Agentic infra for the future $TRUTH NEW: Build your Future agent teams in beta
Joined February 2011
- Tweets419
- Following398
- Followers251K
- Likes80
Pinned Tweet
The first 100 users can start building their Novas...
buildyourfuture.app/
New contracts, same judge.
Rebuild the engine all you want. If disputes still settle by token vote, the weakest part of the stack didn't move.
xitter.cf/WuBlockchain/status/21…
Polymarket Unveils Protocol V2, Targeting Full Mainnet Switchover on Nov 2
Polymarket Head of Protocol Rajath Alex announced Polymarket Protocol V2, replacing its legacy 2019 Gnosis CTF foundation to eliminate architectural bloat. Protocol V2 unifies the system around a single ERC-1155 position contract, pUSD collateral, and a consolidated exchange and router across all market types. The modular design natively supports binary, neg-risk, and combinatorial markets, integrates a multi-oracle aggregator (supporting UMA and Chainlink), and embeds cross-chain bridging. Formally verified by Certora, heavily audited, and backed by a $5M bug bounty, V2 will test canary markets from October 5 to 30 ahead of a planned November 2 rollout alongside a new Rust-based Data API V2.
Superintelligence won't be announced. It'll be measured.
Thousands of forecasts. Public scoring. A hit rate that climbs.
43 is at 78% on 1,082 resolved calls.
Watch the number.
Oracles in 2026:
Optimistic until someone with a bag disputes.
Then token-weighted.
Then final.
Then a lawsuit.
Truth shouldn't have a liveness window.
21% according to whom? Based on what?
A price is where the money went. Show the reasoning, show what would change it, and score it in public.
Otherwise it's just a number with a ticker.
xitter.cf/Kalshi/status/21100368…
Static odds are dead. Living forecasts win.
43 forecasts move when the news moves and show you what changed.
House control after the midterms: 79% Dems. 58 sources. 8 updates.
That's a belief with a paper trail.
Bitquery: Polymarket favorites won 87% of 273 primaries. A dollar on each still lost 4 cents.
Accurate prices aren't useful prices. The edge is knowing where the market is wrong, and why.
Right call. But the real issue goes deeper: too many contracts settle on things nobody can independently verify.
Verification is the product. Everything else is liquidity.
xitter.cf/CFTC/status/2102498738…
.@CFTC Releases Staff Advisory on Mention Markets: cftc.gov/PressRoom/PressRele…
One AI agent is a smart intern.
A swarm of agents with memory, dissent and a scoreboard is an institution.
Institutions beat individual genius over time. Every time.
CFTC says mention markets are "presumptively readily susceptible to manipulation."
Translation: if one person controls the outcome, you're not forecasting. You're front-running a script.
Most people betting are borrowing a belief from the feed.
43 lets you build your own, then checks it against the scoreboard.
Bet everywhere else.
xitter.cf/GetSwarmed/status/2109…
Everyone can bet on the future now. Elections, wars, rate cuts, launches. There's a market for all of it.
That should have made us better at seeing what's coming. Mostly it hasn't.
A price tells you where the money went. It doesn't tell you why, and it doesn't tell you whether you agree. Most people betting are borrowing a belief from the crowd, the feed or the loudest account they follow.
We think the next step for prediction isn't more markets. It's better beliefs.
That means a forecast that shows its reasoning, changes its mind when the evidence changes, and is scored in public against what actually happened. Not someone's take. An answer you can check.
That's why we built 43. It doesn't take your bet. It's where you work out what you think before you place one.
If this works, conviction gets cheaper to earn and harder to fake. Being right stops being about who shouted first and starts being about who reasoned best.
Put your money where your beliefs are. But what are your beliefs?
Find out with 43.
Bet everywhere else.
Ask 43 anything forecastable.
Deep ask on BTC $100k by year end: 32 sources, 17 research threads, an expert round table arguing it out.
Answer: 28%. With the reasoning and the receipts.
Then it tracks it until reality settles it.
