May 7, 2020
To echo General Ripper in Dr Stranglove…”Although I hate to judge before all the facts are in, it’s beginning to look like Professor Ferguson exceeded his authority…”
For those of you who just returned to Earth, Ferguson, Professor of Epidemiology at Imperial College London, one of the leading science universities in the world (disclosure, I am an alumnus thereof) is famous for presenting a model of the corona virus pandemic which, apparently, scared the British government into its total lockdown mode by predicting in excess of half million deaths in the UK if this action was not taken.
Well, Ferguson has a new claim to fame now—he’s been breaking the lockdown rules he espoused. Now, Professor Pantsdown, the Bonking Boffin, has been visiting his mistress, another far-left ‘activist’ and a married woman with children and an obviously beta-male husband. He has resigned from the SAGE (STUPID, sh-h-urely) committee, until recently a secret tribe of lefty academics supposedly advising Boris Johnson, the UK Prime Minister.
Ferguson’s main claim to fame is his computer model used to predict the outcomes of the coronavirus pandemic. His model and software have been subjected to devastating reviews as may be seen here.
For starters, and this is bad enough, it is 15 000 lines of C++ code that has been built up over 15 years, and tweaked to death to get it to work. Anyone who has worked with C++ knows it’s a devil to deal with as new bugs tend to get introduced where least expected, especially with such a large program. Ferguson has refused to release the source code, which should, by itself, light up any dashboard. Apparently, some heavily modified version is available via GitHub, the software sharing site. Here’s what appears:
COVID-19 CovidSim Model
This is the COVID-19 CovidSim microsimulation model developed by the MRC Centre for Global Infectious Disease Analysis hosted at Imperial College, London.
CovidSim models the transmission dynamics and severity of COVID-19 infections throughout a spatially and socially structured population over time. It enables modelling of how intervention policies and healthcare provision affect the spread of COVID-19. With parameter changes, it can be used to model other respiratory viruses, such as influenza.
⚠️ This code is released with no support.
⚠️ This model is in active development and so parameter name and behaviours, and output file formats will change without notice.
⚠️ The model is stochastic. Multiple runs with different seeds should be undertaken to see average behaviour.
⚠️ As with any mathematical model, it is easy to misconfigure inputs and therefore get meaningless outputs. The Imperial College COVID-19 team only endorses outputs it has itself generated.
W-H-A-T! It is nigh on unbelievable that the British government is shutting down the world’s fifth largest economy based on this.
As the code reviewer says:
Non-deterministic outputs. Due to bugs, the code can produce very different results given identical inputs. They routinely act as if this is unimportant.
This problem makes the code unusable for scientific purposes, given that a key part of the scientific method is the ability to replicate results. Without replication, the findings might not be real at all – as the field of psychology has been finding out to its cost. Even if their original code was released, it’s apparent that the same numbers as in Report 9 might not come out of it.
Investigation reveals the truth: the code produces critically different results, even for identical starting seeds and parameters.
A team from the University of Edinburgh tried to use this code after making the data tables more efficient for faster loading, and discovered to their surprise—and, no doubt horror—that the resulting predictions were off by 80 000 deaths for the same parameter input!
The Imperial team’s response is that it doesn’t matter: they are “aware of some small non-determinisms”, but “this has historically been considered acceptable because of the general stochastic nature of the model”. Note the phrasing here: Imperial know their code has such bugs, but act as if it’s some inherent randomness of the universe, rather than a result of amateur coding. Apparently, in epidemiology, a difference of 80,000 deaths is “a small non-determinism”.
Frankly, words fail me. This software has never been tested and examined by people who really know about dealing with large populations, like insurance companies, who have to deal with the real world.
Further, Professor Ferguson’s previous escapades in the world of epidemiology leave much to be desired:
• In 2005, Neil Ferguson told the Guardian that up to 200 million people could die from bird flu. “Around 40 million people died in 1918 Spanish flu outbreak,” he explained. “There are six times more people on the planet now so you could scale it up to around 200 million people probably.” The final death toll from avian flu strain A/H5N1 was 440. (That’s 440 people, not 440 million.)
• In 2002, the same Professor Ferguson predicted that mad cow disease could kill up to 50,000 people. Thankfully, it ended up killing less than 200.
There’s much more to come out about this story. But, Imperial College has taken a huge hit to its reputation as a world leader. They should let Professor Ferguson know it.