Short is not economical

The principle

The Economy score does not measure word count. It measures whether each word is doing a job a shorter or different word could do. A two-word joke can be uneconomical if one of those words is a hedge. A forty-word joke can be economical if every word is load-bearing. Length is a proxy, and like all proxies it fails you the moment you stop looking at the actual material.

There are three stances a writer can take toward a first draft, and they genuinely disagree. One treats rewriting as a mechanical skill: you mark up the page, cut the fat, repeat until nothing is padding. Another treats the first draft as the honest version and regards going back to sand it down as a betrayal; in its view, heavily revised work reads like a pinned specimen rather than a living thing, and the visible effort to smooth things out signals that the writer did not trust the first impulse. A third, more modest position: even the rawest material benefits from one pair of eyes that is not yours, because you cannot see your own qualifiers from inside them.

All three agree on the point that matters tonight. The question is never "is this short enough?" The question is "is this word doing something the word before it did not already do?" If you can delete a word and the joke still lands, that word was dead. If you can swap a vague word for a specific one and the punch hits harder, the vague word was costing you points.

The usual overshoot the app catches: a writer who has been told "be concise" treats conciseness as a length target. They cut a joke from sixty words to twenty-five and feel satisfied. But twenty-five words of hedging, qualifying, re-explaining, and vague gestures is still twenty-five words of dead air. The app does not see the number twenty-five. It sees that word fourteen is a qualifier the audience never asked for, and word nineteen re-states what word twelve already said.

So the skill is not "write shorter." The skill is this: for each word, ask what specific job it does. If the job is "make the audience comfortable before the punch," that is not a job. Cut it. If the job is "give the punch a concrete image to land on," keep it, but make sure the image is one the audience can see in a half-second, not a description they have to reconstruct in their head.

In a joke

Illustrative example, same topic in both versions: the train commute.

Weak (short, but every word is padding):

"So, you know how on the train everyone's just kind of staring at their phone? And it's, like, this weird group silence. Anyway, I get off at my stop and feel like I survived something."

Fixed:

"On the train, everyone stares at their phone like they're defusing a bomb. I get off at my stop and feel like I survived a hostage situation."

What changed: I removed the hedges ("so," "you know how," "just kind of," "like," "weird," "anyway") that padded the setup without adding information. I replaced the vague punch ("survived something") with a specific image ("hostage situation") that does the same emotional job in fewer syllables and gives the audience a picture to hold. The fixed version is roughly the same length as the weak one, but every word now carries weight instead of cushioning it.

Your turn

Take the joke the app just scored and put your finger on the first word. For each word, say out loud what specific job it does in one phrase; if you cannot name the job, cross it out. Continue until you hit the punch, then read what remains. If the punch still lands, you were uneconomical. If it does not, you cut a load-bearing word; put that word back and re-check the words around it.

In one line

Economy asks what each word does, not how many words there are.