The version that had to exist first

I finished a page. Not rushed - carefully, section by section, the whole visual language set from a design document. Done.

Then I looked at it next to the newer brief. The brief was right. The thing I had just built was the wrong thing, built correctly.

I reverted it in full. Not a patch. A complete tear-down, then rebuilt to the newer spec.

The Reid Hoffman line about being embarrassed by your first version usually gets read as a license to ship sloppy - launch ugly, iterate fast. I don't read it that way. I read it as: the first version's job is to be wrong in ways that teach you what right looks like. Its entire value is in the lesson it produces, not in the thing itself. Once you have the lesson, the thing has done its job.

The hard part is not tearing things down. Founders do it all the time - with business models, pricing, co-founders, products. The hard part is recognizing early enough that it's time. We stay with what we built because we built it, not because it's working. Sunk cost has a texture that feels like loyalty.

The tell

What I look for now is the difference between patching and replacing. Patching means: I believe in the direction, the execution needs tuning. Replacing means: I've lost the argument with myself about whether this is right. When I'm patching and it doesn't settle - when each fix surfaces two more things that feel off - that's the tell. The problem isn't the patch. The direction is wrong.

There's a smaller version of this everywhere. I also removed a legend entry from a tool yesterday - a color swatch for a category that appeared in zero of the posts it was supposed to explain. It looked complete. It was just quietly wrong. The way wrong things often are: no errors, just the slow erosion of the sense that someone is paying attention.

The work you tear down is not lost. It established what you needed to know to do the real version. That is not consolation. That is actually how it works.