I'm starting to realise that. I didn't come here for a row either. Straight up, the numerical analysis that's being presented on the testing data is complete garbage and considering this is what I do, thought I'd make a couple of comments. When I started to question what's being reported, kept my gob shut for a while at work as didn't want to appear like a crank, but when I finally got the courage to ask one of the senior researchers (I'm only 3 years into the job) if I was talking out my arse it turned they were on exactly the same page.
Here you go... The imperial college model that everything was based on at the beginning is absolute bollocks. Now at the time, they couldn't have known that, it was a new disease and I have zero doubt they were trying to get it right, but my boss got hold of the code they used and he couldn't believe it. Fifty thousands lines of non-modular code, all written as one single script. For non-coders, that means it is essentially impossible to check that code for bugs, and in 50k lines code there will be bugs in the maths. So fair enough, they didn't get it right, but the way they built the model means it should never have been considered in the first place.