay in touch at all than to do it with the right tool.

• I am quite un­der­whelmed by this ar­ti­cle. The first half neatly sum­ma­rizes some statis­tics and plans that have been pro­posed, and the po­ten­tial im­pact of these. I like this a lot, and the strong line (fur­ther down in the ar­ti­cle)

If you’re a poli­ti­cian and you see that one op­tion is to let hun­dreds of thou­sands or mil­lions of peo­ple die with a miti­ga­tion strat­egy and the other is to stop the econ­omy for five months be­fore go­ing through the same peak of cases and deaths, these don’t sound like com­pel­ling op­tions.

ac­cu­rately sum­ma­rizes my feel­ings on the mea­sures I’ve seen so far.

How­ever, al­most im­me­di­ately af­ter the first half the ar­ti­cle changes tone dra­mat­i­cally, and I feel un­com­fortable with most of the re­main­der. The en­tire ar­gu­ment hinges on be­ing able to con­trol ‘the Dance’. Per­son­ally I think there is lit­tle to no chance of this work­ing, due to the sheer amount of pub­lic co­op­er­a­tion and gov­ern­ment co­or­di­na­tion re­quired for the mea­sures you sug­gest. I’m will­ing to dis­miss the ex­am­ples of South Korea and Sin­ga­pore as unique cir­cum­stances that do not read­ily ap­ply to a lot of west­ern coun­tries (for ex­am­ple Sin­ga­pore is tightly con­fined ge­o­graph­i­cally and im­ple­mented their mea­sures very early), an ar­gu­ment which is pre­sented and im­me­di­ately dis­missed in the ar­ti­cle with­out elab­o­ra­tion.

I would love to see a more quan­ti­ta­tive anal­y­sis of what this dance would look like in a west­ern coun­try. The ar­ti­cle states that the im­pact of con­tact trac­ing, travel re­stric­tions, big crowds and plau­si­bly other mea­sures are vastly un­der­stated—what is the quan­ti­ta­tive im­pact of these? After the ex­cel­lent in­tro­duc­tion I was ex­pect­ing an equally high level sum­mary of this new strat­egy, but in­stead it was only mo­ti­vated by anal­ogy and rhetoric.

• Yes, the bul­let point list is for Dis­cord. I only tried Jitsi for an hour or two, but the qual­ity of the call was hor­rible and the ses­sion crashed when a sec­ond per­son joined (so re­ally I could just see my own face, pix­el­lated). When I tried to down­load the Ubuntu client I got dpkg er­rors and a failed in­stal­la­tion. Maybe Jitsi is ac­tu­ally great(?), but I’m not very tempted to try again with­out out­side help.