With Christmas around the corner, this variant cannot have better timing. I really expect that this variant takes over Delta before New Years, no matter what we do, unless a complete lockdown for Christmas.
Despite conventional wisdom and intuition, numbers don't back up that the U.S. SARS-CoV-2 curve exhibited any obvious change at Christmas (or Thanksgiving) in 2020.
Why focus on COVID hospital census and rate of change?
(1) We've repeatedly observed that case reporting trends are a mess around holidays. There's at least one day of low reporting of test results (and hence cases) due to the holiday, followed by a few days of higher than normal reporting due to the catch-up. Christmas is a particular mess: almost certainly more than 1 day of lower reporting, and then New Year's Day occurs just 1 week later. Reporting of daily hospital census, by contrast, is not much impacted by weekends or holidays.
(2) U.S. COVID numbers last year were increasing prior to these holidays. That they continued increasing after the holidays doesn't tell us much, especially if the rate of growth was decreasing. Those are simply the curves that we'd expect to see.
And that last point - rate of growth continuing to decrease - is largely what we saw right through the Thanksgiving and then Christmas / New Year's season.
We can't be certain of the counterfactual, so perhaps rates of change in hospital census would have declined faster without holidays. That's a subtle point that's difficult to prove one way or the other.
I have trouble taking Dr. Feigl-Ding seriously given that it seems like he's called roughly 30 of the last 3 legitimate reasons for concern, but maybe that's just my bias.
Agreed that if it was Feigl-Ding on his own I wouldn't take it seriously. The other sources very much don't do that, and his thread seemed useful in context.
Bergstrom generally a trustworthy source, Feigl-Ding a sensationalist one. Personally I can't stand Feigl-Ding's style, and think he has bad filter for communicating. But have to admit sometimes I do get information from his tweets I would have otherwise missed.
Just something I'd like to point out: when you wrote "We all agreed on Omicron and now we have to type Omicron all the time?", you probably meant "We all agreed on Nu and now we have to type Omicron all the time?", but that reference to Nu got caught up in your Find and Replace sweep you talked about at the start of the post.
There is one thing that I'll predict: natural immunity from previous infection only wanes much, much more than vaccine-induced immunity. The reason is that the recent variants have already been headed this way. Also, I've been right about every awful thing I've predicted (it's all in Coronavirus Updates on FB.) I still can't figure out how to get this psychic ability to work on Powerball numbers, though.
Hmm, I think the reason they passed over Xi is to avoid offending the PRC, who would be understandably annoyed if everyone associated a new and deadly variant with their leader.
Yup. I suspect this is Zvi’s sly sense of humour, not an oversight on his part. But I won’t be surprised if right wing media silos try to make it stick.
>I think this final graph is a bit confused here, unless ‘the original strain’ here means Delta. Delta had about a 120% advantage over 'the original strain’ or 70% over Alpha.
These kinds of transmissibility advantage numbers need to be taken with a mountain of salt. First of all, there are good reasons to think that if strain A has X times the transmissibility of strain B, and strain B has Y times the transmissibility to strain C, that strain C may *not* have XY times the transmissibility of strain A. First of all, there's the obvious fact that transmissibility depends on a host of ambient factors, and these kinds of "transmissibility advantage" calculations are never taken at the same time. Secondly, I wonder if stochastic factors could also play a role, think e.g. https://en.wikipedia.org/wiki/Intransitive_dice . But most importantly, we see this in the data. Different strains don't behave as if they had a fixed transmissibility advantage, even if it seems so initially, and one variant consistently "displaces" another one.
The market reaction is more worrying, but the market reaction is probably more based on measures instituted by various governments, and less on quantificaiton of individual risk.
So, if omicron really could be as vaccine evading as some fear it may be, shouldn’t we see evidence of it really soon? After all, in the original detection countries, not many are vacced — so type the virus of any vaccinated case presenting in the hospital, and see how many are omicron.
Put another way, shouldn’t each 24 hour period we _don’t_ have clinical evidence of vaccine breakthroughs at scale from omicron give us some additional confidence that it isn’t fully vaccine evading?
(I recognize there can be lots of reasons that this approach could be selecting on the dependent variable, but still think it tells us something)
Data gathering is currently on the level of saying the majority of those in hospital are unvaxxed, when the majority of the population is also unvaxxed. So not great. It's a statistical question, we won't know for a while for sure. BioNTech and Moderna will also do lab tests.
I considerably prefer Substack to Wordpress because I never got into the RSS habit, but check emails regularly, so getting email notifications when new posts are up is helpful. I consume your posts via Firefox's inbuilt TTS reader, so other formatting considerations are irrelevant to me (though, and I realize this would be a lot of work for a niche audience, so I mention it only for completeness' sake, it would be nice if tweets and other text-heavy images had their contents copied in plain text).
With Christmas around the corner, this variant cannot have better timing. I really expect that this variant takes over Delta before New Years, no matter what we do, unless a complete lockdown for Christmas.
Despite conventional wisdom and intuition, numbers don't back up that the U.S. SARS-CoV-2 curve exhibited any obvious change at Christmas (or Thanksgiving) in 2020.
Here's a thread looking at that period, focusing on COVID positive hospital census and its rate of change - https://twitter.com/dave_tagge/status/1385387774283370496
Why focus on COVID hospital census and rate of change?
