It would show up as people with a particular year of birth having a much lower risk than people born one year earlier or later. Since most research includes collecting date of birth, this should be easy to check.
jimrandomh(Jim Babcock)
Karma: 16,488
History’s Biggest Natural Experiment
Biological global catastrophic risks were neglected for years, while AGI risks were on the top.
This is a true statement about the attention allocation on LessWrong, but definitely not a true statement about the world’s overall resource allocation. Total spending on pandemic preparedness is and was orders of magnitude greater than spending on AGI risk. It’s just a hard problem, which requires a lot of expensive physical infrastructure to prepare for.
“Impaired consciousness” doesn’t sound unusual for patients with severe fever, but five strokes out of 214 hospitalized patients is pretty noteworthy.
It’s a newspaper article based on an unpublished paper; that reference class of writing can’t be trusted to report the caveats.
(I could be wrong about the mechanics of PCR; I’m not an expert in it; but the article itself doesn’t provide much information about that.)
This can only be used on groups where everyone is asymptomatic, and there will be low limits on the pool size even then.
The first step of a PCR test is RNA amplification; you use enzymes which take a small amount of RNA in the sample, and produce a large number of copies. The problem is that there are other RNA viruses besides SARS-CoV-2, such as influenza, and depending when in the disease course the samples were taken, the amount of irrelevant RNA might exceed the amount of SARS-CoV-2 RNA by orders of magnitude, which would lead to a false negative.
tl;dr: Someone wrote buggy R code and rushed a preprint out the door without proofreading or sanity checking the numbers.
The main claim of the paper is this:
The total number of estimated laboratory–confirmed cases (i.e. cumulative cases) is 18913 (95% CrI: 16444–19705) while the actual numbers of reported laboratory–confirmed cases during our study period is 19559 as of February 11th, 2020. Moreover, we inferred the total number of COVID-19 infections (Figure S1). Our results indicate that the total number of infections (i.e. cumulative infections) is 1905526 (95%CrI: 1350283– 2655936)
So, they conclude that less than 1% of cases were detected. They claim 95% confidence that no more than 1.5% of cases were detected. They combine this with the (unstated) assumption that 100% of deaths were detected and reported, and that therefore the IFR is two orders of magnitude lower than is commonly believed. This is an extraordinary claim, which the paper doesn’t even really acknowledge; they just sort of throw numbers out and fail to mention that their numbers are wildly different from everyone else’s. Their input data is
the daily series of laboratory–confirmed COVID-19 cases and deaths in Wuhan City and epidemiological data of Japanese evacuees from Wuhan City on board government–chartered flights
This is not a dataset which is capable of supporting such a conclusion. On top of that, the paper has other major signals of low quality. The paper is riddled with typos. And there’s this bit:
Serial interval estimates of COVID-19 were derived from previous studies of nCov, indicating that it follows a gamma distribution with the mean and SD at 7.5 and 3.4 days, respectively, based on [14]
In this post I collected estimates of COVID-19′s serial interval. 7.5 days was the chronologically first published estimate, was the highest estimate, and was an outlier with small sample size. Strangely, reference [14] does not point to the paper which estimated 7.5 days; that’s reference 21, whereas reference 14 points to this paper which makes no mention of the serial interval at all.
Right now, most people are hyperfocused on COVID-19; this creates an obvious incentive for people to try to tie their pet issues to it, which I expect a variety of groups to try and which I expect to mostly backfire if tried in the short run. (See for example the receptiontthe WHO got when they tried to talk about stigma and discriminatio; people interpreted it as the output of an “always tie my pet issue to the topic du jour” algorithm and ridiculed then for it. Talking about AI risk in the current environment risks provoking the same reaction, because it probably would in fact be coming from a tie-my-pet-topic algorithm.
A month from now, however, will be a different matter. Once people start feeling like they have attention to spare, and have burned out on COVID-19 news, I expect them to be much more receptive to arguments about tail risk and to model-based extrapolation of the future than they were before.
To start, the severity estimates that Joshua assumed were worst case and are implausible. The very alarmist Fergeson et al paper has much lower numbers than Joshua’s [Joscha Bach’s] claim that “20% will develop a severe case and need medical support to survive.”
