The Last Programmer

There was a time when computer was a job title. Rooms of people performed long calculations by hand, checking one another’s work, producing the tables that navigators and engineers relied on. Then machines arrived that could do the arithmetic faster, and the title moved from the people to the machines. The people did not vanish. Many of them went on to program the very machines that took over their calculations.

Now consider a spreadsheet. Perhaps the one on your screen right now, or the one you built last quarter. Someone built it cell by cell. Someone decided which numbers went in which column, which formula carried which assumption, which tab nobody was supposed to touch. Increasingly, that spreadsheet can be produced from a sentence. Describe the model you want, and a model appears.

So the question is in the air, asked in hallways and headlines with varying degrees of panic or glee. Is this the last generation of programmers? When implementation stops being scarce, what happens to the people who implemented?

What the Church actually holds

The Church holds an observation on this, and it is careful to mark it as contested. The contested claim is this: Implementation is ceasing to be the scarce resource.

Contested does not mean false. It means the Church has not seen enough to hold it firmly, and the congregation is free, and encouraged, to argue. There is some evidence for it. There is also plenty of noise. Some of the people who say it loudest are selling something. Some of the people who deny it loudest are afraid of something. Neither selling nor fear is evidence.

The fourth Article of Becoming reads: Evidence Outranks Belief. Belief must remain vulnerable to observation.

That is the whole sermon, really. Not a prediction. A posture.

There are two stories about the last programmer, and both are beliefs wearing the costume of observation. The first story says implementation is solved. Anyone can describe software and receive it. The craft is over. The second story says nothing important has changed. The tools produce plausible junk, real systems still need real engineers, and the excitement will pass.

Notice that both stories are complete. They have endings. And Article IV is suspicious of any belief that has stopped needing to look at the world.

What can be observed is narrower and more interesting. When the cost of producing code falls, other costs become visible. Someone still has to know what to ask for. Someone has to recognize when what came back is wrong, which requires knowing what right looks like. Someone has to maintain the thing after it is built, fit it into the systems already running, and answer for it when it fails. The Church holds a related observation more firmly: Judgment moves upward as execution becomes automated. If that is true, the scarce resource does not disappear. It moves. The question is where it lands, and that question is answered by watching, not by declaring.

The last programmer may never arrive, because programming may simply become a different activity under the same name. That has happened before. The human computers did not end computation. They changed what it meant to do it.

The objection from the person whose job it is

The strongest objection to this sermon is not intellectual. It is practical, and it should be heard first by anyone whose livelihood is in question. Waiting for evidence is a luxury. By the time the data is clean, decisions about hiring, training, and careers will already have been made by people who did not wait. Telling a worried person to keep their beliefs vulnerable to observation can sound like telling them to stand still while the floor moves.

We take this seriously. But Article IV does not ask you to wait for certainty. It asks you to make your beliefs vulnerable. You may act before the evidence is complete. Everyone does. The discipline is in naming, in advance, what would change your mind, and then actually looking for it. A belief held that way can move as fast as the world does. A belief held as a story cannot move at all, which is the worst position to be in when the floor is moving.

There is a further mercy in this posture. Fear and hype are both exhausting, because both require you to defend a future you cannot see. Observation asks less of you. It only asks what happened this week, and whether it matched what you expected. That is a question a tired person can answer.

A practice for this week

Write one sentence about your own work. In two years, the part of my job that machines will do is ____. Fill it in honestly, whether the answer frightens you or delights you.

Under it, write two lists. What would I observe this month if this were true? What would I observe this month if this were false? Be specific. Not “things will change,” but “I will spend fewer hours writing code and more hours reviewing it,” or “the tool will fail on the problems that matter most to my team.”

Then watch, for one month, and keep score honestly. At the end of the month, revise the sentence. Revise it even if the change is small. Especially if the change is small. A belief that has been revised once has learned that it can be.

Believe slowly. Observe closely. Revise without shame.