I have updated the plot on the COVID19 page as ususal, and here is a new version of the one that tries to align the cuves with each other by shifting them in time:

This is scary in the sense that many places will probably look like Italy within days. Exponential grows is unintuitive and easily surprises systems and people that function linearly. This video does a great job illustrating the numbers behind an epidemic - highly recommended watch!

There is hope, of course, and the dire situation in Italy need not be a prediction for the rest of the world. This is because we already have an example of a large outbreak that was brought under control: What South Korea does works! They report a declining number of new cases and have tested 3.6‰ of their population, more than any other country.

Social distancing is a major tool in this effort, more on this in this Twitter thread.

Mellofinalen gick av stapeln igår i Stockholm trots allt. Jag hoppas att jag har fel men jag skulle bli förvånad om det inte bli ett synligt uppsving av COVID19-fall om några dagar. Om inte antalet redan har ökat så mycket då att det inte längre syns i statistiken alltså.

Jag följer noga ett internationellt gäng smarta och kunniga människor på Twitter; epidemologer, virologer och besläktade yrken. Och jag slås med häpnad av diskrepansen mellan deras budskap, baserad på färskaste siffror och kunskap om viruset, och det som man får höra från Folkhälsomyndigheten, med statsepidemolog Anders Tegnell i spetsen.

Så sent som i förrgår påstod han att det inte finns någon lokal spridning i Sverige utan bara importerade fall som man har bra koll på genom smittspårning. Detta medan antalet bekräftade fall är samma som i Italien för bara två veckor sedan.

Tegnell medgav visserligen att han har missbedömt situationen i Kina men idag fortsätter han att argumentera mot drastiska åtgärder. Det blir inte bara aningens obegripligt när Taiwan, Singapore och framför allt Sydkorea visar att det faktiskt går att begränsa spridningen.

I en färsk analys har Sverige högsta R-värde (d.v.s. hur många fall som orsakas av ett befintligt fall) av alla länder som ingår i studien. Det ligger kring R=4 i Sverige. Sydkorea och Hubei-provinsen i Kina har däremot lyckas pressa ner det till R=1 eller t.o.m. under. Spridningen ökar alltså inte längre där.

Exponentialkurvor är luriga och inkubationstiden gör att man hela tiden ligger efter. Åtgärder som känns rimliga i dagens läge kommer att vara helt otillräckliga imorgon, och det som skulle ha gjorts idag känns helt överdrivet än så länge. Men det behöver göras ändå. Det kräver lite mod av beslutsfattarna och utsätter dem för kritiken att överreagera. Låt oss hoppas att en överreaktion är det vi beklagar i snar framtid, inte en situation som i Italien, orsakad av farlig tvekan.

Jag håller därför helt med DN's ledare att det är dags att inse allvaret och inte vara rädd för att "orsaka panik". För det första är panik inte lika farligt som man tror, för det andra eroderar velandet förtroendet för myndigheterna. Det förtroendet kommer att behövas snart, när sjukvården överbelastas rejält.

Slutligen, att "sparka Tegnell" är retoriskt mednat och är inte nödvändigtvis rätt åtgärd. Jag har ingen insyn i hur mycket han själv ansvarar för budskapet och hur mycket som kommer "ovanifrån". Men ett ändrat budskap och drastiska åtgärder behövs.

Ändring 12/3: Jag har lagt till ett "inte" i rubriken.

Since I just tweeted it, I might as well put up this plot here as well:

It shows the latest reported number of COVID19 cases outside China, on a logarithmic scale. If you are not used to this, just know that the curve steepness corresponds to the growth rate of cases, or if you will the doubling time. The steeper the curve, the shorter the time it takes get double the number of cases.

There are several noteworthy things here, I believe. First off, like the last post claimed, S.Korea has managed to flatten the curve more than any European country. Only Singapore, Hong Kong and Japan seem to do even better, but they have had a far smaller outbreak for now.

I sure hope for them that Iran is not doctoring its numbers and that the slowing down in their curve is real.

The steep slope from Belgium, where I am at the moment, it do not like at all. It is still small number statistics and depends on the rate of testing, i.e. what fraction of actual cases is discovered.

Homecountry Sweden is not doing so well either, especially considering that the numbers in the plot are not scaled by population size. At least they yesterday reported the number of 3000 tests carried out, which does not seem ridiculously low for 100 cases. I have asked the authorities for better stats on this, but they will only become available next week.

Scott Sumner asks whether Europe and America are bungling the Coronavirus outbreak:

Unlike the Chinese, we had several months to plan and then launch an all out effort to address the problem with very aggressive testing of the population to spot outbreaks. And yet both the US and Italian governments seemed unprepared for the outbreak when it finally occurred. [...] If Europe and the US fall into recession we should not blame China. Indeed, many analysts expect the Chinese economy to be almost back to normal by mid-April. Instead, we need to look in the mirror.

Looking at the numbers seems to confirm this. China has reduced its R0 drastically and S.Korea has made an immense effort in testing 140.000 people, and their increase in deaths and news cases has slowed more than Italy's.

Meanwhile the US mess up their testing and the Rest of Europe seems to be slow as well. Authorities still have the main message that all is under control and hesitate to employ drastic measures. At least in Germany and Sweden where I follow the news.

I just started a continuously updated page at /covid19

This graph plots the currently reported case-numbers of a few countries on a logarithmic scale. I have applied a shift of a few days, see the legend, to make the curves match up roughly.

I find remarkable that they can easily be made to line up, as soon as the numbers increase above the noisy few.
Now one could conclude from this that Italy is 4 days behind S.Korea in its outbreak, but the numbers depend on how much testing each country does and S.Korea does a lot more than most. Italy already has more deaths and its outbreak more likely has progressed further than this graph would suggest,

What S.Korea is doing seems to be working. This is good news because it shows that not only dictatorial countries can succeed with their containment measures. Let's hope Europe learns quickly from this.

This is not the flu.

The age dristribution seems to matter a lot. This is just a projection that among other things ignores differences in treatment capabilities.

MIT takes the lead among the bis universities and suspends all foreign travel and meetings until mid-May.

A short interview with Yuval Noah Harari who, as usual, has interesting things to say.

Tyler Cowen tries to explain why people are recting so differently to the developments. I think he is right; this is certainly part of what happens at the meta-level.

A random set of internet gems that I enjoyed recently, starting with the Coronavirus COVID19. Three Twitterers that seem to knowledgable, up-to-date and not hyping-up matters are Kai Kupferschmidt, Helen Branswell and Ian M Mackay. Plus, there are good articles coming from down under.

Don't bury the lede, not even in grant applications.

The International Space Station in LEGO. I like that this is an official set, not a custom design or one-off sale. Speaking of which: My Science Tower is still waiting to be built.

Quit the news! (PDF) It's bad for you!

Robin Hanson comapares Parasite and Joker. This made me think less of Parasite and want to watch Joker. I have downloaded it, but not gotten around to it yet.