COVID-19: a sourced timeline
Origins: the hypotheses, and what they were called → What the US government said about vaccines and transmission → Shoshin: a beginner's mind updates on this record →
Nothing on this page is our own reporting. Every row is somebody else's document, named and linked, and the link is the point of the row.
The horizontal axis is time. The vertical axis carries no magnitude and no ranking. Rows are stacked in date order for one reason: to give every document a place to sit where you can see and click its citation.
Every source is labelled by what it is. primary means the instrument or study itself. reporting means a journalist's account, used only where no primary document exists. no document means exactly that, and it is stated rather than papered over.
Where we made a judgment call, it says so on the row. Dates that a source left ambiguous, and superlatives that depend on a definition, are marked "our call" in the open. The vaccine rows are one standing call worth stating here: each is placed at the date of its first authorization anywhere, not its first dose, because authorization is the act that leaves a document. Where the two differ materially, the row says so.
Our voice appears in exactly two places: the choice of which events share this axis, and the lines marked "our call". Everywhere else the page reports what a source says and attributes it. We have a position, stated below in its own words so you can argue with it rather than having to infer it.
The argument, stated plainly
This is ours, and it is the only claim on the page that is. The events above, each individually documented, are placed on one axis because we think the sequence shows something the events do not show apart: public health authority was applied unevenly, and the unevenness tracked the politics of the activity rather than its epidemiology. Beaches on an open coastline were closed by signed order three weeks after the first published finding that transmission outdoors was 18.7 times less likely than indoors, and stayed closed after a second paper found one outdoor cluster in a Chinese national dataset; one mass gathering was studied as the country's largest superspreader event while another was endorsed by 1,288 public health professionals; the official who signed the orders dined in violation of his own guidance. Meanwhile the vaccines arrived in an order set by which regulator was willing to move first, not by which vaccine was best evidenced — the earliest authorization on this axis had no Phase 3 data and was for soldiers only.
We are not asking you to accept that from the picture. The picture is only an index of documents. Follow the links, then decide.
The strongest objections to it
These are not strawmen. If we lose the argument, we expect to lose it on one of these.
- The events are not the same kind of thing. A health officer's order, a working paper, a signed letter and a dinner are different categories of act by different actors. Lining them up on one axis creates an appearance of a single agent behaving inconsistently when in fact these are separate institutions, in separate jurisdictions, answering to separate incentives.
- Juxtaposition implies causation and coordination that no source here demonstrates. Nothing on this page shows that the people who closed the beaches are the people who signed the letter, or that either influenced the other. A timeline is a very effective way to insinuate a link without ever having to make the claim explicit.
- The open letter made an argument the framing here flattens. Its authors did not claim protests were epidemiologically safe. They argued that racism is itself a public health emergency and that the response to outdoor protest should differ from the response to protests demanding an end to the mitigations. That is a contestable position, but it is a position, not simple hypocrisy — and it is worth reading in the original before deciding.
- The outdoor evidence was thinner in March than the timeline makes it look. Nishiura was a preprint, not peer reviewed, built on 110 traced cases in one country, with a confidence interval running from 6 to 58. Qian's data came from a population that was locked down, so outdoor exposure was suppressed in the sample itself. "Less likely outdoors" is also not "safe when a hundred thousand people share a beach" — the closures were argued as crowd-density control, not as claims about sand. Placing a preprint next to an order does not establish that the officials had, or should have credited, a finding that would have changed the decision.
- Vaccine authorization dates measure regulators, not vaccines. Reading the vaccine rows as a race is a category error. The order reflects which agency used which emergency power, how much risk each state was willing to carry, and where each manufacturer filed first — and a military-only clearance with no Phase 3 data is not the same kind of act as an authorization following published efficacy results, however similar the two rows look.
- The list is a sample, and we picked it. A different set could support a different conclusion. The selection is the argument, which is exactly why we say so rather than presenting this as a complete record.
How to argue with this page
You don't have to persuade us. This page isn't published the way an article is, where the author keeps the pen and you get a comment box underneath. Everything you see here — every date, every link, every word of the argument above — sits in a single public file that anyone is allowed to edit. You change what you think is wrong and send us your version. We either accept it, publicly, or we have to say publicly why not. Both the change and the refusal stay on the record where anyone can read them.
- Something here is factually wrong. A date, a name, a quotation. Correct it and show the document you're correcting it from.
- You have a better source. Especially one that replaces a journalist's account with the original document. The 30 April Orange County beach closure is the open question: if a signed order exists, whoever finds it has done more than this whole page has.
- An event is missing — particularly one that undercuts us. Those are worth more here than ones that help us, and they get added on the same terms as anything else.
- The argument itself is wrong. Rewrite it. It is the one part of this page we actually wrote, so it is the part most owed a rewrite.
And if you'd rather not argue with us at all, take the whole thing. You are free to make a complete copy of this page — our events, our sources, our argument and all — and publish your own version of it under your own name, with different events and a different conclusion. You need nobody's permission and there is nothing to ask for. That isn't us losing an argument; it is the argument working as intended, because now two versions exist and both can be checked against the same documents. Send us yours and we'll link to it from here.
The buttons below are how that actually happens. It runs on GitHub, which is a tool programmers use to edit things together and to keep an unforgeable record of who changed what; you do not need to be one to use it, but you will need to make a free account. In its vocabulary, sending us your version is a "pull request" and taking your own copy is a "fork".
See the file everything lives in Send us a change Take your own copy Or just tell us what's wrong
All events
| Date | Event | Grouping | Source |
|---|