There are two important aphorisms to remember when consulting ‘experts’ ; the first is Warren Buffet’s comment that “You should never ask a barber if you need a haircut” , or in a similar vein, ask a General if you need to go to war or if he needs more money for weapons. The second is the old expression from the Soviet Union, “to a man with a hammer, everything looks like a nail.”
Thus to ask Bill Gates, who has spent the last decade promoting the cause of vaccines, if the solution to Covid-19 is a vaccine rather than a ‘cure’ is to ask the man with the hammer if Covid looks like a nail. Equally, the ‘barbers’ at Sanofi and GSK as the two biggest makers of vaccines are likely to agree with him. This however is different to the ‘experts’ who underplay the RIRO impact on most forecasting models, playing on the natural desire for certainty in what is, to use the title of Mervyn King and John Kay’s excellent new book, an era of ‘radical uncertainty.’
RIRO, or Rubbish In = Rubbish Out is a variation on another adage that employing a faster computer will simply get you the same wrong answer more quickly if your inputs are wrong. Most common is to observe correlation between historic variables and infer causation and then introduce more and more ‘dummy variables’ to explain why the model didn’t work as a predictive tool. These ‘new’ variables, be they global warming that failed to appear as predicted because actually it is being hidden in the oceans’ or the idea that the failure of Covid deaths to hit the initially huge projections is due to the underestimated impact of a variable such as social distancing. This latter example, eagerly embraced by those wanting to justify the lockdown, continues to lead to exaggerated ‘forecasts’ for countries such as Sweden.
Amongst all this, it is sometimes useful to follow William of Ockham and his famous razor and look at a really simple model of the behaviour of an exponential variable and the ultimate exponential saturation. The following is based on this article by analyst and blogger Dimitri Orlov. As noted above, he rightly points out that while models can be created with enough parmaters to explain everything, in doing so they lose any prediction functionality.
http://cluborlov.blogspot.com/2020/04/a-simple-model-of-coronavirus-pandemic.html#more
Taking the Occam’s razor approach of minimal variables and a three component model of mid point, maximum and growth and based on a simple mathematical model known as a logistic function, we can calculate how natural occurring phenomena follow an exponential growth function before reaching exponential saturation – i.e there are fewer and fewer hosts to occupy. Orlov follows this approach for a model of Global Covid (with some modest adjustments for China) and he is within + or minus 2% of the actual data being reported, which in our view makes it worth some consideration.. Moreover, rather awkwardly for some, he points out that while different countries have been taking different measures in terms of social restrictions, the virus is spreading in exactly the same way as it would have done anyway, i.e following the logistic function. In other words it makes no difference! Like much of his writing, it could be a bit awkward for politicians. It is well worth a read.
For what it is worth, here is how his model predicts the pandemic peaking and ending. It looks like we have already passed the peak and will be done by the end of May, but remember, according to Orlov, the outcome has nothing to do with the official response, the virus simply did what viruses do.
Already Past the Peak(?)
