Make the machine write your one-star review first

The exercise is simple. Take the thing you are about to send out — the essay, the deck, the business plan, the CV — and hand it to a model. Then tell the model who to be.

Seth Godin sets out the instruction as a single line: ask for a detailed, specific negative review of the kind that might appear on Amazon, in ArtNews, at a PhD defence, from a disappointed venture capitalist, or from a recruiter.

The choice of critic is the whole trick. A generic request for feedback produces generic feedback. Naming the venue and the role forces the output into a shape you will actually recognise when it lands in your inbox.

Godin argues three things follow. Some of the one-star reviews turn out to be ones you are happy to have earned. His own example: a single star for a novel because the vocabulary is sophisticated.

Second, the criticism is easier to hear because nobody said it. He describes the output as a parody of a person rather than a person, which is what makes the useful parts of it readable.

Third, you learn that some critics cannot be satisfied. Criticism is what critics do, and seeing that on a screen makes the negative voice in your head look less authoritative than it did.

He also attaches a warning, which is the part a founder should probably underline. He advises against making it a habit, because the negativity gets stuck in your head. If that happens, his suggested remedy is to ask the same model for glowing reviews instead.

For anyone raising money, the applications are obvious enough. A model briefed as a disappointed investor will write the memo that kills your round, and it will write it for free, on a Tuesday, before you have spent a partner meeting finding out the hard way.

The same applies to hiring. Ask for the recruiter's rejection of your own CV, or of the job advert you are about to post, and you get the objection stated plainly rather than inferred from silence.

Godin also published the model's review of his own post, which called the idea a pseudo-profound productivity hack and accused it of building a machine for manufacturing comfortable rationalisations. That is the honest limit of the technique, printed by the person recommending it.

The counter-argument he then published is the more practical one. Most work never receives meaningful feedback at all — it gets silence, a thumbs up from a relative, or rejection letters that say nothing useful.

That is the case for the exercise, and it is a modest one. It does not replace a critic who has actually read your work. It fills the gap left by the critics who never showed up.

The cost of trying it is one upload and one sentence. The cost of not trying it is finding out what the disappointed investor thinks while sitting in the room with them.