Understanding Is a Moral Obligation

I inherited a codebase once where I could feel the previous person’s confusion. I do not mean that poetically. I mean it was in the structure. It was in the naming. It was in the layers that described what but never why. Every file was a small monument to someone who had shipped before understanding what they were shipping, and I spent weeks, actual calendar weeks, trying to understand what someone else could not be bothered to understand themselves. And in those weeks I could feel the thing starting to happen that I am most afraid of in this work, which is that I was becoming them. I was inheriting not the code but the confusion, the way a room inherits the posture of the person who sat in it last, and I was sitting in the room, and the posture was beginning to settle into me, and I could feel it.

That experience changed something in me I have been trying to articulate ever since, and I want to try to articulate it here, knowing I will say it badly, because it is the thing I believe most strongly and also the thing that makes me sound, when I say it aloud, the most insufferable.

Understanding is the work. Not typing. Not shipping. Not velocity. Not the small warm coin the body mints when a pull request turns green. Understanding. I know how that sounds. I know it sounds like a thing a person says to make themselves feel superior to the people who are shipping while the person saying it is not. I have been both people. I have been the one shipping without understanding, and I have been the one who stood at the edge of shipping and refused, and I have been the one who lectured, and I have been the one who was lectured at, and the only thing I can say honestly from inside all of that is that understanding is the work anyway. It does not care whether I am currently embodying it. It is the work whether I am doing it or not.

I want to build this slowly because I think there is a place people skip over and the skipping is where the whole thing falls apart.

We are surrounded by people who do not understand what they are doing. I do not mean stupid people. I do not mean lazy people. I mean people who learned, correctly, that sounding confident is safer than being honest. People who learned that shipping fast gets them promoted and understanding deeply gets them nothing. People who learned to produce output before they learned to see the problem, and who have now been producing output for so long that the not-seeing has become invisible to them, not because they are hiding it but because the muscle for seeing was never built in the first place and you cannot miss a muscle you never had. These people are everywhere. They are in engineering. They are in medicine. They are in journalism. They are in government. They are in my relationships. They are sometimes in me. They are, on a bad Tuesday, mostly me.

The reason understanding matters, the reason I have come to think of it not as a virtue but as an obligation, is because of who pays when you do not do it. And it is almost never you.

It is the engineer who inherits your confusion at three in the morning with the logs open. It is the on-call person reading your comments like a ransom note, trying to figure out what you meant. It is the doctor who uses the device you built when it fails in a way you never tested, because the failure mode required a kind of understanding you did not have time to build. It is the person in a relationship with you who trusted that you understood them, and who is slowly learning, in small undramatic ways, that you did not, that you never actually tried, that what you called knowing them was a shortcut you took because the actual knowing felt like it would take too long.

When you act without understanding, you are not taking a risk. You are transferring one. The transfer is silent. You never see where the cost lands. You do not hear the three-in-the-morning call because you are not the one answering it. You do not watch the device fail because you are not in the room. You do not feel the moment your silence wounds the person you love because the wounding happens inside them, in a place you have not been invited into, and you continue on with clean hands and a high opinion of yourself, and the person you love begins, quietly, to expect less from you, and the expecting less is the cost, and the cost is not yours, and that, more than anything else, is what I mean when I say understanding is a moral obligation. The morality is not in the understanding. The morality is in the transfer.

Understanding is slow. I have said this and I will say it again because I know you are reading fast, and I read fast too, and the problem we share is that we learned to read fast in a world that rewarded speed, and we carried the reading speed into domains where the speed is the enemy. Understanding feels, from the inside, like standing still. It feels like everyone else is shipping while you sit at the desk and stare at the problem like an idiot. It feels unproductive. It does not appear in metrics. There is no ticket for “sat with the problem until it revealed its shape.” No one gets promoted for comprehension. We have constructed performance review systems that reward action and that reward nothing for understanding, and then we are surprised when the systems we inherit are brittle, and we are surprised when the people we love feel unseen, and we should not be surprised, because we built the incentives that produced exactly this, and the incentives are still running, and they are still producing, and they will keep producing until something in us refuses the incentive and pays the cost of the refusing ourselves.

I want to say something about relationships, because I think the engineers reading this are going to try to make the essay only about code, and it is not only about code, it is about everything, and if I let the engineering stay the whole subject I will have done to this essay the thing I am arguing against.

Understanding in a relationship means letting the other person be a person with an interior you do not have access to. It means sitting with someone long enough for their anger to stop seeming confusing and start seeming coherent. It means watching how your decisions land in their body, not their mind, their body, the flinch, the tightening in the chest, the small change in the rhythm of breathing that you can see if you are paying attention and that you will miss if you are not. It means noticing the silence that follows a thing you said, and not filling it with more of your own voice, and letting the silence tell you what it is trying to tell you. It means being willing, sometimes, to be wrong about them, to find out that the thing you were sure of about them was wrong, and to let the being-wrong reorganize you rather than arguing it away.

When you do not do this, you transfer the risk. You transfer it to the people who love you. You transfer it to the people who depend on you. You transfer it to the people who built some portion of their life around the premise that you saw them, and who are slowly learning you did not, and who are not always going to tell you, because telling you would require a kind of confrontation they are not sure they have the energy for. I have been on both sides of this. I have been the one who did not understand and made someone else pay. I have been the one who was not understood and paid quietly for years. Neither side is the villain. There is no villain. There is only the cost, and the cost is real, and the cost is invisible, and the invisibility is part of what makes the whole thing so hard to see in time.