(1) We've repeatedly observed that case reporting trends are a mess around holidays. There's at least one day of low reporting of test results (and hence cases) due to the holiday, followed by a few days of higher than normal reporting due to the catch-up. Christmas is a particular mess: almost certainly more than 1 day of lower reporting, and then New Year's Day occurs just 1 week later. Reporting of daily hospital census, by contrast, is not much impacted by weekends or holidays.
(2) U.S. COVID numbers last year were increasing prior to these holidays. That they continued increasing after the holidays doesn't tell us much, especially if the rate of growth was decreasing. Those are simply the curves that we'd expect to see.
And that last point - rate of growth continuing to decrease - is largely what we saw right through the Thanksgiving and then Christmas / New Year's season.
We can't be certain of the counterfactual, so perhaps rates of change in hospital census would have declined faster without holidays. That's a subtle point that's difficult to prove one way or the other.
Thoughts on this Carl Bergstrom thread? https://mobile.twitter.com/CT_Bergstrom/status/1464058581502738435
I have trouble taking Dr. Feigl-Ding seriously given that it seems like he's called roughly 30 of the last 3 legitimate reasons for concern, but maybe that's just my bias.
Agreed that if it was Feigl-Ding on his own I wouldn't take it seriously. The other sources very much don't do that, and his thread seemed useful in context.
Bergstrom generally a trustworthy source, Feigl-Ding a sensationalist one. Personally I can't stand Feigl-Ding's style, and think he has bad filter for communicating. But have to admit sometimes I do get information from his tweets I would have otherwise missed.
For example this tweet was useful info: https://twitter.com/DrEricDing/status/1464542390350487554
I think Bergstrom again makes good points in his thread.
Just something I'd like to point out: when you wrote "We all agreed on Omicron and now we have to type Omicron all the time?", you probably meant "We all agreed on Nu and now we have to type Omicron all the time?", but that reference to Nu got caught up in your Find and Replace sweep you talked about at the start of the post.
There is one thing that I'll predict: natural immunity from previous infection only wanes much, much more than vaccine-induced immunity. The reason is that the recent variants have already been headed this way. Also, I've been right about every awful thing I've predicted (it's all in Coronavirus Updates on FB.) I still can't figure out how to get this psychic ability to work on Powerball numbers, though.
Hmm, I think the reason they passed over Xi is to avoid offending the PRC, who would be understandably annoyed if everyone associated a new and deadly variant with their leader.
Yup. I suspect this is Zvi’s sly sense of humour, not an oversight on his part. But I won’t be surprised if right wing media silos try to make it stick.
They're trying. It will work about as well as things usually work.
BNOnews is great. They also have a map tracking spread of omicron: https://bnonews.com/index.php/2021/11/omicron-tracker/
>I think this final graph is a bit confused here, unless ‘the original strain’ here means Delta. Delta had about a 120% advantage over 'the original strain’ or 70% over Alpha.
These kinds of transmissibility advantage numbers need to be taken with a mountain of salt. First of all, there are good reasons to think that if strain A has X times the transmissibility of strain B, and strain B has Y times the transmissibility to strain C, that strain C may *not* have XY times the transmissibility of strain A. First of all, there's the obvious fact that transmissibility depends on a host of ambient factors, and these kinds of "transmissibility advantage" calculations are never taken at the same time. Secondly, I wonder if stochastic factors could also play a role, think e.g. https://en.wikipedia.org/wiki/Intransitive_dice . But most importantly, we see this in the data. Different strains don't behave as if they had a fixed transmissibility advantage, even if it seems so initially, and one variant consistently "displaces" another one.
See e.g. https://cspicenter.org/blog/waronscience/is-the-delta-variant-really-more-than-twice-as-transmissible-as-the-original-strain-of-the-virus/ and https://cspicenter.org/blog/waronscience/the-british-variant-of-sars-cov-2-and-the-poverty-of-epidemiology/ on two blog posts by the excellent Philippe Lemoine on this theme.
The market reaction is more worrying, but the market reaction is probably more based on measures instituted by various governments, and less on quantificaiton of individual risk.
So, if omicron really could be as vaccine evading as some fear it may be, shouldn’t we see evidence of it really soon? After all, in the original detection countries, not many are vacced — so type the virus of any vaccinated case presenting in the hospital, and see how many are omicron.
Put another way, shouldn’t each 24 hour period we _don’t_ have clinical evidence of vaccine breakthroughs at scale from omicron give us some additional confidence that it isn’t fully vaccine evading?
(I recognize there can be lots of reasons that this approach could be selecting on the dependent variable, but still think it tells us something)
Data gathering is currently on the level of saying the majority of those in hospital are unvaxxed, when the majority of the population is also unvaxxed. So not great. It's a statistical question, we won't know for a while for sure. BioNTech and Moderna will also do lab tests.
Remember, experiments are still banned.
I considerably prefer Substack to Wordpress because I never got into the RSS habit, but check emails regularly, so getting email notifications when new posts are up is helpful. I consume your posts via Firefox's inbuilt TTS reader, so other formatting considerations are irrelevant to me (though, and I realize this would be a lot of work for a niche audience, so I mention it only for completeness' sake, it would be nice if tweets and other text-heavy images had their contents copied in plain text).