I'd guess that the UK probably ends up being pretty much the same as Italy, Spain, France etc in terms of death rates.
In other words, it blew the natural advantage of being (mainly) an island and so having an easier border to control. Part of the reason it did this, I'm pretty sure, is because all the policy was so much influenced by IC dusting off its flu models which made some key errors for addressing this virus:
- Closing borders doesn't work. Nope; viz Oz and NZ, with death rates <1% per capita vs the UK, and where border closing was probably the single most important policy move. Apart from screwing things up for the UK, this IC modelling error taken from its flu modelling also played a big part in the WHO's early advice against border closing, and similarly for the US. I note that the UK is still waffling about border closings, which is mind-boggling from my Oz point of view.
- Case-based contact tracing & isolation isn't important. Nope; again viz Oz and NZ where early on this advice was rejected because front-line public health experts dominated the decision making, rather than modellers, and where intensive every-case tracing has been immensely important in stopping the thing close to the border. I'd also note that contact tracing appears to have been a huge reason why Austria, Germany and other Euro countries, lacking the natural island advantages of the UK, Oz, NZ etc have nevertheless kept death rates to a fraction of the UK's.
- Isolation doesn't work well; at least 80% of cases will not comply. Nope; completely invalidated so far by experience everywhere. People have been amazingly compliant with self-isolation recs and orders.
- Locking things down will have a really big impact. Well, maybe not so big. Oz generally didn't lock down as much as NZ and has better outcomes by some metrics; look at the grindingly slow improvements in Italy etc from lock-downs, which the UK seems to be experiencing also; and generally it has been a vastly weaker tool than border closings+tracing/isolation.
I haven't heard IC or Neil Ferguson say anything about maybe having been wrong, and as far as I can see they don't show anything about the performance of their model over time, just the latest cut where of course everything gets re-fitted to the history:
https://mrc-ide.github.io/covid19estimates/#/ Not very classy.