The soil · a toy

Soil tester

Paste a post and see which engagement-bait tells it triggers. A transparent phrase-matching toy with known false positives — not a judge of intent, context, or any community's rules.

Read this first. The Soil Tester is a toy. It matches phrases and formatting that often appear in engagement bait — “comment YES below”, “tag 3 friends”, “link in bio”, one-line-per-paragraph broetry, hashtag walls — and adds up weights into a score. It cannot read intent, account history, or the rules of the place you're posting in. A sincere “what do you think?” scores the same as a fake one. Use it to notice habits, not to judge people.

Soil tester · v1.4 · runs in your browserNothing leaves this page
Try one:
⌘ + Enter

How it scores

Each detector below looks for one family of tells. The first match in a family counts once, with the weight shown; continuous signals (emoji density, hashtag count, very short lines, ALL CAPS) add a little more. The total is squashed onto a 0–99 scale with diminishing returns, then banded: Heirloom (under 22), Lightly tilled (under 48), Cultivated (under 74), Factory farm. The source is readable at /assets/farm-score.js; nothing is hidden.

Detector Weight Looks for
Comment harvesting 30 “comment X below”, “drop a…”, “reply with”, “DM me”, the 👇 emoji
Tag / giveaway bait 28 “tag 3 friends”, “giveaway”, “to enter”, “follow + RT”, “winner announced”
Link-gating 26 “link in bio”, “comment and I'll send you…”, “want the template?”
Share / RT solicitation 24 “RT if”, “share this”, “repost”, “save this post”, “bookmark this”
Engineered outrage 22 “they don't want you to…”, “delete this”, “say it louder”, “controversial but”
Manufactured curiosity hook 16 “unpopular opinion”, “hot take”, “nobody talks about”, “here's why/how”
Broetry formatting 15 six or more lines where most are four words or fewer
Fake-question prompt 14 “thoughts?”, “agree?”, “am I the only one”, “who's with me”
Authority / growth flex 13 currency amounts, “0 → 10k”, “in just 30 days”, “went viral”, follower counts
Thread / listicle scaffolding 12 🧵, “a thread:”, “7 lessons”, “1/12”
Emoji, hashtag, ALL-CAPS stuffing 8–18 density rather than any single phrase

Known false positives

These are documented, not hypothetical. The research behind this guide inspected the scoring code and found that it will penalise:

The 0–99 output looks precise. It is not calibrated against any labelled corpus. If your score surprises you, read the signal list — the reasons are the useful part.

What to do with a result