Entropy April 2018

Entropy has lots of definitions, but the one I am referring to means the inevitable and steady deterioration of a system. Basically the passage of time tends to have a deteriorating effect on things. This includes common uses, like expiry dates on foods, and less common uses, like the effort necessary to sustain a relationship, a business or a set of cultural values. I’d argue that being aware of the impact of entropy is necessary in order to sustain almost anything you care about. Any sustained endeavor benefits from fighting entropy. So what is it?

One antidefinition of entropy would be as follows. I think the Nassim Taleb concept of antifragility is essentially the opposite of entropy. Almost everything, once formed, is constantly moving towards entropy. It takes constant effort to ensure that things move away from entropy and towards antifragility. That said, there are design decisions you can make when building a system, relationship, business or set of values that can make them more antifragile and less likely to entropy (meaning you can extend their shelf life).

One way to do this would be to consider what lasts. The test of time is an excellent test of the antifragility and rate of entropy that something experiences. When I got married it felt like a good sign that we had been friends for over a decade before we started dating. And I think the same applies to programming, where certain approaches, tools and patterns have existed since the beginning of computing and some are new. The ones that still exist despite being invented fifty years ago are likely to outlive the new ones.

I would argue that things which do not last as long have much less value. I would strongly prefer something to last my lifetime or longer to something that has enormous impact but fizzles out. Having had a household long enough for all the poor quality items I bought when I originally moved out to have been replaced at least once has given me an appreciation for quality and things that last.

The scary part about entropy is that there can never be such thing as passive investment. Income that is passive ends up being income that does not exist, sooner or later. Likewise relationships take constant investment to stay the same quality that they start with. Even startups seem to be like this, if you get distracted it has an immediate and lasting impact on the success of that project. You have to fight it constantly in order to avoid the effects. Ideally you have to outrun it while laying down protection from future entropy. If you can get better as a result of entropy, you’re in a great position.

So the main goal of an endeavor should not necessarily be to maximize the impact it has at any given time, but rather to maximize the length of time that it can remain antifragile in the face of entropy. Assuming things compound, the longer something lasts, the more beneficial, regardless of the rate of growth. And to consider how long the useful life should be before making decisions that could shorten it.

My dentist told me that teeth with cavities will have to be crowned within 30 years of the filling. Modern concrete condo buildings tend to have a useful life of between 50 and 60 years, and are unlikely to last even 100 years the way less effecient but more expensive steel buildings do. All this just goes to illustrate that things we assume are permanent are not and that entropy and antifragility should be carefully considered during the design phase to maximize the value of work being done. Work is not virtuous in and of itself, and a lot of work fails the Lindy test. If it doesn’t matter in a year, it almost certainly doesn’t matter at all.

Even survival itself is an exercise in fighting entropy (cell death) and maximizing antifragility (resilience in the face of a challenging and changing environment). People assume that all the modern technology we have was a rock that had to be pushed up a hill one time and then people would benefit from the technology forevermore. This is simply not the case. It takes constant and considerable effort to even maintain the technologies of today. Many of them could die or at least seriously regress (a third of people think Facebook itself is “the internet” and are not aware of the other tens of billions of interesting websites) without constantly fighting entropy. We used to have faster software that didn’t crash as much. But people failed to invest in maintenance, and now we have slow software that you have to pay every month for.

When making a design decision, think about what the decay rate of those design decisions might be. Literally everything decays on a physical level, if not a qualitative one, so this must be considered and weighed against all other forms of possible upside to determine whether something is worth pursuing. Further, almost anything new goes through a hype cycle. It is not enough to have early success in business: your first customers needed your new product more desperately. It is just as hard to grow and stay large, because you must constantly fight entropy. Worry more about the churn rate than the growth rate. Hire a tailor rather than shopping. Invest in the mid level products that have the quality but are missing the luxurious branding. Fight entropy.