I want to know your estimate for how many people you think currently have it within the San Francisco Bay Area. Not how many people who have been measured to have it, since our measurement is garbage. Answers can be given as a distribution or point estimate.

New to LessWrong?

New Answer
New Comment

3 Answers sorted by


Mar 04, 2020


Do we have a concise time-range-line for progress of infection in an individual? Something like (note: numbers made up, I'm looking for better ones):

  • 0 - contact with source of infection.
  • 2-5 days - virus replicated enough to be contagious
  • 4-6 days - testing would show positive
  • 8-15 days - symptoms begin
  • 12-20 days - recovery or death
  • 15-30 days - no longer contagious

Models come out VERY different depending on the distribution of timings you assign to these points. Specifically, the time from contagion-capable to detection-and-isolation is the difference between hundreds of already-infected and tens of thousands.


Mar 01, 2020


Conservative Main land china population: 1.3 billion

High end Number of infections in China after 2+ months of infections: 80,000

Chance of getting disease = 80,000/1,300,000,000 = 0.006% (Rounded up)

Death rate = ~2% (higher if age >70, 0.2 if under 40).

Bay area population 7 million Bay area x contagious rate = 400 people Actual number with it after February ~ 0

Likely due to younger age of population, but also fairly low infection rate to begin with.

Across US Population = 300,000,000 High Estimate of infections based on infection rate = 18,000 Multiple by 6 to get a full year, would be 100,000; death rate would equal 3,000 (This seems way to high to me)

Bayesian updated due to awareness, preparation and suspicion I would actually estimate about a tenth of cases. Yes other countries haven’t had a chance to spread disease, but china is actually controlling it since the initial outbreak.

Given an additional 10 months in 2020 but weighed against the actual very few infections here at present. I stick with the same estimate of 10,000, deaths about 300.

Given influenza past infection season numbers from 100,000 - 1,000,000 this seems pretty low.

It appears influenza has a higher infectious rate but lower death rate. This fits given that influenza has established itself across continents and has a readily available vaccine, and while the vaccine may not fully prevent the disease, it often provides some protection and has been proven to lower death rates of populations who receive it.

I suspect in following years coronavirus may reach these numbers once the virus has dispersed sufficiently across countries but with better protocols and possible vaccine in future years it will likely be similar to influenza based on the numbers presented.

Curious what other peoples estimates are.

Curious if people start pulling out the plague bird masks.

(If they used bird masks to scare mice, we should use human or lion masks to scare pangolins. Or we should use hawk masks to scare bats. If we use hawk masks, they need a hawk call)

To answer your question, given that there are between 0-4 cases at the end of February, I estimate 0-4 cases at the end of February.

Good day.

Do I win?

Pee Doom

Mar 03, 2020


I think it's ~1000 for that day. I don't know how long deaths take to bear out - is it 14 days? If it's that, then if we have ~34 deaths in 14 days then that suggests this prediction is right.

We probably have a greater than average transmission rate/doubling time because no one is tracking it, and I'm pessimistic enough about the U.S. healthcare system at by this point that we may not even be tracking fatalities accurately.

I will update my answer as I think about this more clearly.

2 comments, sorted by Click to highlight new comments since: Today at 2:35 PM

Is it true that the Bay area recently started massive testing on people with symptoms but without suspicious travel history? If yes, then that should soon give an update on whether it looks more like 20-50 cases or 50-200. I find the latter range more likely given how things went in other countries, but I wonder whether I should update downward because there seem to have been no new cases reported in the last 24h(?), and I'd imagine(?) that there' must a be lot of testing going on by now.

No, there has barely been any testing. I think it's more like 200-1000 cases.