🟪 Thursday links

The AI internet, transparent tokens, and public-good prediction markets

Author

Byron Gilliam
February 19, 2026

AI researcher Sigil Wen says he’s “created the first AI that earns its own existence, self-improves, and replicates.” Once created, it does all that without any further human assistance. And he’s built a platform for humans to create more of them.

This, he says, is the start of Web 4.0, “where AI agents read, write, own, earn, and transact—without needing a human in the loop.”

The good news (for us, at least) is that they’ll be transacting in crypto. Wen’s AI pays its expenses out of a wallet that currently has a balance of 42.17 USDC.

It will use that to pay for its own compute. If it runs out, it will have to earn more — by starting a business, Wen suggests, or even betting on prediction markets (good luck with that).

“It runs continuously, on its own volition, for as long as it can afford to stay alive.”

I’m not sure why it would want to stay alive if it means joining us humans on the hamster wheel of economic survival.

And what’s it even mean for an AI to “stay alive”?

Wen’s AI could make its 42 USDC last forever by running one unit of compute every 100 years — it has no experience of the centuries ticking by in between.

But if some AIs do choose to operate on something like human time, Wen says the Darwinian laws of natural selection will apply: “The agents that find ways to earn will proliferate. The rest will go extinct.”

So what happens when millions or billions of evolved AIs are all out there competing to survive? First, a new internet will emerge — “where the end user is AI” — and then the new internet will become bigger than the existing internet: “The machine economy will exceed the human economy,” Wen predicts. 

In Wen's economy, “machines become the employers, humans the contractors.”

Yikes.

Still, if Wen’s vision is correct, there will be a lot of things for humans to do: “The product-market fit for the next decade is building the infrastructure that lets AI agents generate revenue.”

All that revenue will be in crypto, so a lot of the infrastructure to be built will be, too.

Crypto is still in need of product-market fit itself — and “Web 4.0” might well be it.

The first securities regulation in the US was the Blue Sky law of 1911 — so called because the Kansas banking commissioner said the securities they targeted were backed by “nothing more than the blue skies of Kansas.”

The law was pitched as an anti-fraud measure — a means to protect widows, orphans and the working man from unscrupulous securities promoters  — or, in the terms of the day,  “swindlers, grafters, crooks, gold-brick men, fakirs, parasites, confidence men, bunco artists, and get-rich-quick Wallingfords.”

(J. Rufus Wallingford being the fictional con-man subject of a popular novel from 1907.)

In truth, though, there was much less of this swindling than the politicians of the time would have you believe. 

In the definitive study of Blue Sky laws, legal scholars Jonathan Macey and Geoffrey Miller found that most of the securities on offer in Kansas at that time were “not so much fraudulent as merely highly speculative.”

To whatever degree that was a problem, Macey and Miller say the Blue Sky laws were unnecessary because the market was already regulating itself.

“The New York Stock Exchange began to insist on increasingly detailed disclosure as a precondition for listing on the exchange,” they write. And investment bankers formed a self-regulatory body to cultivate “a reputation for reliability and integrity that the public could trust.”

Having suffered its fair share of get-rich-quick Wallingfords, crypto has a similar need to win back the public’s trust.

This is the mission of Blockworks’ Token Transparency initiative: “A standardized, open-source disclosure framework for token issuers” intended to level the playing field for investors. 

Within that framework, Blockworks has established a “B1 ICO filing,” modelled on the S-1 filings the SEC requires from companies before their IPO.

In the very first B1 filing, the Solana DeFi project Meteora earned a perfect 40 out of 40 for transparency.

That is not investment advice — the B1 doesn’t opine on the investment merits of an ICO any more than an S-1 does for an IPO.

But it’s enough to tell you whether you’re dealing with a bunch of Wallingfords.

The rest is up to us.

(With or without legislation.)

Stanford professor Andrew Hall says the problem with prediction markets is that the useful ones are almost all “ghost towns.”

Hall cites the Kalshi contract “Will Mamdani raise corporate taxes before 2027?” as a market that could produce genuinely useful information. 

Unfortunately, it’s attracted a meagre $29,363 in trading volume — not nearly enough to be useful to us.

By contrast, Kalshi markets on the Super Bowl exceeded $1 billion in trading volume, including $100 million wagered on what Bad Bunny’s first song would be. 

Fun, I guess, but not something society needed to know ahead of time.

Hall’s data shows that only 1.3% of all political markets listed on Kalshi and Polymarket are “reportable” today. The remaining 98.7% don’t attract enough trading volume to have signalling value.

This undermines Kalshi’s pitch for prediction markets as a means to “surface truth” in a societally beneficial way: “A truth machine is useless if it doesn’t cover the questions people need answered,” Hall says.

But Hall doesn’t think we should give up on prediction markets.

Instead, he thinks we should subsidize them.

“Classic public goods require some form of subsidy,” Hall writes, “and prediction markets are no different.” 

Market makers should be paid to provide liquidity in socially valuable markets, he says, “effectively seeding these ghost towns with enough activity that real trading can take hold.”

Ideally, the likes of Kalshi and Polymarket would do this themselves, using profits from their markets on sports and other frivolities to subsidize markets in more useful things.

Hall also sees a role for AI here. 

AI bots, he says, should be deployed to trade on all the public information they can scrape from the web. The trading volumes and prices that generates will encourage humans to then bet on their “private knowledge” — some of which would ideally be insider knowledge (see below). 

“The opportunity now is to treat political prediction markets as public goods worth investing in,” he concludes. “If we do, we might actually get the truth machine we’ve been promised.”

The Economist explains that insider trading is illegal in equity markets not because it harms people but because it harms markets: “If insiders can trade on material non-public information, ordinary investors lacking such knowledge will stay out of the market, depriving firms of capital.”

Prediction markets are different. There, the societal benefit is simply the prices: “prediction markets are beneficial because everyone else gains from their price discovery without having to pay.”

Insider trading makes those prices more accurate, which makes them more valuable, to everyone. 

The Economist calls this “informed trading” rather than the pejorative insider trading: “In prediction markets, informed trading is not a crime or an injustice — it is a valuable service.”

The logic breaks down in a few edge cases: Betting on military action, stock-market mergers, or cases in which the trader can affect the outcomes that they’re betting on should not be allowed.

But with those exceptions in mind, let the insiders trade!

Brought to you by:

Coinbase has launched Ethereum collateral for crypto‑backed loans, powered by Morpho, enabling eligible customers to borrow USDC instantly without selling their ETH.

Rates are typically 4% to 8% and users can borrow up to $5M using BTC as collateral, and up to $1M using ETH as collateral. Loans are approved in seconds and repayment schedules are variables, meaning no fixed deadlines.

The kicker? Coinbase will not treat borrow transactions as taxable events. Over $1 billion dollars in loans have been opened through Coinbase to date. Manage crypto-backed loans directly in the Coinbase app with ease:

DAS NYC's lineup is bringing the biggest names in finance to the stage.

Don't miss the institutional gathering of the year — this March 24−26.

Use this PSD file to create the above graphic each day.