The R numbers coming out of UK epidemiology are garbage, Racaniello is right about that. They aren't physically grounded in any way, they're just arbitrary fudge factors brute forced to make their equations plot graphs that look like reported government stats so far. The same model will routinely calculate totally different values of R for different areas of the same country at the same time, without variants or anything like that.
Also the claims about variants being "more infectious" are - when you dig in - hopelessly confounded by seasonal effects that many epidemiological papers were at that time still ignoring.
The seasonal effects are trivially controlled for by comparing the growth rate of the variant to the growth rate of the wildtype at the same time. And that's exactly how the original analysis was done, because the people doing it were not incompetent.
For your seasonal effect explanation to make any sense, the effect would have to have only applied to Alpha but not to baseline Covid at the same time. And then it's not a seasonal effect, is it? It's an actual difference in the behavior of the two variants. The same goes for any other similar confounder that you try to manufacture.
Not the analysis I saw, where Delta's growth rate was compared to Alpha in the same time window, not the start of alpha's own growth period. Because yes, they are totally incompetent. You can look at graphs of the changing proportions to see visually that that's not much difference in how fast they took over.
Also the claims about variants being "more infectious" are - when you dig in - hopelessly confounded by seasonal effects that many epidemiological papers were at that time still ignoring.