My friend Duncan (
https://en.wikipedia.org/wiki/Duncan_Campbell_(journalist)) told me off for jumping on the Neil Ferguson hate bandwagon :
"Looking at code in a critical way is also good, my friend, but hang
on hard to your goolies and don't get swept over by internet
rage-raising that has nothing to do with science or public health.
Ferguson and his team have been doing the same jobs internationally
for at least 2 decades before Johnson and Cummings showed up to and
chose to write cohorts of our population and elders off to quick and
terrifying deaths, alone, and without medical help, in pursuit of
dumb old ideas and new fantasies.
He and Imperial will be around when Britain is finally shot of these people.
Because of the gravity of what they did in February and March, and
the lethal effects, they are desperate for any distraction, as Trump
as is. The attacks on Ferguson come from there, as did the spying on
his personal life and then the hate campaign against him in
tabloids. We have seen it all before, have we not?
Take a deep breath, and see if any codepicking nitpicks actually
affect the scientific outcomes. For example, how does Imperial's
predictions compare with the University College, Edinburgh or John
Hopkins models.
Have proper peer-checked papers been published that make criticisms
of the IC model?
Most important.. Have you gone and got the code, run it, checked the
issues and claims and how they would affect outcomes. Have you
read the other and related NERVTAG papers about it ?
I would have expected everyone's models to essentially be at least 10
years old, because they were all put together after SARS. Its not a
crisicism, at all. There's been no real huge pandemic since.
If there is a stochastic element set to diffferent runs, then of
course each run will be different, and drive a probabilistic
outcomes. All these points need to be checked, to be responsible.
If you'd like to dig further, there is definitely rigourous checking
needing to be done. Suggest you do all the checks above, and come back"