Now I want to say something about the machines, because the machines change this and also do not change it, and both are true.

The machines can help you understand faster. This is real. I want to start here because I do not want to be one of the people who makes the argument against understanding-without-AI and pretends the argument is about understanding when it is really a territorial defense of an older way of working. The machines can explain things you could not parse alone. They can reframe a problem so you see it from an angle your assumptions were blocking. They can show you the shape of something you could not see from inside it. This is a gift. I want to name it as a gift before I complicate it.

What the machines also do, and what I am watching happen to my friends and to myself, is that they make it easier than at any previous point in human history to not understand and not notice. Because the machine will produce an answer that looks right. It will compile. It will pass the tests you wrote. It will ship. And you will not know what you built. You will have produced a working system you cannot explain, cannot debug, cannot defend in a room full of skeptics, cannot maintain when the thing breaks in a way the tests did not anticipate. You will have become, in a small and a new way, a ghost, and the ghostliness is available now to people who would never have chosen it if they had known what they were choosing, but the choosing happened one small shortcut at a time, and by the time the accumulated shortcuts added up to a life they could not explain, it was too late for any single choice to be the one they would have made differently.

The machines will keep getting better. The barrier to producing something that functions is going to approach zero. And I think what happens next is the quiet sorting that I am already starting to see, where the people who use the machines to understand faster, to comprehend more deeply, to sit more precisely with problems the machines helped them see, are going to compound their advantage in a way that looks, from outside, almost unfair. And the people who use the machines to avoid understanding will fall behind in a way they will not be able to diagnose, because from the outside they will still be shipping, and the shipping will still look like progress, but the understanding will have thinned, and one day the understanding will be thin enough that they will be replaced by the tool they thought was helping them, because the tool can produce without understanding too, and it can do it cheaper, and cheaper wins when nobody is watching the understanding.

The thing that separates the two groups is the willingness to stop.

To stop before you build, and ask whether you understand this well enough to explain it to the person who will inherit it two years from now, when you are gone and your memory has faded and they are alone with the thing you made. To stop before you publish, and ask whether you understand this well enough to defend it in a room full of people who will pull it apart and find what is underneath. To stop before you ship, and ask whether you understand this well enough to debug it when everything breaks and everyone is looking at you and the logs are a wall of red. To stop, before you speak to someone you love, and ask whether you understand where they are coming from, whether you understand why they are reacting the way they are, whether you understand how the sentence you are about to say is going to land in a body that is not yours.

Almost always, the answer is no. That is not a failure. That is the beginning. The no is the door. The no is where the work starts, because it is only after the no that the sitting-with-the-problem becomes possible, and the sitting is the whole thing, and the sitting is what most of us refuse to do because the sitting looks, from outside, like nothing, and we have been trained to produce visible somethings, and a visible nothing feels, in the body, like a small death.

Understanding compounds. I keep returning to this because I think it is the part people miss. Each thing you genuinely understand makes the next thing easier to understand. Each time you sit with a problem until you actually see it, you are laying something down in your mind that will reduce the cost of every future problem you encounter that has a similar shape. It is the one investment that does not show up on any dashboard and pays dividends for years. We celebrate action because action is visible. We celebrate speed because speed is measurable. Understanding is invisible and slow and it is also, I have come to think, the only thing in this work that is actually worth anything in the long run, and the shortfall between what we celebrate and what is actually valuable is, I think, most of what has gone wrong in the field, and in our lives, and in the rooms where we are supposed to be loving each other well.

This is how it scales. One person, one decision, one moment of I will figure this out later, and the ripple spreads. Ten people, and you have a system where nobody understands anything, where everyone is just trying to keep it running, where the meaning has drained out without any single person noticing the draining, because the draining happens below the level of the metrics, and the metrics are the only thing anyone is watching. Eventually the thing breaks and no one can say why, and the asking-why becomes a political act rather than a technical one, and the people who end up being blamed are not the people who failed to understand, they are the people who happened to be in the room when the failure finally became legible, and the failure was distributed across everyone who preceded them, but distributed costs do not show up on dashboards either.

Understanding is a moral obligation because the alternative is quietly asking other people to pay for your ignorance, and they will pay, and they will not know your name, and they will never get to ask you why you did not do the work. They will pay in time and in frustration and in burnout. They will pay in relationships that fracture because you did not take the hours the understanding would have cost. They will pay in their own sense, slow and cumulative, that they are not being seen, and that the not-being-seen is normal, and that they should expect less, and the expecting less is what will break them, or break the thing between you, and by the time the break arrives it will look like it came from somewhere else, but it came from here, from this, from the small repeated refusal to stop long enough to actually know what you were doing.

Do the work.

I am trying to. I fail at it regularly. I have failed at it this week, on a codebase I wrote myself, because the understanding was inconvenient, and I took a shortcut, and the shortcut will cost someone, probably me, probably in about three months. I am writing this, in part, so that tomorrow I have to look at it. That is the only use of writing a thing like this down. It is a small promise to a future me who will be tempted, again, to move on without understanding, and who will, I hope, remember the sentence he wrote on a Wednesday in December and let the sentence slow his hands for long enough that the understanding can catch up.

Sit with the problem. Understand it. Then act.

The present will not reward you for this. The present rewards the shipping. The future might. The future is built out of the understandings we accumulated when nobody was watching, and the future belongs, mostly, to the people who did the understanding without the reward, because the people who needed the reward to do it eventually stopped doing it when the reward stopped arriving, and the rewardless understanding is the only kind that compounds.

That is the work. That has to be enough.