[dalore]

Yes, this confirms what a few people have been trying to say, that we do have some form of prior immunity, it's not a novel virus in that everyone is 100% susceptible, and that anti body tests won't show up prevalence. In addition means we need a lower rate for herd immunity, if we haven't already got it in some places like New York and London. So the models of exponential growth was unfounded and it follows more a natural gompertz curve.

[graeme]

Not quite. Carl Bergstrom had a good thread on this. Basically, we saw a certain rate of growth. If this rate of growth happened despite half the population having T cell immunity, then it means the R0 must be higher than we thought, and therefore the herd immunity threshold in the susceptible population rises.

You do get a somewhat lower overall herd immunity threshold, but less than you thought.

https://twitter.com/CT_Bergstrom/status/1290744393796694016

[timr]

That doesn't make any sense. The key part of Bergstrom's logic is here:

"But ex post, holding the epidemic trajectory constant, it means that the disease is more transmissible per contact event than you expected."

...but that's wrong. The epidemic trajectory isn't constant (R0 is not an innate property of a virus; it changes with context), and it's pretty likely that we've completely mis-measured it by sampling from the wrong population. Our estimates for R0 are mostly model extrapolations from PCR testing data, which are not uniformly sampled from the population. If anything, in fact, they're heavily biased toward the susceptible population.

In other words, because we're sampling disproportionately from the most susceptible population, our estimates of R0 already seem high. If you sampled randomly from the population of less-susceptible people, you would see that the R0 is much lower than you originally predicted.

On top of that, there's absolutely no reason to believe that cellular immunity would make it seem like R0 is lower in any particular population: T-cell-mediated immunity is a late response. It can reduce the severity of an infection, but it's still highly likely you'd show up as a positive PCR test, and count toward an R0 estimate.

So you have a measurement based on a test that isn't strongly affected by cellular immunity, derived from a population biased toward the most susceptible individuals. If you have to doubt something here, you doubt the estimate of R0 (i.e. the model), not the observations of reality.

[mzs]

I don't see what your point is. Gompertz is just one type of a logistic function. A first year epidemiology course covers them saying that they model most outbreaks and you have example exercises with them where you learn how to calculate things like when will 1m people be infected. One way to think about them is that they have two phases, the first dominated by an exponential curve and the second by a logarithmic. This might be what you are confusing.

[hcknwscommenter]

" In addition means we need a lower rate for herd immunity"

It's like you didn't even read the article. At MOST, it means we MIGHT need a lower rate for herd immunity.

[acqq]

> In addition means we need a lower rate for herd immunity

No, it actually doesn't, the math functions differently than what is naively believed. See:

https://statmodeling.stat.columbia.edu/2020/08/03/math-error...

The R0 would have to be corrected to the higher value, which would mean that the herd immunity threshold for the susceptible part of population would have to be higher(!)

Also, even among completely asymptomatic a high viral load is surely observed, implying that they aren't "immune" in the sense that they aren't the ones who continue to spread the infection -- but even for that, more research is needed.

Which means, until we know more, that those who are in danger aren't less in danger at all until they get a vaccine. With all the consequences we already know, like in the states which were strongly hit.

> the models of exponential growth was unfounded

I don't believe anybody ever claimed that the exponential growth would be indefinite. However, exponential growth through the susceptible part of population in one important phase of the spread (before any interventions are taken) was indeed a good enough approximation of what was observed.

Edit: re: the comment below: "it still means that herd immunity is easier and faster to achieve" -- no, not necessarily, see what I wrote about the "asymptomatic." For all we know at this moment, they get the virus and they spread it. If they just don't get sick it is then different from them not being involved in the transmission. And if some amount of people are identified who can't get it, the result can still mean "for two more years it has to be like it was up to now."

