Fraser Nelson may well "be thick", as he suggests in his own tweets. Although I suspect from his tweets that he is being disingenous and trying to get the answers he wants rather than the truth. Certainly, he is irresponsible interviewing via twitter where Prof Medley can only give answers to complicated questions in 280 characters... If he genuinely believed he had a scoop that SAGE were holding back or manipulating information, he would have set up a proper interview to ask those questions.
He (and everyone on this thread) would do well to read the SAGE introduction to epidemiological modelling to understand the limitations of modelling. We might get fewer posts about SAGE getting it wrong predicting 100k cases a day in the summer if people actually understood more about modelling.
Modelling is not an attempt to predict the future, it is a description of what will happen if X, Y and Z assumptions are true. In reality, human behaviour (which has an effect on transmission and outcomes), is very difficult to predict, which will have a impact on the model's accuracy. I think we can all predict that people will not behave the same way in a lockdown now as they did in the first lockdown but how can anyone know to what extent people will adhere to the rules?
Modelling can never exactly replicate reality and therefore no individual model will give a perfect description of the future.
Sometimes we ask modellers to produce ‘reasonable worst-case scenarios’. This is where the parameters of the models are chosen to generate a challenging, but plausible range of possibilities describing what might happen in a highly pessimistic scenario. For example, this could be a scenario of no further policy changes, with some assumptions made more pessimistic, such as the emergence of a hypothetical variant of a virus that evades immunity. These sorts of scenarios are designed to inform government planning for more extreme situations.
www.gov.uk/government/publications/introduction-to-epidemiological-modelling/introduction-to-epidemiological-modelling-october-2021
As for omicron being less virulent than delta, there is very little evidence of that. There is some - it appears to be less effective at infecting and replicating in lung tissue. That is not enough evidence to assume that it is less virulent and decide policy based on that. If we are wrong, by the time we know that, it will likely be too late to take action. As Prof Medley says, modelling that scenario does not inform anything. "Decision makers don't have to decide if nothing happens."
Furthermore, virulence is only one factor that determines severity of disease. Immunity levels (you need to take into account that different vaccines seem to provide different levels of immunity, as does immunity from infection, and immunity wanes over time), genetics, comorbidities (people having other diseases), age etc. You can't compare the UK to South Africa because our popuations are not the same.
FWIW, analysis of the data in the UK (Imperial) suggests that severity of disease is the same for delta and omicron here so far. That comes with a huge caveat that we have very little data so far so may be inaccurate.
It's also likely that transmission may well be higher in the UK. There are many factors that will affect transmission that are not determine by the virus's inherent transmissibility eg it is summer in SA (people are inside less, the virus survives longer outside the body in humid, cold conditions), restrictions, mask wearing (compulsory in public places in SA).