Posts

Sorted by New

Wiki Contributions

Comments

Very well presented. However, there is another aspect that most people seem to be missing. I come from an area of physics and engineering where I have dealt with control systems in the face of severe lag. Long time delay between changing a control parameter and seeing its effect.

We have a situation in the current pandemic where there seems to be a roughly 2 week period from the time a person becomes infected (and contagious) and to when it finally presents observable symptoms. Everyone in the country is pushing for reopening the bars and restaurants, and they are understandably impatient. But when the pandemic first appears to be tapering off is precisely the wrong time to reopen everything. You need to give the incubation period a chance to pass. Yet, repeatedly, people follow their impulses and governors respond to business pressure.

When a system has severe phase lag, as in this case, humans are notably incapable of dealing properly with them. We tend to see oscillatory behavior develop in the magnitude of system reaction - infections in this case. That is a signature of severe phase lag, and the simplistic control response based on immediately perceived threat levels. What we need is some feed-forward, or derivative (based on rates of change) injected into the control system to help stabilize the response.

(A bit of integral should be called for too... don't stop with control until the integrated or accumulated infections have declined to acceptable levels. You need all three terms - integral, proportional, and derivative, in this case. And all we have seen is a control response based on proportional impressions. This is notoriously prone to oscillation, just as we have seen.)