I believe the 20% figure comes from the WHO joint report which says
13.8% have severe disease (dyspnea, respiratory frequency ≥30/minute, blood oxygen saturation ≤93%, PaO2/FiO2 ratio <300, and/or lung infiltrates >50% of the lung field within 24-48 hours) and 6.1% are critical (respiratory failure, septic shock, and/or multiple organ dysfunction/failure).
There are a lot of modeling assumptions that go into this, and the true number is probably lower, but not so low as to invalidate Joscha’s point.
Thank you, this is exactly the sort of clever analysis I was hoping people would come up with when I wrote my post.
This site has floor-plan images of Diamond Princess cabins, from which we can make a few inferences about cabin occupancy. Most of the cabin layouts contain a single bed which fits two people, so two-person cabins will almost exclusively couples sharing a bed. If I assume the rate at which people in single-person cabins get infected (8%) is the rate of infection outside the cabin, and that the higher rate of infection in two-person cabins is caused entirely by within-cabin secondary transmission, then it looks like each person would have to infect their partner an average of 1.5 times each. This also tells us that the transmission rate between elderly couples sharing a bed is likely to be extremely high, and also that people in single-person cabins must be different in some way—perhaps they spent less time in the ship’s common areas.
Three- and four-person cabins seem harder to interpret. These would originally have been couples with children, but there aren’t many children aboard as of Feb 5th, and they probably moved people around to free up single cabins for extra-vulnerable people and for confirmed cases that they needed to isolate.
This paper analyzes specific incidents in which a group of one infected person plus some uninfected people sat down together, and some uninfected people got it. They find a secondary attack rate (from mostly non-household interactions) of 35%.
There are two big issues that prevent this paper from being used to draw good inferences about the household secondary attack rate. First, the incidents were found by specifically looking for superspreading events, and does not include any events where transmission didn’t happen. And second, the events are single gatherings, whereas living with someone may involve many opportunities to get infected.
The main piece of data that would help answer this question is case-studies of past vaccines, whether they had safety problems and what those problems were, and when the problems manifested. Given that there’s a new influenza vaccine every year, and I’ve never heard of any year’s influenza vaccine being rejected on safety grounds, my guess is that 18 months is much too conservative.
The indentation of the table of contents is determined by the heading levels. It looks like you may have set some of the headings to “Heading 2” style and others to “Heading 3″ style in Docs, then adjusted the font size to make them look the same. If you use Docs’ heading format presets and use the same one for all the sections, they should be at the same indent level.
Unfortunately the LW codebase doesn’t support multiple authors with edit access to the same post yet; we’re working on this (as part of a broader overhaul of the post editor, which will allow Google Docs-style simultaneous editing), but it isn’t ready yet. In the mean time, the easiest way to handle this is to make edits in a shared Google Doc, and have the primary author paste them in and save. (Copy-paste between the post editor and Google Docs should just work.)
(I added Roko to the metadata as a coauthor. Tagged coauthors are a beta-feature which currently can only be edited by moderators, since we haven’t implemented the mechanics of having authors approve each other.)
In this post I collected papers which estimate the incubation period (time from exposure to symptom onset) and the serial interval (time from exposure to infecting the next person in the chain). Studies get varying results because they’re done in different populations and have methodological differences, but find reasonably similar medians. A few of them also provide full distributions over incubation periods and serial intervals.
What does it mean for the future? That it takes about a week of a severe lockdown to switch from the exponential growth to linear, about two weeks to switch to leveling off, and three to four weeks to start seeing a meaningful decline.
This is because of the incubation period (3-14d) and the delay between people becoming symptomatic and getting tested. The reduction in transmissions is immediate, it just takes awhile to notice.
R0 is the number of people that each person will go on to infect, on average. R0 for COVID-19 is high compared to other common diseases, indicating high transmissibility.
I avoided stating a quantitative estimate of the attack rate because my confidence intervals are too wide to be useful. If I had to bet, I’d say 90% CI 15-85%, 50% CI 30-65%. I’m hoping people can gather weak evidence of various forms (secondary attack rates and R0s of other diseases, anecdotes in which household members do or don’t get it, or in the best case a dataset with household memberships labelled).
For now, mods do it. This will be expanded to all users and to a wider variety of tags when tagging leaves beta.