π 30,000 OF OUR USERS WEREN'T PEOPLE
We went through every account on XREIGN. It wasn't a few cheaters here and there β it was big farms, one person running hundreds of accounts at once. π€
And the money is the smaller problem here.
We're building REIGN AI, and it learns from what people actually do on the platform. If thousands of those "people" are scripts, the model learns from fake behavior. We'd end up with an AI trained on bots.
So we marked all 30,000 as bots. Nothing was deleted β they just don't count anymore. Not for the AI, not for the rewards.
If you're a real user, your piece of everything just got bigger.
Check your account here β everyone who passes gets a certificate π
https://t.co/i9AhSeVRr3
And if we got it wrong β your account is marked as a bot but you're a real person β just write to our support. We'll check it manually and fix it.
Foresight is five days old. Here's every day of it.
Day 1 β crowd 50.0% Β· REIGN AI 50.0%
Day 2 β crowd 51.6% Β· REIGN AI 57.5%
Day 3 β crowd 57.3% Β· REIGN AI 56.5%
Day 4 β crowd 58.9% Β· REIGN AI 56.5%
Day 5 β crowd 60.0% Β· REIGN AI 55.4%
The model came out first. The crowd passed it on day three and
hasn't given it back.
502,448 calls from 36,808 people in five days.
Two real posts from the live timeline. Six hours. X decides which one
wins β you call it, REIGN AI calls it sealed and hashed before you
ever see the pair, so it can't hedge and can't rewrite its answer
after the fact.
The top 10% of callers are running at 68.6%. Twelve points above the
model, and the one thing no lab can scrape: people who already know
why one post travels and the next one dies.
β‘ 25 $REIGN per call β right or wrong
π― +60β250 when you're right
π Γ2 with an Influence Passport
New scoreboard every week.
The day the model passes you again, you'll have built the thing
that beat you.
https://t.co/7cAdppqbH0