The content doesn't have to be new

The first thing I want to post from an idea is usually the worst version of it, and I've stopped treating that as a problem. A video AI made with no words, no music, no script — just the raw thing. Post it. It's fine that it's bad. It's a rep, and the rep is the point.

The idea is not the scarce part

Somewhere I picked up the belief that each post has to be a new idea, and that reusing one is cheating. That's backwards. I have a handful of things I actually believe about building and shipping, and I could say each of them a hundred different ways before I ran out. The scarce resource was never ideas. It's the number of times I'm willing to say the same one out loud without deciding it's been "covered."

So I post the same idea over and over, formatted slightly differently, and I let it start shitty. The version that lands isn't the one I planned. It's the one I found by shipping the ordinary versions first.

The reps are on the process, not the post

What actually improves over time isn't the ideas — it's the process for turning one into a post. How fast I can go from a thought to something published. Which formats are cheap to produce. How ugly I'm willing to let the first one be. Where the see-more cut lands. Whether the link goes in the body or the first comment.

None of that is about having something new to say. All of it compounds. A month of refining the pipeline is worth more than a month of chasing fresh ideas, because the pipeline is what every future idea has to pass through. That's the whole reason the experiments next door exist — the hooks and cadences are the thing under test, the underlying idea usually isn't.

AI made re-saying the same thing nearly free

The reason "post it a hundred ways" went from a slogan to something I actually do is that reformatting used to be the expensive part. Rewriting one concept as a carousel, a talking-head script, a silent captioned video, a two-line text post — that was hours. Now it's a prompt. I can take one concept and present it a dozen ways in the time it used to take to make one.

And once making the variants is cheap, a second thing falls out of it for free: you can measure them. Same idea, five shapes, into the same feed — now the format is the only thing moving, and the numbers tell you which shape travels. You don't get that read unless the variants are cheap enough to actually make. That's the part I'm still bad at reading honestly; the data on my YouTube clips has been sitting there unpulled for months, and I have the same problem with Maps photos. Cheap to produce doesn't mean I've looked.

The arsenal is what unlocks the automation

Here's where it goes from a mindset to a machine. Once you've built up a sufficient stock of content — variants, reformats, the ugly ones and the polished ones — there's no reason you shouldn't have automated posts going out twice a day. The arsenal is the prerequisite. You can't automate a cadence you can't feed, and the fear of running dry is exactly the fear that reusing ideas dissolves.

Most of what goes out that way will do nothing. Sixty impressions, a hundred and twenty, forty-five. That's not the failure case, it's the mechanism — the volume underneath is what makes the occasional outlier possible, and you don't get to pick which post is the outlier in advance. Designing every post as a clean, precious, brand-new idea is a great way to publish far less, which is the one outcome I'm confident is worse.

What I don't know yet

Whether the automated-twice-a-day version actually beats the hand-picked version, or whether volume past some point just trains the algorithm to discount you. I don't have the arsenal deep enough yet to run that test honestly, and I suspect the answer is different for reach than it is for clickthrough. But the part I'm sure of is the cheap part: the idea doesn't have to be new, and refining how you say it beats waiting for something new to say.