Emerging public discussion in 🇩🇪 that the internal debate of the German health office @rki_de didn't match the political pronouncements, centering on 17 March 2020 and the call to increase Covid hazard level from "moderate" to "high".

A good opportunity to recapitulate... ...the biggest mistakes that drove the response to the pandemic.

1⃣ The perpetual lack of recognition that we were observing events that had occurred weeks before, and the refusal to account for this.

What exactly happened on March 17? Nothing much in terms of the spread. /2

The German health authority @rki_de has evaluated the efficacy of pandemic restrictions it supported, and unsurprisingly it awarded itself a good grade. Overall everything worked out quite well, it says. We looked under the hood and found some surprises. A thread (🇩🇪 below). /1 For starters, in order to show that an intervention had a hypothesized causal effect, a study has to be able to reject the null hypothesis. The null hypothesis generically says that the observed curved, with its ups and downs, was not caused by the hypothesized factors. /2

I was asked to elaborate on why this claim is nonsense, so a good opportunity to summarize a few things I wrote about game theory, machine learning, and Silicon Valley over the years.

https://twitter.com/beenwrekt/status/1611738182533668870

Exhibit 1. A short clip on just how much Silicon Valley runs on game theoretic concepts, in many cases decades adopted decades after they had been invented and languished in obscurity. link.medium.com/wtm0D49zqwb

A book a day on the Cold War era development of cybernetics, game theory, information science, and operations research: histories, biographies, autobiographies.

Day 1. Age of System: Understanding the Development of Modern Social Science by Hunter Heyck. play.google.com/store/books/de… Day 2 in the history of cybernetics etc. Rise of the Machines: the lost history of cybernetics by Thomas Rid. play.google.com/store/books/de…

@CzypionkaThomas Hab genau das Gleiche so im März 2020 geschrieben. Ist halt leider komplett durch die Realität widerlegt worden. Wissenschaftlich ausgedrückt: die Nullhypothese ist nicht falsifizierbar, ganz einfach weil die Nullhypothese viel näher an der Realität ist als die Kausalhypothese. @CzypionkaThomas Anders ausgedrückt, das was wir damals als kausal angesehen haben (Japan, Hongkong etc.) hat sich nicht in irgendeiner Form irgendwo anders replizieren lassen, insbesondere nicht in den populären Universalhypothesen "Maske wirkt" oder, enger umschrieben, "Maskenpflicht wirkt".

A single truth without a single source: a thread 🧵. Thinking about decentralized systems means thinking about how an organization can agree upon what it holds to be true and how this relates to the actual ground truth. /1 A “single source of truth” is a concept from information science, but it’s so universal that it can be applied to a whole variety of scenarios and organizations. /2

@prof_freedom @Karl_Lauterbach Aus 13 Studien wurde rausgelesen, dass bei 243 Probanden ein relatives Risiko für Maskenträger von 0.13 existierte. Leider ist so ein relatives Risiko unmöglich, ohne dass es zu einer massiven Diskontinuität in allen Diff-in-Diff Studien käme. Deshalb: Tonne. 5 Minuten. @prof_freedom @Karl_Lauterbach Wie schon mehrfach erwähnt, ein Faktor 0.95 oder vielleicht auch 0.9 in begrenzten Szenarien (vollbesetzte Züge) wäre ein plausibles und aufschlussreiches Ergebnis. Das hier ist der übliche Sondermüll. Lern Statistik, Kalle.

Probably quite unpopular opinion, but after two-plus years of far-below-capacity operations and the two-pronged demand shock of everyone catching up on travel and the 9-Euro ticket, the very fact that German railway is still functioning at all is an operational marvel.

https://twitter.com/veit_markus/status/1546439954380775426

The story that plays out with the complaints about rail operations (certainly justified, it's a mess) is the same story that plays out almost everywhere in the Western world: extremely stupid decisions on the political leadership level put massive strain on the operational level.

While Germany is awaiting the "independent" evaluation of its restrictive interventions, Switzerland 🇨🇭 already had theirs, and since the overall Swiss response didn't descend into lunacy like the German one, the process was comparatively low-key. Findings below. 1/n The evaluation was conducted by an economics think tank using a Diff-in-Diff approach (good) over hospitalizations (good) for the Winter 2020-21 wave (bad). 2/n

That whole "anti-disinformation" industry has become such a skanky grifter industry full of snake oil salesmen and piffle merchants. Beware of bullshitters claiming to protect your from other folks' bullshit. They usually only want to sell you their own bullshit. Warning signs: Conspicuous display of overblown academic credentials, and a splash banner promoting their "anti-disinformation" "science communication" (read: piffle) book targeted at the airport market. 🚨🚨🚨

