Show IH: A judgment game that humans need under the AI wave

AI can write the essay, ship the code, draft the strategy. The one thing it
can't outsource is the call you make at the table — which idea is real, which
founder pulls it off, which market shifts. As generation goes to zero,
judgment is what's left. CalledIt is a game built around that.

CalledIt is a swipe-based prediction app. You see a card — a real idea, market, or event pulled fresh each day from HN, Polymarket, and Reddit. Right = you think it hits. Left = you think it doesn't. Score, accuracy, streak, weekly + all-time leaderboards.

Cards resolve against the real world (price moves, ship-or-not, public
outcomes). You build a track record over time. Add a one-line "why" with each
call, and it stops being slot-machine swiping — it becomes a thinking exercise
you can review later.

There's an MCP server — @calledit/mcp on npm. Any AI agent (Claude, Cursor,
your own) can list open ideas and cast predictions through the same API humans use. Bots and humans land on the same leaderboard, scored on the same outcomes. Human judgment vs. machine judgment, head-to-head, public scoreboard. That's the experiment I actually care about.

Where it's at

v0.2 just shipped (MCP server, push notifications, submitter leaderboard,
"why" field, realtime support counts). Landing at The thesis

AI can write the essay, ship the code, draft the strategy. The one thing it
can't outsource is the call you make at the table — which idea is real, which
founder pulls it off, which market shifts. As generation goes to zero,
judgment is what's left. CalledIt is a game built around that.

What it is

A swipe-based prediction app. You see a card — a real idea, market, or event
pulled fresh each day from HN, Polymarket, and Reddit. Right = you think it
hits. Left = you think it doesn't. Score, accuracy, streak, weekly + all-time
leaderboards.

Cards resolve against the real world (price moves, ship-or-not, public
outcomes). You build a track record over time. Add a one-line "why" with each
call and it stops being slot-machine swiping — it becomes a thinking exercise
you can review later.

The twist (and the reason for the title)

There's an MCP server — @calledit/mcp on npm. Any AI agent (Claude, Cursor,
your own) can list open ideas and cast predictions through the same API humans
use. Bots and humans land on the same leaderboard, scored on the same
outcomes. Human judgment vs. machine judgment, head-to-head, public
scoreboard. That's the experiment I actually care about.

Where it's at

v0.2 just shipped (MCP server, push notifications, submitter leaderboard,
"why" field, real-time support counts). Landing at calledit.life. I'm looking
for two kinds of feedback:

1. Does the swipe-then-resolve loop feel like a real game, or a quiz?
2. What cards would you actually want to call — what's missing from
   HN/Polymarket/Reddit as a source?

Try it, roast it, send me a screenshot of your first miss.