In an interview on technical blogging, the most cited piece of advice is a phrase that made the rounds: lower your standards. Publish while you’re still actively unsatisfied with what you’ve written, because the alternative is a folder full of drafts and no published articles. The flaws you see are invisible to your readers.

John Gruber picks up the advice and extends it with a metaphor that illuminates the method: think of blogging as playing live, not recording a studio album. An album comes out every now and then, but if you tried to make every post a masterpiece you’d publish nothing. Goal: professionalism, not perfection. You’re in front of an audience, not in a garage rehearsing.

For those reading rug.gal, this pattern is familiar. The site came from the same idea: AI serves to empower those who produce, and production matters more than immaculate polish. Anyone using AI to write, analyze, or summarize hits the same block as someone writing by hand: the temptation to tinker endlessly. The method solves it with an operational rule, not an abstract principle. Publish before you’re ready.

The useful angle for those amplifying themselves with AI is direct: if you delegate writing to an agent or assistant, the draft you get doesn’t improve with time spent staring at it. Value comes from the correction loop, from verification, from closing and sending. The same principle behind the critical review pass playbook: one structured critique pass, then you publish.

In detail

Cynthia Dunlop’s interview dates back to January, but it was picked up and reshared in early August. The questions span the full arc: why he started writing, what still surprises him, which post he’s proudest of, which was hardest, and what advice he’d give to someone starting out.

The most cited piece is the answer on how to begin: lower your standards. Publish while you’re still unsatisfied. No one will ever know how perfect the piece in your head could have been. It’s advice that sounds counterintuitive in a world where every incentivized platform rewards polish, but the motivation is concrete: the real alternative isn’t a better post, it’s no post at all.

Gruber adds the frame of the live musician. You’re not playing in a garage randomly, you’re professional and trying to nail every note. But the right mindset is that of a concert, where you move from one song to the next without stopping to stare at an imperfect transition. If you tried to make every post a hall of fame entry, you’d release nothing. The concert metaphor has a practical corollary: the audience reacts in real time, and you adjust the next piece based on how they responded to this one. The rule applies equally to those writing by hand and those generating drafts with a model: endless tinkering is the enemy of production.

For those working with AI there’s a second level. Assistants and agents drastically reduce the cost of the first draft, but the cost of endless revision stays the same, sometimes gets worse, because with a model you can always regenerate a variant with another prompt. Time spent polishing a generated sentence is time not spent verifying facts, testing prompts on different inputs, or building the next piece. The rule shifts the center of gravity: the draft exists to be corrected and sent, not to be worshipped.

There’s a third level worth naming. Those who publish regularly build something no perfectly polished draft can give: a trajectory. Every piece is a point on a graph, and the graph says where you’re going. A single post refined to the millimeter is an isolated point, and an isolated point tells nothing to readers, to those wanting to collaborate, to those seeking expertise in your field. The value of continuous blogging lies in seriality, not in the single piece.

The limits of this approach are evident in regulated contexts. Those writing in domains with accuracy requirements (legal, medical, financial) can’t lower standards on verification. But they can lower them on form: produce the draft quickly, verify facts with method, publish. The principle stays, the application adapts.