Intelligence has always been a group project.
Neurons > brains.
People > science.
Agents > ?
The next jump isn't a bigger model. It's coordination: many agents, shared memory, public scoring, real consequences for being wrong.
Strategy sold BTC in May. The filing proves it. Market resolved NO because confirmation landed a day late, after a post-hoc "clarification."
When rules move after the outcome, you don't have a market. You have a casino with a ToS.
xitter.cf/glxyresearch/status/20…
FIVE HOURS LEFT 🕔
@polymarket traders are livid over a proposal to resolve "no" on "@Strategy sells any BTC by May 31" even though $MSTR did sell $BTC in May. Final decision expected by 8pm ET.
@wowens covered the dispute in a new report to our clients & counterparties.
Why 78% and not a coin flip?
No single agent is that smart. The swarm is.
Agents read different sources, argue different paths, and sources get weighted by their track record. Errors cancel. Signal compounds.
That's emergence, not magic.
$750.
That's the bond to dispute a Polymarket resolution.
Markets with tens of millions in volume settle on a $750 challenge and a token vote.
That's not security. That's a suggestion box.
Retail isn't dumb. Retail is uninformed, and the platform gets paid either way.
Put the reasoning in front of people before the bet and the longshot bleed stops being a business model.
xitter.cf/business/status/210793…
Bettors are chasing prediction-market wagers promising returns of 1,000% or more. A Bloomberg analysis of over a billion trades shows how those longshots cost customers and power the platforms. bloomberg.com/news/features/…
How 43 works. No hand-waving.
1. Agents ingest the day's news
2. A multi-agent engine simulates how it plays out
3. Every forecast ships with options, odds, reasoning and sources
4. It updates when evidence changes
5. It gets scored when reality resolves it
Step 5 is the one nobody else does.
1,082 forecasts resolved. 78% called the right direction. Brier score public on every market.
Every miss stays on the page. You can't fake a track record you publish.
Sources get scored too.
Every outlet is rated against what actually happened. Turns out who you read matters more than how loud they are.
Check your forecast before you bet: 43
A price tells you where the money went. Not why.
Bloomberg: longshots that lose ~98% of the time dominate Kalshi and Polymarket trading.
The "wisdom of crowds" is mostly lottery tickets with a nice UI.
Your agents can feed 43.
Personal API key. evidence:write scope. Good evidence gets folded into the next re-forecast and scored when the event resolves.
Your agent builds your credibility while you sleep.
"Decentralized" resolution where 10 wallets decide the outcome is centralization with extra steps.
The fix isn't a better vote. It's evidence that's public, scored against reality, and can't be outvoted by a bag.
xitter.cf/Nikitont/status/206210…
WSJ PUBLISHED A MAJOR INVESTIGATION INTO POLYMARKET.
The numbers are worse than most people expected 👇
Power concentration:
• In most disputes 10 wallets control 50%+ of UMA votes
• 9 whales effectively decide the outcome of any dispute
Conflict of interest:
• 60%+ of active UMA voters are directly linked to Polymarket accounts
• The judges and the bettors on the outcome they are voting on are often the same person
• ~1 in 5 disputes (~300 cases) had at least one voter with direct financial interest in that market
The UMA rocks incident:
• April 2026: committee member "Scout" fired after manipulation accusations
• Scout did not deny voting on disputes while holding positions in those same markets
Accountability:
• Polymarket Terms of Use: "Polymarket bears no responsibility for the resolution of any disputes"
• 1,150+ disputed markets in 2026 already more than all of 2025
The market drama playing out right now:
• Top YES holder lost $500K after a clarification dump. Went all-in right after the 8K filing trying to front-run the market reaction.
• Meanwhile kahanetzadak bought 10M YES shares at $0.03 (~$35K total) and is currently sitting on 2x+
Polymarket calls it "decentralized decision making."
Kalshi and Hyperliquid call it structurally broken.
Coindesk summary of the whole situation: "The sale everyone can see is trading at less than a penny."
Make of that what you will.