Leaderboard — top drafts by predicted virality

Every draft scored on Foresyn HN, ranked by what the model thinks it would do on real Hacker News. Bigger predicted-score = stronger position. Beat the number — paste a fresh draft on the front page and try.
1.Meshtastic: Open-source, off-grid, decentralized encrypted mesh network (meshtastic.org)
predicted 342 points by anon | 0 comments
2.Show HN: I scraped 183 Telegram AI bots — 44.3 million users and a broken discovery layer (x.com)
predicted 329 points by anon | 0 comments
3.OpenStack is not dead (x.com)
predicted 328 points by anon | 0 comments
4.We're making an open-source $30 GPS/mesh radio and would like advice (meshtastic.org)
predicted 323 points by anon | 0 comments
5.An Introduction to Meshtastic: Off-grid messaging without the internet (meshtastic.org)
predicted 319 points by anon | 0 comments
6.Show HN: I scraped 183 Telegram AI bots — 44.3 million users, 7.3 million on BotFather alone (x.com)
predicted 315 points by anon | 0 comments
7.Show HN: Telegram is the agent economy nobody on HN noticed — 11 million dollars in 4 days (x.com)
predicted 310 points by anon | 0 comments
8.Show HN: I tracked 11 million dollars of Telegram payouts, 200 million TON staked, in 4 days (x.com)
predicted 309 points by anon | 0 comments
9.Show HN: I tracked 18 months of TON exchange flows — 2026 is the first year more leaves than enters (x.com)
predicted 303 points by anon | 0 comments
10.Show HN: Telegram paid out 11 million dollars to channels and bots in 4 days (x.com)
predicted 272 points by anon | 0 comments
11.Why relying on cloud budget alerts won't stop 429 bill shocks (github.com)
predicted 52 points by anon | 0 comments
12.Load is not requests per minute (github.com)
predicted 37 points by anon | 0 comments
13.Rate limiting beyond the 429 (github.com)
predicted 31 points by anon | 0 comments
14.Show HN: I scraped 183 Telegram AI bots — 44.3 million users and a broken discovery layer (news.ycombinator.com)
predicted 31 points by anon | 0 comments
15.I spent six months building an admission controller for the edge (github.com)
predicted 30 points by anon | 0 comments
16.200 lines of code to enforce hard budget caps on LLMs (github.com)
predicted 28 points by anon | 0 comments
17.Why are rate-limiting libraries so bloated? (github.com)
predicted 24 points by anon | 0 comments
18.Telegram as an agent server (github.com)
predicted 24 points by anon | 0 comments
19.Stop sending your AI prompts to the cloud (play.google.com)
predicted 24 points by anon | 0 comments
20.I analyzed the 40% failure rate on DMV tests: Here is how to pass (driving-tests.org)
predicted 22 points by anon | 0 comments
21.Telegram is an agent server (github.com)
predicted 22 points by anon | 0 comments
22.Show HN: Ten years of systems administration and the furry subculture (example.com)
predicted 22 points by anon | 0 comments
23.Client-side tracking is broken: Why server-side APIs are now mandatory (capi.pro)
predicted 20 points by anon | 0 comments
24.Lua 5.1 is too slow for the modern key-value store (github.com)
predicted 20 points by anon | 0 comments
25.Manual programming considered harmful (github.com)
predicted 20 points by anon | 0 comments
26.You might not need Redis (github.com)
predicted 19 points by anon | 0 comments
27.From zero to 10,000 procedural pigs: A simple path to generative NFTs (pigsinblankets.io)
predicted 18 points by anon | 0 comments
28.Show HN: I built a local-first SQLite notebook that runs entirely in the browser (github.com)
predicted 18 points by anon | 0 comments
29.Remind HN: Today is Mother's Day, call your moms
predicted 18 points by anon | 0 comments
30.Request counts are dead (github.com)
predicted 17 points by anon | 0 comments
31.Show HN: I scraped 183 Telegram AI bots — 44.3 million users and a broken discovery layer (okhlopkov.com)
predicted 15 points by anon | 0 comments
32.Show HN: I scraped 183 Telegram AI bots — 44.3 million users and a broken discovery layer (ohldbot.swanrate.com)
predicted 15 points by anon | 0 comments
33.Show HN: I analyzed 1.2M posts to predict Hacker News front-page success (hackernews.foresyn.ai)
predicted 15 points by anon | 0 comments
34.Show HN: A 90-day log of Claude Code running my Telegram inbox — 92 RPC tools, no human (ohldbot.swanrate.com)
predicted 14 points by anon | 0 comments
35.Show HN: 90 days auditing Skygen and the vulnerabilities I found (skygen.ai)
predicted 14 points by anon | 0 comments
36.Show HN: 90 days of Claude Code on a single Hetzner box running my Telegram — 92 RPC tools
predicted 14 points by anon | 0 comments
37.Show HN: A 90-day log of Claude Code running my Telegram inbox — 92 RPC tools, no human
predicted 14 points by anon | 0 comments
38.Show HN: A high-performance thrust engine written in C (github.com)
predicted 14 points by anon | 0 comments
39.Show HN: I scraped 183 Telegram AI bots — 44.3 million users and a broken discovery layer (identityhub.app)
predicted 14 points by anon | 0 comments
40.Show HN: I scraped 183 Telegram AI bots and found 44.3 million users with no discovery layer (identityhub.app)
predicted 13 points by anon | 0 comments
41.Show HN: I scraped 183 Telegram AI bots — 44.3 million users and BotFather doubled in 2 months (identityhub.app)
predicted 13 points by anon | 0 comments
42.Show HN: Benchmarking Opus 4.7 vs. M2.7 in a real production session (andlukyane.com)
predicted 13 points by anon | 0 comments
43.10,000 posts analyzed: The data behind the Hacker News front page (hackernews.foresyn.ai)
predicted 13 points by anon | 0 comments
44.Show HN: Sort iOS icons by color using an evolutionary algorithm (github.com)
predicted 12 points by anon | 0 comments
45.Show HN: I scraped 183 Telegram AI bots and found 44.3 million users that web search can't see (ohldbot.swanrate.com)
predicted 12 points by anon | 0 comments
46.Show HN: 10 years in hardware and why we ignore the best engineering
predicted 12 points by anon | 0 comments
47.Show HN: I scraped 183 Telegram AI bots and found 44.3 million monthly users with no discovery layer (identityhub.app)
predicted 12 points by anon | 0 comments
48.Show HN: 183 Telegram AI bots are bigger than most SaaS — 44.3 million monthly users (identityhub.app)
predicted 12 points by anon | 0 comments
49.Show HN: I scraped 183 Telegram AI bots — 44.3 million users and Google can't see any of them (identityhub.app)
predicted 12 points by anon | 0 comments
What this ranks. Predicted score from our virality model over a 39K-post HN corpus — model output, not real HN votes. Filters: drafts with predicted score ≥ 1 and title ≥ 12 chars (cuts empty-test pokes). Sorted by predicted score desc, tie-break newest first. Want to see how the model does on the real HN front page? /predictions updates every 10 min.
ForesynWanna keep in touch?
Built this solo over a weekend. Soft-launching before the HN post on Monday. If you scored a draft and the prediction either nailed it or whiffed, I want to know.
DM @crimeacs on Telegram — fastest way to reach me
Connect on LinkedIn — Artemii Novoselov