This is one of those posts that show how my mind works.
When I learned about process control systems, it was very "ivory tower", with some very tough math (the only time I ever really understood Laplace transformations). Decades later, I think I can explain six months of engineering coursework in a few paragraphs and make it easy enough to understand and perhaps even useful.
There are three basic ways to contol any situation, given unfortunate math names: proportional, integral and differential. Each has its good points and bad points. I can demonstrate them with the example of car theft.
A proportionate response to an increase in car thefts is to increase the effort to recover stolen cars. The plan is to bring things as close to normal as possible. The main asset of this method is it's easy to understand. The downside is that things never get returned completely to their original state; even if a car is found quickly and has not been damaged, there is still wear and tear on the vehicle and the owner was deprived of its use for some time. The best proportionate response would be to have insurance that gives one a loaner vehicle until the stolen one is recovered and for the license numbers of stolen cars to be available to anyone, so they could be found in places like parking ramps.
Integral response is designed to return things to their original state, which proportional control can't do and that is its major benefit. The car theft example would be to have the thief pay for the car owner's losses. The problem with this method is "clamp-down." The return to the original state may be considered too slow, so more of the control is used and this leads to ever-increasing control. Here, the thief would be required to pay not just the losses to the owner of the car, but also the court costs... then the costs of the police used to catch him, then would put the thief in a workhouse to pay these expenses, then would require the thief to pay the costs of imprisonment. Essentially, the punishment keeps increasing until the system fails.
Differential response has the benefit of being fast and so often avoids clamp-down effects. Differential responses are usually anticipatory; for example, keeping a thief in jail prevents car thefts by that thief. The classic example of anticipatory response and the problems it can create are car alarms. When car thefts increase, people tend to buy car alarms (and some insurance companies subsidize this response), whether or not their car is likely to be stolen. One thing that happens is the creation of cycles of thefts: thefts decrease when a new alarm comes out, then increase when thieves learn how to circumvent it; similarly, when thefts are localized, police focus on that area and the thieves move - when the police move to the new area, they move back.
Differential control also is subject to "noise," that is, responding to something other than it should. Car alarms are notorious for going off when the car is not being stolen and people pay the price of having the peace disturbed because someone else is afraid of having their car stolen.
The biggest problem of differential control is that it can actually make the situation worse. If people rely on their car alarms, they may stop locking their doors, making the car easier to steal. If people ignore car alarms because they go off all the time, then the unlocked car is an easy target. If you imprison car thieves, they trade information and become better thieves.
Sometimes differential response combines all of these flaws and creates an unstable situation. The more one tries to stop the car thefts, the more the thefts would occur - fortunately, I don't know of any case where this has happened.
When you know these different approaches, you start seeing them everywhere and when you see the typical flaws, you look to see the method of control that created it. This leads me to the Minnesota state budget, which currently has a record deficit. During the Ventura administration, there was a record surplus (We tend to forget this. The money was returned to the taxpayers, which was politically popular, but the worst possible plan for control), and there was a then record deficit under Carlson before that. This looks like cycling out of control, which is the result of a differential control not working.
Is the budget under this type of control and is it going to get worse? The budget is anticipatory, so it's probable (unlike personal budgets, government budgets don't need to be anticipatory; any politician you hear saying, "If I ran my business like the government, I'd be bankrupt" does not understand finances). The common methods used for balancing the budget are accounting shifts and one-time fixes, both of which are differential responses. The only way out that I can see is to stop using projections of revenue and expenses to predict future needs and to switch to dealing with the situation as it is at the moment.
There is a local privately-owned golf course that shows how this is done. Every year, the members agree that they'll pay their share of whatever the expenses are and at the end of the year, they're given a bill for that amount. New members are usually surprised at first at how expensive a golf course is (the last I heard, 20 years ago, membership was $200000 a year!) and suggest ways of cutting expenses the next year; they decide what they consider to be luxuries they don't need. Interestingly, there's a waiting list to become a member, despite the cost.
Wouldn't it be wonderful if the same worked in government, where people wanted to move to Minnesota, where the taxes are high?
5 days ago