Edit 2: re the comment below: "That maths you link to literally says that the threshold for herd immunity among the entire population is indeed lowered." No it doesn't, not in the sense "it is better". Try to work from R0 = 2.7 which was observed. Assume 50% can't even spread. The R0 for the remaining 50% is then not 2.7, but has to be scaled up to 5.4(!) Which means, e.g. that the 90% of the 50% have to become immune to reach the herd immunity. And if up to now e.g. 10% of these were infected in past 6 months, that means 6 months * 8 = 4 years more to go. Whereas, when no 50% "immune" population exist, if with given R0 2.7 the threshold is e.g. 60%, and we have 10% infected, we have only 5 times 6 months to go, i.e. 2.5 years(!). You see, it's nothing so obvious like one likes to believe. Like Ed Young said: "Immunology Is Where Intuition Goes to Die." Search for the article (it's not about this calculation, but about how simple mental models don't work with immunology on other levels too).

The herd immunity threshold is dependent of R0, and that is exactly what they try to explain in the post. But like I've said, we don't even know how many of "asymptomatic" are part of all equations. And which equations are the right ones. The "right" models could be even more complex than what I've used here.

Edit 3: re: Citing Judith Curry. Judith Curry, the only qualification having from being one of the most favorite "experts" to be cited by climate deniers. But what she claims is still wrong: https://www.skepticalscience.com/skeptic_Judith_Curry.htm

[codeulike]

That maths you link to literally says that the threshold for herd immunity among the entire population is indeed lowered. And that was the original posters point.

I think you're getting too caught up in the details. Its still good news, and it still means that herd immunity is easier and faster to achieve

[acqq]

> Its still good news, and it still means that herd immunity is easier and faster to achieve

It is not even clear it's "good news", it doesn't "still mean" at all, read carefully under "Edit 2" of my parent post. It can also mean that it will take longer. R0 must be reevaluated, and threshold is f(R0). We don't even know the "correct" function for the threshold, even less which people contribute in which way. We don't know how much were infected up to now, the high numbers weren't random samples, prevalence in global population is much lower than in a few hot spots where a lot of deaths happened. "Math is hard." Immunology is complex.

[guscost]

> R0 must be reevaluated

I'm at the point of concluding: "All R0-based models are useless for a disease like this." The true dynamics are far too complex, and this assumption that we can conclude anything by plugging linearly-varying (or constant!) parameters into a trivially non-linear model is absurd.

[hcknwscommenter]

"That maths you link to literally says that the threshold for herd immunity among the entire population is indeed lowered. "

The maths you refer to do not make that judgment at all. They indicate that, assuming all the assumptions that go into those models are true (they aren't), herd immunity may be lowered, but not by much.

[dalore]

Your maths fails in real life. Seems to be only 10-20% of anti body prevalence is needed for herd immunity due to t-cell protection that wasn't taken into account. The Diamond Princess cruise ship, perfect floating petri dish for experiements, bore this out. Same with London and New York, reach the herd immunity levels of 10-20% and died out.

Try this maths https://judithcurry.com/2020/05/10/why-herd-immunity-to-covi...

[tathougies]

> So the models of exponential growth was unfounded and it follows more a natural gompertz curve.

The models of exponential growth were always unfounded. Exponential growth would have you believe that in a few short months, a sum larger than the entirety of the human population would get the virus. We knew then that it was ridiculous. The question wasn't if it was exponential or gompertz, but what the coefficients were.

Those who parroted the exponential curve were just trying to use scare tactics to drive clicks to their websites.

[dalore]

At the start the only person trying to say it was a gompertz curve was Nobel Prize winner Professor Levitt. Everyone was using the Imperial College model which did say exponential growth and of which the governments around the world used to make policy. So to turn around and say now the models were unfounded is not useful as they were the models that were followed to make huge policy decisions. So it wasn't just people driving clicks, but governments making drastic changes to peoples lives based on these flawed models.

https://unherd.com/thepost/nobel-prize-winning-scientist-the...

[beagle3]

There were quite a few others; it just happens that Michael Levitt is the highest profile person on that list, and essentially the only one to therefore be globally quoted;

But there are iirc ex surgeon-general equivalents in at least Germany and Israel and independents like Karl Deninger who had been saying essentially the same, with data to back themselves up, since March.