It's easy to look at a company's trajectory after the fact and assume the signs were obvious. They rarely are, in the moment. The companies that eventually became unicorns spent meaningful stretches of time looking, to almost everyone outside a small technical community, like nothing in particular — a repository with a modest following, a Hacker News post that got a few hundred upvotes and was forgotten by the next morning, a Product Hunt launch that didn't trend.
What actually distinguishes the pattern from noise
The difference between a startup that fades and one that compounds isn't usually visible in any single moment. It's visible in the trajectory across multiple independent signals over time — sustained contributor growth rather than a single spike, cross-platform corroboration rather than isolated mentions, and novelty scores that stay elevated rather than decaying back to baseline after the initial post.
The three collectors feeding this domain
- GitHub Trending — repository momentum, commit velocity, contributor diversity
- Hacker News — front-page appearances and sustained discussion engagement, not just submission
- Product Hunt — launch activity and follow-up engagement over the following weeks, not just launch day
The OpenAI precedent, again
Twelve entities currently tracked in this domain show signal profiles matching OpenAI's own 2017–2019 pre-recognition pattern — a period when OpenAI itself looked, to most of the public, like a research lab among many. The pattern that eventually became obvious was visible in the signal data years before it became obvious in headlines.