Wednesday, May 27, 2020

Models That Have Been Tried and Found... Trying

The earliest observers of the sky realized immediately that the motions of those faraway objects were predictable. It's now a pop-culture assumption that any enigmatic ancient construction was probably sited and aligned in accordance with the rising and setting of various significant lights in the sky.

These days, astronomers are aware of millions of objects flitting about in our celestial neighborhood, many of which, were they to collide with the surface or -- at sufficient speed -- atmosphere of our own world, could snuff us all out in very short order. As new asteroids are detected, they are observed and catalogued, and these observations are modeled on computers to see whether there is any danger of a new Chicxulub-style extinction-level event.

Modeling asteroids involves amassing data about each object's mass and motion, based on known values developed by mathematicians over several centuries. The more individual bodies are added to the model, the more complex it becomes -- but even now the job is profoundly simple compared to some other disciplines that have attempted to use computer models.

When you are dealing with, for example, living things such as viruses that are prone to mutation in extremely short timespans, your modeling assumptions have a half-life that can fall well within the decision loop of any agency or government. Using such a model to attempt to prescribe policy is irresponsible at best -- and potentially criminal at worst.

Yet that is precisely what happened with the Wuhan coronavirus pandemic of 2020. A complete post-mortem of policy responses and their consequences may well find, and many expect them to find, that the model-driven response cost more lives than had the response followed more established, time-tested patterns.

Epidemiology is only the most dynamically complex discipline to have been subjected to flawed computer modeling. Decades of flawed economic policy have been driven by computer models. Climate models have been used as a bludgeon to attempt to impose extreme changes on world civilization in the name of urgency.

Someday astronomers may announce that the Earth is at risk from a possible asteroid impact. They will defend this claim based on their computer models of celestial motion, and given the challenges and timeframes involved they will argue that some type of response is urgently demanded to fend off a global apocalypse.

Ideally, that response would be forthcoming, and it would be successful.

But not if the political overreach of economic, climate, and epidemiological modelers lead people to disregard all such efforts as garbage.

You can't expect to model every system accurately in a computer when you don't even know what you don't know about that system. Economists acknowledge that in any group of seven of them you will get at least eight opinions. Climate and epidemiology are vastly more complex, and there is no sign that the modelers in those fields are anywhere near as humble as their astronomer and economist counterparts.

Don't blow this for humanity, you over-educated idiots. If your arrogant stupidity prevents us from averting another Chicxulub, it will be your fault.

No comments:

Post a Comment