I appreciate the reply, and I agree that the concentration of antibodies in the blood has been observed to drop with time. That's relatively common in many diseases though, and doesn't mean that the patients no longer have any useful immunity. The test thresholds were set for best (but still imperfect) discrimination between known positives (mostly severe cases) and known negatives, and there's no specific reason to believe they predict when a recovered patient becomes susceptible again. They're also testing for IgG, when we know that T cell and IgA immunity are important. One of your papers mentioned IgM, which is expected to drop quickly to undetectable levels while the patients retain immunity (not to say you suggested otherwise, of course).

I also agree that whatever immunity patients get after a mild or asymptomatic case is likely to be weaker than after a severe case. That's one case where my calculation above could be wrong--if there are many reinfections but the first or second case is always very mild, we might be much more likely to miss those. That would still be good news for the patients, though bad news for the overall population if they're still comparably infectious.

Finally, even if a patient's immunity degrades to the point they no longer exhibit sterilizing immunity (i.e., the virus still replicates a little at first), in most diseases they won't get as sick as the first time. So even if the coronavirus becomes endemic (which seems relatively likely, since many countries will lack the resources to eradicate it even with a vaccine), I expect the cost in mortality from whatever reinfections do occur to be far lower than what we're seeing now. The opposite of that (antibody-dependent enhancement) does occur, and was a specific concern here because vaccine studies for the original SARS showed evidence of that. So far vaccine studies for SARS-CoV-2 do not, though.

I actually thought the SF Chronicle article wasn't terrible, more pessimistic in its conclusions and tone than I would be given the same evidence but with many of the points above. Their headline seems irresponsible to me though; even if durable sterilizing immunity were impossible, a vaccine that cut the IFR by a factor of ten would be tremendously valuable to the elderly. All that nuance is lost when people just say "reinfection is possible". I was probably too strong to say "sowing public panic", but I do believe your comments paint a falsely pessimistic picture of the current science, and that this false pessimism can be harmful later (e.g., by causing people not to seek a vaccine because of something they half-understood about immunity). Specifically, I also believe the absence of confirmed reinfections out of places like Sweden is strong evidence that immunity usually lasts >3 months. If you were claiming that reinfection might be common after a year, then I'd be much less sure (though I'd still guess probably not based on the original SARS).

In any case, I certainly agree that younger people shouldn't get themselves deliberately infected in search of whatever immunity that affords (though the death rate among young people is low enough that I doubt reinfection would change the calculus for anyone considering that either way). I also agree that the USA response has been terrible, and resulted in a lot of avoidable death--I'm not sure, but it seems possible to me that just with universal mask use and good hygiene (like in Japan), we could live otherwise normal life with negligible spread.

ETA: And here's a paper showing neutralizing antibodies for at least three months (the limit of the study, which they're continuing) in New York. It seems beyond any reasonable doubt to me that immunity lasts three months, and I believe you're simply wrong to question that. Longer gets more speculative, but I think it's quite likely.

https://www.medrxiv.org/content/10.1101/2020.07.14.20151126v...