DOOM: "Omicron Covid cases ‘doubling every two to three days’ in UK"

[jamesbrown]

I would worry less about whether scientists have an agenda and more about whether they are producing something useful (and I say that as a scientist myself). It seems to me that they are not very good at forecasting covid epidemiology (e.g., look at the history of the Imperial model, among many others). Creating vaccines, otoh.

[AtW]

It seems to me that they are not very good at forecasting covid epidemiology (e.g., look at the history of the Imperial model, among many others). Creating vaccines, otoh.

Change in behavior changes models - 146k dead so far from Covid and Imperial said 500k (without any changes in behavior at the time when no vaccines were available), if anything Imperial was on a low end.

[OwlHoot]

Yes, it's more complicated than just antibodies but the fact is that people are catching it twice, and catching it after being double jabbed - but in general doing better than they would be if they didn't have some immunity.

Also, let's not forget the medical advances aren't just about vaccines. There has also been development of drugs to ameliorate Covid symptoms.

[jamesbrown]

Change in behavior changes models - 146k dead so far from Covid and Imperial said 500k (without any changes in behavior at the time when no vaccines were available), if anything Imperial was on a low end.

That is a very weak argument. If that were true, the predictions would've become better over time as they learned to better incorporate the behavioural response or, as a minimum, scenarios for different actions. Some good examples here.

https://twitter.com/julianHjessop/st...08978070466572

The reality is that forecasting covid is very hard indeed, as with many other things that involve a complex aggregation of physical/human factors and non-linear feedbacks like, say, the stock markets. We should mostly ignore these covid forecasts because they have very little skill and tend to be biased high, which is even worse.

[AtW]

That is a very weak argument.

It's a very good argument specifically for that Imperial study - it said that there will be 500-600k dead in the UK _IF_ virus is allowed to rip through to get that magical "herd immunity".

We know that 146k actuall died despite lockdowns, and at best 20% of population were "naturally" infected, so it follows that 500-600k dead count was actually almost spot on, again if virus was allowed to rip through, which thankfully did not happen.

[jamesbrown]

It's a very good argument specifically for that Imperial study - it said that there will be 500-600k dead in the UK _IF_ virus is allowed to rip through to get that magical "herd immunity".

We know that 146k actuall died despite lockdowns, and at best 20% of population were "naturally" infected, so it follows that 500-600k dead count was actually almost spot on, again if virus was allowed to rip through, which thankfully did not happen.

Yours is an untestable proposition because it didn't happen. On the other hand, there are some eminently testable predictions, as indicated in the thread I posted. The testable predictions turn out to be woeful. Thus, your argument is based on unverifiable speculation about an unknown (what the response was, what the underlying death rate might have been). We should focus on the predictions that are actually testable (because, if these are bad, we can have very little confidence in the untestable ones). Predictions that are inherently untestable are largely non-scientific, even if they may have some policy impact.

[AtW]

Yours is an untestable proposition because it didn't happen.

Global thermonuclear war also did not happen, yet models that suggest hundreds of millions will die pretty quickly are testably correct, just not worth testing it.

Same here - imperial model was clearly correct in terms of forecast deaths (if virus was allowed to go unchecked).

[jamesbrown]

Global thermonuclear war also did not happen, yet models that suggest hundreds of millions will die pretty quickly are testably correct, just not worth testing it.

Same here - imperial model was clearly correct in terms of forecast deaths (if virus was allowed to go unchecked).

This is a facile analogy because, very obviously, the impact of a nuclear explosion and epidemiological forecasting have nothing in common in terms of the uncertainties involved and we have direct evidence for the former whereas the latter involves a counterfactual. 1/10.

[d000hg]

Have we got an R for Omicron yet? When they say Omicron is far more transmissible than Delta, which was already far more transmissible than Covid Classic, which was itself already very transmissible, it makes me think surely there must be an upper limit. I got the impression that even with Delta, you were almost certain to catch it if you were in close contact, similar to chicken pox.

edit: https://en.wikipedia.org/wiki/Basic_...tious_diseases

I think they must have downgraded Delta's R value, because I'm sure it was right near the top of the list before. Looks like anything up to 3X worse is possible! DOOM!

[xoggoth]

Don't forget to sanitise your Christmas cards before you open them this year.