Essentially anything's possible. It's possible that if I bought ten lottery tickets in a row, I could win the jackpot with every single ticket, ten times in a row. It's possible, it just happens that the associated probability to this event is ridiculously low.
1/
1/
It's also possible that SARSCoV2 could become more virulent than ebola and MERS, it could spread through the human population like wildfire, and send humanity back to the stone ages. It's possible, it just happens that the associated probability to this event is ridiculously low.
Essentially anything and everything is possible, though some events are more likely to happen and some are worse. This makes risk assessment difficult. In essence the question becomes how can we weight both the impact, and the likelihood of something bad happening. Tricky!
3/
3/
Predicting risk is indeed super tricky and as a consequence an understandable pressure on those who have to predict risk is to err on the side of caution, i.e. "better safe rather than sorry", even when they appreciate they may end up way off with their predictions.
4/
4/
This might shed some light on the possibly enigmatic - and arguably somewhat flippant - recent comment by Neil Ferguson that he was happy to be wrong as long as he was pessimistically wrong.
5/
5/
More importantly, the question is how to translate probability and magnitude of risk of a rare event (e.g. emergence of a viral 'variant')into policy. Interestingly the 'precautionary principle' that has often been invoked in that context is actually utterly useless.
6/
6/
To go back to what is the probability of SARSCoV2 mutating into a super lethal strain with 33% associated mortality, we can't really answer it precisely, though we can approach it in a Bayesian inference framework, which happens to be how our brain tends to work.
6/
6/
In short, we can set some priors (based on what we know or think we know), which we update with the data we have collected, and from there produce a posterior, i.e. an expression of our degree of belief (probability) that the event may happen.
7/
7/
One problem is that the choice of priors can be highly subjective. One person may think "no way this can happen", another "humanity is obviously doomed". Such strong prior won't be 'overruled' by the data, which happens to be extremely limited at this stage for SARSCoV2.
8/
8/
Alternatively, we may be willing to accept we have no clue, and set so called 'flat priors'. This may well be sensible, but given the limited data available, we may as well decide to go to the pub, rather than rack our brains.
9/
9/
There may be one way out. We know of ~200 other respiratory viruses in circulation that cause 'common colds'. Most are rhinoviruses, but four are endemic coronaviruses. Everyone essentially has been infected multiple times by them during their lifetime.
10/
10/
I can see no good reason personally why SARSCoV2 would be intrinsically more virulent than most of them, with the obvious difference that SARSCoV2 entered the human population only recently (late 2019), and no one had been exposed to it previously.
11/
11/
Back to my priors. We got ~200 respiratory viruses. They circulate year in, year out. To the best of my knowledge none of those viruses ever became 50x more virulent, but appreciate the data from >100 years ago may not be good enough for us to have recorded such an event.
12/
12/
So I set my prior for the emergence of a much more lethal viral lineage in any year as a distribution centred around ~0/20,000 (i.e. ~0/200 viruses x 100 years), which is a super-tiny number.
13/
13/
I would want to see very strong data before I were prepared to update my belief (posterior) that the emergence of a super-lethal SARSCoV2 lineage in any year is anything but negligibly small.
14/
14/
Loading suggestions...