The thermodynamics of user laziness

how adoption really works

One of the things I have said and believe, with love, is that users (that is, all of us) are lazy. That’s a good thing! As makers of all stripes, we love what we make, and we think other people will. But the reality is: we are all busy, and don’t have time for something that makes our lives better. We’re “lazy”. This isn’t just in software - any product that involves a user choice has to reckon with this.

But what does that mean, precisely? What’s “better”? One thing that’s been commonly said is something either has to be totally new (that is, solve a job that has no solution at all) or be “10x better”. Why?

I look to thermodynamics for this, specifically the second law (stick with me!) that says that entropy always wins. In a closed system, entropy can only go up - things become more disordered. Ok, but why, then, do we have order at all? If a cup of hot coffee can only cool down (in a room temperature room), how is it hot ever? Why doesn’t that violate the law of thermodynamics?

Because of that definition of “closed system”. If you look far enough out - say, to include the solar system, entropy in that system as a whole is increasing. The sun is increasing entropy massively as it burns fuel, which sends some heat here, which ultimately winds up heating that cup of coffee up. All along the way, bits of entropy are increasing to deliver that last small payload of order.

This is how it is with user laziness. Often companies will design a new product that’s a little bit better, and wonder why it doesn’t get adoption. Users are lazy! This is better! Why aren’t they using it more?

The answer is that closed system again: users are lazy in the totality of their lives. If the effort to move to a new habit plus the savings of the new habit isn’t a net win in terms of overall effort, they won’t do it. A little bit better plus a hassle to learn something new and remember to change defaults ends up being more work, and never happens. It’s only when there is so much novel value that the effort of adoption is swamped by that value (10x better or a solution where none existed before) that users will move.

This seems obvious but it happens over and over again. Throw a rock and you can see this in almost all of the work being done with generative AI right now. There are some genuinely good and transformative things being done, and some of those will be successful. But there are also many, many, incremental improvements being made in the hope of changing user behavior, which is very unlikely to work, even within an app, much less across it (within the app this is known as engagement, usually). Even really disruptive things like image generation won’t work if the additional work asked of the user to get to a usable result still nets out to a negative return on time.

It’s hard to do because we love what we build, but it’s always important for a maker to take in the “thermodynamic totality” of the user, and ask in an honest way, whether the solution is really, truly easier for them overall. If you can’t honestly answer that, no amount of wishful thinking or good intentions or funnel packing will move them.