The great achievement of the Scottish enlightenment was the development of science. Science is based upon open and transparent claims, disputes, and data. Science has been fundamental to our material progress – and science will be key to responding and overcoming the current coronavirus challenge. But science can only work if the claims of scientists are open to public discussion and criticism. This is why journals publish scientific papers, and why peer review exists – to review for possible errors.
Of course, science doesn’t progress upon “experts” making pronouncements. Science makes progress based upon debate and discussion and logic and analysis. The whole point of the scientific enlightenment was that it rebelled against former calls to authority (“the Church said so”) and in favour of public debate and discussion of important issues.
The models used to predict the spread of the coronavirus in the Isle of Man should be published – so that they can be reviewed and analysed by outsiders in the community. Not only will feedback from outsiders help to improve the models (and fix possible errors) but transparency in public scrutiny will also give the public greater confidence in the accuracy of the underlying analysis. Our scientific method recognises that there is no one expert that knows everything – but rather, by sharing information openly and transparently, we can learn things as a community. If there is an error in this blog, other people can identify the error, point it out, and fix it. This iteration process can’t happen to documents produced in secret.
Transparency is especially important when the scientific predictions are used to impose significant restrictions on our community. We need to ensure that any such rules are based on good science – because if there is an error, it should be fixed. If the science is accurate, public scrutiny will confirm it, and increase compliance with the rules imposed upon our community.
One area of particular concern to me is that in yesterday’s Manx Government media briefing, the Government said its modelling was based upon “exponential” spread of the disease (see around the 7:35 mark of the briefing). Firstly, it is very possible that the Government misspoke – exponential is a common and simple word, and it is possible that they simply used the wrong word. If that’s the case, then publishing the modelling will clear up that error.
Secondly, exponential growth might be the right basis for the spread of the coronavirus disease – but I am very, very, skeptical of this. Exponential models are based on, literally, exponential growth. That makes sense in some circumstances (eg, compound interest) but it doesn’t make sense in all areas. Exponential growth makes sense if you are projecting growth in an unlimited domain (eg, there’s no maximum cap to your bank balance from compound interest) but it does not make sense in a more limited domain.
Exponential growth can also be very misleading when applying it the area of human affairs. Humans are not like gas molecules bouncing randomly around a chamber – they form networks which we normally call families, friends and communities. Coronavirus won’t jump randomly from one person on the Isle of Man to another person on the Isle of Man – it will jump from people to people who have contact with each other. It will spread through human networks. This is why it is important to keep distant from others.
Rather than use exponential models to chart the connections in a network, there is another, separate, branch of mathematics – graph theory. One key problem with using exponential models is that it will predict the disease to spread much further and broadly than graph theory would produce – because graph theory recognises that not every infected human is connected to every other uninfected human. Graph theory recognises that humans are not random molecules of gas bouncing around in a chamber – graph theory recognises that humans are connected to other humans through networks such as families, households, accommodation and so on.
So, if the Government has mistakenly used an exponential model, rather than a model based upon networks, there are two key problems with the predictions:
a) You come up with projections which forecast very high spreads of the disease
b) You would fail to properly account for the disease to have an outbreak in various clusters.
Graph theory fixes both of these problems – it recognises that people can only spread the disease to people they have contact with (directly like living in the same house or indirectly like through fomites). Further, it recognises that when the disease gets into a certain community, it will rip through that community (whether it be a family home or a larger shared accommodation facility) very dangerously.
In addition, graph theory and exponential growth can deliver very similar results in the initial spread of something: When you are starting from a very small base, network growth and exponential growth can appear very similar. But the differences will diverge over time, as networks become either fully infected (hence, the disease cannot spread further) while exponential growth does not account for that properly.
Of course, the Government’s modelling might have accounted for the network effects of this disease. In which case, it should be published to give the public greater confidence in its predictions.