I posted a graph to show how ons numbers appeared to be too low when 20 to 29 year olds were high, but I don't remember seeing a zoe graph
The graph you showed, I never said it was a Zoe graph?
Zoe is often a good indicator and I would imagine is far better an indicator than current case numbers
Sorry, I just cannot see any possible way that a narrative that "Zoe is accurate 'cos random people are faithfully recording their information in Zoe". works against the same narrative "Cases are inaccurate 'cos random people are not doing tests"
Both systems rely on people opting in to testing, and the Zoe one is considerably more onerous on the individuals than the testing one.
I have no reason to believe the models for an R of 0.6 to be right, but equally no reason at all to believe Zoe either, all of them are extremely sensitive to their models of behaviour and sampling, given all of behaviour, sampling, testing and the virus itself has changed completely in the last six weeks, it's pretty unlikely that any are any good.