Breaking The Rules

Walking back from the track the other day, I came upon a red light, looked to see there were no cars coming and crossed the street. Growing up in NY, that’s just what you do. But the eight or so other people gathered at the crosswalk near the Universidade de Lisboa didn’t budge — they were still obediently waiting for the signal to turn green.

(My first thought was to write a dystopian story about a municipality that as an experiment increased the time for each light to turn green — both for cars and pedestrians — by a quarter second every week to test their obedience. Both parties waiting at ever longer lights even though no one was in the road.)

My second thought was when irrational rules like, “Wait at the light even if there are no cars coming” should be broken. Maybe not for an eight-year old, but as a seasoned pedestrian from NY, my judgment was usually* sound on that front.

*I have occasionally had to sprint a couple times and wondered what the hell I was thinking deciding to channel Usain Bolt just to save 20 seconds on my foot commute.

My third thought was whether this passes the Kantian Categorical Imperative, “Act only on that maxim you would will to be a universal law” test, i.e., should every adult pedestrian cross when there are no cars coming, and what would be the effect of that? And then, more broadly, should everyone simply set aside laws and rules generally they deem not to be especially efficient or practical in a given case? Is it really a problem, for example, if I embezzle a few grand from my company to qualify for a bigger mortgage if I know I’ll pay it back as soon as the bank gives me the green light?

The slope from crossing on red as I see fit to embezzling and returning the funds undetectably isn’t that slippery though. It’s not that the downside is worse than getting run over by a car, but it’s harder to predict. And even if I were to pay it back without incident, those entrusting me with access to the funds would still view it as a violation in a way few seem to mind if I simply cross the street on a red light.

So no, you should not break inconvenient rules solely because you’re pretty sure there won’t be consequences for doing so. But you should also not stand like a donkey on the curb waiting for the light to turn if you can see no cars coming for 300 meters in either direction, just because it’s a rule.

The obvious conclusion is you should sometimes break impractical and inconvenient rules, and you should sometimes adhere to them, depending on a variety of variables too complex to assemble into a clear formula here. It would be hard, if not impossible, to come up with a rule for when you should adhere to and when you should break rules because that rule, being a rule, would be self-referential, i.e., a rule can’t govern whether to break itself without falling into a logical infinite regress.

It’s just not feasible to come up with a purely rule-based system of appropriate actions exhaustively accounting for every situation in which you might find yourself. Deciding what to heed and what to ignore is something that requires feel.

It’s a fairly basic point, but in the era of artificial intelligence where some people have high hopes for machines that have better judgment than people, it’s one, in my opinion, worth making. Machines operate according to their programs — the rules that govern them. To the extent governing rules are insufficient, machines themselves are ill-suited to a task and should not be charged with it. People use the expression, “it’s not checkers, it’s chess,” but as machines (which are much better than people at chess) are given decision-making roles, we might modify it to, “It’s neither checkers nor chess, but real life.”

. . .

The limitations of rules-based programs to navigate complexity remind me of another shortcoming of machines, and that’s their inability to generate truly random numbers. If there’s a rule behind its creation — and computers need rules (programs) — then it’s not random. To create something random, you’d need a rule that defied rules, which was the same problem we had above.

The best way to come up with a random number, then, one that can’t be solved by figuring out the program that created it, is to incorporate nature into the process. Imagine a wind chime whose notes were recorded by a computer whenever it rang. That would be truly random, something no algorithm could crack.

(It might theoretically be possible to map out all the variables that cause the chimes to ring out exactly as they do, but not practically so because the math in complex systems is too daunting.)

We too are complex systems, co-existing in even more complex biological and social ones. We not only impoverish our understanding but are liable to make a mess of our lives (and those of others) should we try to model our thinking after machines.

That is, unless you’re in a game of checkers.