Accountability requires a neck
Knowing what to do next was never a data problem. It won't be an AI problem either.
Action requires three things:
- Context — the tacit layer, the stuff that never makes it into a slide
- Motive — what compels you, and what you'll trade for it
- A neck — owning the consequences, warts and all
As an analyst I sometimes had the context. I borrowed motive secondhand, from whoever's OKRs were on fire. But my neck was never on the line. Every recommendation I shipped was free advice from someone with nothing at stake.
"Don't raise prices for short-term gains."
They raised prices.
That wasn't a communication problem. No amount of storytelling-with-data can overcome a missed quarter. AI can log context and motive. But who's making the call?
Context lives in fog, and that isn't an accident. The first post in this series asked whether AI would exacerbate organizational theater or eliminate the need for it. Neither. Nobody is coming to force accountability on anybody. Fog is free, fog is comfortable, and fog is literally air cover. Nobody writes down the call they made in the grey. Why show your hand?
But fog now has a price. When AI can execute on command, the command becomes the bottleneck. A company whose humans are decisive in the open pulls ahead; the one still adjudicating whose number is right stalls. Same market, two clock speeds. AI doesn't make companies honest. It makes withholding expensive.
With that, context stops evaporating. Engineering already runs this play — decision records, post-mortems, commit messages — because nobody wants to re-litigate a settled argument. Now at the decision level, written down as they happen, rulings become a replayable ledger, not a knowledge base. The next conflict doesn't inherit a doc it has to retrieve. It inherits the ruling.
Motive can be written down too, no longer ghost-authored through metric definitions (every definition is a goal wearing a number). Investors fixed this ages ago with the bet memo: thesis on paper, expected return attached, graded at exit. Write the bet down next to what you expect it to do, and it has a byline. The fog keeps thinning.
That leaves the neck. There's still no artifact for a neck — the only one it ever leaves is a chopped one. You can't punish a model, so accountability has nowhere to land. With context and motive on the record, everyone knows what to do next. So who owns the commitment? The job concentrates toward that last human step: setting expectations, grading the bet, owning what follows, conviction not decks. When the bet misses, there's no fog to retreat into. That's ok.
When the fog clears, all that's left is to write your name.