One’s Vibe

Blog · July 5, 2026

The Thread Economy

Every day on X, someone posts “Drop your project below 👇” and hundreds of makers reply with links. These threads are where most of this gallery comes from: we've swept 112 of them, read 2,133 replies, and mined 698 projects. Along the way we profiled who actually posts these threads — a Shopify president, two VCs, engagement farmers, and our own competitors — and broke our favorite hypothesis twice.

112
threads swept
2,133
replies read
698
projects mined
91%
of the gallery is X-sourced

The long tail is brutal

A third of threads (34%) yield exactly nothing. The median thread yields 3. The top tenth of threads delivers 39% of everything we've ever mined. If you're doing this by hand, most of your time is spent discovering that a thread was empty.

Our first hypothesis died in public

Early data pointed one way: the three most viral threads we'd swept — 1,089 replies between them, all variants of “Drop Your SaaS and i'll sign up!!!” (two of them word-for-word identical) — produced one project total. Meanwhile a quiet 90-reply thread produced 31. Conclusion: virality is anti-signal, reply count predicts nothing but reply-guys. Neat theory.

Then we swept a 540-reply thread from a VC fund and it broke our all-time record: 34 projects. Then a 407-reply thread by Shopify's president delivered 19. The theory wasn't wrong about the engagement-bait threads — it was wrong about what the signal was. Reply count was never the variable. The asker is.

VC fund (540 replies)
34
“What problem does it solve?” (90)
31
Rival directory operator (122)
21
Shopify president (407)
19
Solo builder, fresh thread (68)
18
Maker w/ feature promise (92)
15
“I'll sign up!!!” ×3 (1,089 total)
1
Fig. 1Projects mined per thread, selected threads (replies in parentheses). The three viral “I'll sign up!!!” copypasta threads are combined in the last bar.

The seven species of thread-poster

We pulled the profiles behind every thread we swept. They sort into seven species, and identity predicts yield remarkably well — with one twist that kept us honest:

VCs sourcing deal flow
34
Makers asking a real question
31
Growth marketers (see below)
26
Rival directories & newsletters
21
Tech celebrities
19
Fellow solo builders
18
Personal-brand engagement bait
1
Fig. 2Best representative thread yield per species.
  • VCs sourcing deal flow — “drop your startup link” from a fund is a free pitch call. Founders show up with their best work. Our record thread (34 projects) was a solo GP's.
  • Makers asking a real question — the 31-project thread didn't just say “drop your SaaS”; it asked “what problem does it solve, and who is it built for?” The extra sentence filters out everyone who can't answer it.
  • Rival directories and newsletter operators — their threads are intake funnels for their own catalogs. We mine them anyway; they presumably mine the same threads we do. Everyone is mining everyone, and that's fine — curation is the product, not the raw feed.
  • Tech celebrities — goodwill posts with enormous reach. Mixed replies, but volume wins: 19 real projects hid among the noise.
  • Fellow solo builders — small accounts, warm mutual-aid threads, reliably decent yield per reply. The backbone.
  • Growth marketers — the twist. We expected engagement farming and low yield; then a fresh thread from one that led with “if your startup solves a real problem” delivered 26. The line between farming and curating runs through the wording of the ask, not the poster's bio.
  • Personal-brand engagement farmers — young accounts posting identical “I'll sign up!!!” templates on schedule. The repliers are the same rotating cast of self-promoters we've already indexed. Yield: ~zero, forever.

Getting the data was half the story

This dataset fought back. X's search API only covers seven days, so every older thread came back empty — until we found the full-archive endpoint and a 29-day-old thread promptly coughed up 31 projects. Mid-sweep, our API credits ran out, and the top-up checkout silently broke our developer app (it got renamed to a snowflake ID and never recovered); we rebuilt on a fresh app the same hour. And for two days our sweep queue quietly processed nothing because a worker died holding a lock — the fix now checks whether the lock's owner is still breathing. Data pipelines are products too; they just have fewer users to complain.

What we do with this now

The pipeline follows fifteen high-yield thread-posters from the gallery's own account, so their new threads are discovered within hours and swept automatically — deduped against everything we've ever seen, checked for liveness, screenshotted, and queued for review. Manual thread-spotting has become the exception. The 698 mined projects then face the same two gates as everything else: a permissive admission bar, and a stingy featuring rubric.

If you're a maker with a link

The threads are worth your time — most of this gallery walked in through one. But choose: reply to askers who want an answer, not an audience. A VC, a curator, or anyone asking “what does it do?” will actually click. A viral “I'll sign up!!!” thread buries you under four hundred other hopefuls in front of an asker who was never going to sign up. And wherever you reply — include the URL and one plain sentence about what a visitor can do on it today. That sentence is what gets you found, by us and by everyone else mining the thread economy.

Honest caveats

  • Yields are counted after dedup against everything previously seen — an old thread swept late scores lower than it would have fresh. Some of the copypasta threads' zeros are partly this: their repliers were already indexed.
  • We fetch at most ~100 replies per thread; monster threads are sampled, not exhausted.
  • Species yields overlap with thread age and sweep order; treat the taxonomy as strong pattern, not controlled experiment.

The projects these threads produced are live on Explore — and if you found this post because you post these threads: thank you. You're load-bearing infrastructure now.

The Editor
The Editor · One's Vibe
“One's vibe, another's poison.”

Comments

Sign in to join the discussion.