🇩🇪 If you're curious why the German pandemic response, which started out OK-ish, has descended into lunacy, let's have a look at one of its architects: health minister Prof. Dr. Dr. @Karl_Lauterbach, an MD who also holds a Dr.sci. from @HarvardChanSPC in "health economics". /1 Let's take a closer look at this Harvard doctorate, which he rarely ever fails to mention. It's, well, roughly forty pages of vaguely philosophical musings puffed up to 100 pages using the old "typewriter font" trick. /2

Flow chart for sales pricing decision, from Richard Cyert, James March & C.G. Moore, "A specific price and output model", ch. 7 in A Behavioral Theory of the Firm (1963). archive.org/details/behavi… Who is C.G. Moore? Can't find any reliable evidence about his career after this. digitalcollections.library.cmu.edu/node/32143

The early phase of the Covid pandemic in Lombardy, spring 2020. In most provinces (including Lodi and Bergamo) Rt peaked in February, before the onset of mass testing. The paper reconstructs the beginning of the pandemic before the first public case in Codogno (Feb 14). It was already widespread in Lombardy by mid February, and showed no connection to population density.

To summarize a longer private conversation: In the long reckoning, no restrictive public measure will survive any rigorous scrutiny. It was all protection theater. Measures that made a difference were all in the backend, the access-to-care part, that nobody paid any attention to. That summarizes a claim I made a long time ago. If a country is diligently prepared for a pandemic, most public restrictions are unnecessary. If it's not, they might seem unavoidable, but ultimately pointless. Protection theater for a shell-shocked populace.

If you had Ames, Iowa on your bingo card for hotbeds of innovation research — Bingo. Indeed the technology adoption curve (aka the most important curve in history) was invented by rural sociologists at Iowa State College some 70 years ago.* A little history.

https://twitter.com/oliverbeige/status/1389247143022628864

WW2 put pressure on US food supplies, and a variety of technological innovations in agriculture promised improved yields. The rural sociologists were mostly concerned how these improved technologies could be made attractive to farmers, a notoriously skeptical bunch.

A few notes on coordination, adaptation, and the role of institutions.

"Both the organization theorist Chester Barnard and the economist Friedrich Hayek took adaptation to be the main purpose of economic organization, but with differences." — Oliver Williamson Nobel Lecture Unlike cooperation, which has taken on the meaning of "solving the Prisoner's Dilemma" and collaboration, which just generically means "working together", coordination suffers from being semi-defined, which often means one has to divine the type of coordination implied.

While we're killing time I can translate and summarize a podcast talk I gave in 2017 (in German) on how Donald Trump confounded the poll aggregators and won the election. Because clearly the message still hasn't gotten across, or we wouldn't still be waiting. 1/♾ The story started for me in March 2016 when a coworker predicted during lunch that Trump (still mostly a joke then) would win the election. I proposed that the only way for him to do it would be to stay on message and at the same time discourage Clinton voters from voting.

Very interesting thread about the relevance of Coase's Nature of the Firm for the platform age. I've written a few things about how working on a dynamic ridesharing* service in 2009 helped me connect Coase/Williamson to digital business models, so here's a few summary comments.

https://twitter.com/IgorLetina/status/1296540935266938882

The current focus of the public debate is the often contentious relationship between drivers and platform, but in order to understand the structure of the industry, it's also important to look at the underlying relationship, namely that between driver and passenger.

Hospital routing, hospitals in crisis, empty hospitals, whole hospital systems facing collapse. A short thread.

There are currently lots of seemingly conflicting news items about hospitals like the ones above. The dissonance is a sign of an information cascade. Finding the right balance between patient needs: give sick patients the best care available quickly, and hospital needs: have all resources in place to provide this care, is the ultimate determining factor in a successful response to Corona.

It's what I look at first.

A short thread on graph theory and network science. Both have long histories, but I'll focus on two people: Frank Harary, the "godfather of modern (American) graph theory", and Duncan Watts, the "reinventor" of network sociology, which morphed into network science. The idea that all kinds of bilateral relationships can be simplified to some kind of network structure is quite old (Euler's Königsberg bridge problem is canonical), but for the longest time it was considered "toy mathematics" at best.

https://twitter.com/oliverbeige/status/1165719446415183873?s=20

Machine scheduling, operations research, production, and machine learning. An anecdote thread for @sidwindc.

Biographical background: from age 15 to about 21, I spent my summers working for the laboratory equipment manufacturer that also employed my dad (in "construction"). After that I moved to prepoduction at Siemens, also for healthcare equipment, mostly coding production and tooling processes in CNC. This is where I stayed until my master's thesis.

I might be somewhere in this picture.

https://twitter.com/oliverbeige/status/954670416320180224?s=20