Opinion by: Sasha Shilina, founder of Episteme and researcher at Paradigm Research Institute
Prediction markets are trying to become serious financial plumbing, and regulators are noticing. Tennessee’s cease-and-desist orders against Kalshi, Polymarket and Crypto.com, followed by a federal judge temporarily blocking Tennessee from barring Kalshi’s sports event contracts, show how quickly “information products” can be treated as unlicensed wagering when legitimacy is contested.
The next argument matters more than the court filings. If prediction markets claim to aggregate information, the system has to know whether it is hearing many independent humans or one actor echoing itself.
That flaw is not theoretical. A Columbia-affiliated analysis of Polymarket’s onchain activity concluded that artificial trading accounted for an average of about 25% of buying and selling over the past three years, with the share varying significantly over time. If a market’s displayed signal can be materially shaped by synthetic activity, then the platform is not just a forecast engine. It is also an incentive engine for whoever can cheaply manufacture participation.
That is already a problem in event markets. It becomes fatal in science.
Citizen science without Sybil resistance becomes synthetic consensus
Scientific prediction markets are the obvious next frontier because they can tighten feedback loops. A market price can update faster than peer review, and resolution can be anchored to measurable artifacts: data sets, pre-registered endpoints, replication outcomes and benchmark performance.
Science is also where open participation breaks fastest. If it is cheap to impersonate a crowd, it is cheap to manufacture “consensus,” and in science, consensus shapes funding, replication priorities and reputations. Manipulation does not only distort a chart; it distorts what gets studied.
The romantic version of science markets is “citizen science”: many participants, many independent views, one shared signal. That only works if “many participants” is a social reality rather than a number that can be minted.
Decentralization does not solve an identity problem
Some builders assume decentralization fixes manipulation because everything is transparent. Transparency helps audits, but it does not prove personhood.
If the base unit is a wallet, influence scales with the number of wallets an actor controls. In incentive-heavy environments, that becomes a business model. The moment there is money on the line, identities will be manufactured, and the market will faithfully price the manufactured reality.
Scientific markets are uniquely vulnerable to one specific failure mode: premature convergence. A coordinated cluster of accounts can seed early liquidity, nudge prices into an apparently confident range and create the appearance of consensus before public evidence should justify it. Humans see a settled-looking probability and hesitate, even when uncertainty should still be the honest state of the world.
This is how a market can start looking predictive without actually aggregating knowledge.
Sybil-resistant projects already exist, and markets should use them
The controversial claim is also the practical one: Prediction markets, especially for science, should require Sybil resistance at the participation layer. These are safeguards that stop a person, or a bot operator, from cheaply creating many fake accounts to impersonate a crowd and distort outcomes.
This does not require doxing users. The requirement is narrower: Make “showing up as many” meaningfully expensive while keeping “one human” feasible.
There is no single best approach, and that is good news. The ecosystem already contains multiple strategies with different trade-offs.
Related: Massachusetts judge bars Kalshi from offering sports bets
One class of approaches uses credential aggregation: Human Passport (formerly Gitcoin Passport) aggregates “stamps” from different identity sources into a score intended to raise the cost of bots and multi-account farming.
Another class uses social-graph uniqueness: BrightID frames this as “proof of uniqueness,” using a privacy-first social identity network so apps can enforce “one account per person.”
A third class uses curated registries with dispute processes: Proof of Humanity is designed to create a trusted list of humans verified by a decentralized community, with challenges/disputes routed through Kleros-style arbitration.
A fourth class uses privacy-preserving biometrics to enforce uniqueness at the wallet layer: Humanode Biomapper positions itself as “private on-chain biomapping,” linking a unique, live human to an Ethereum Virtual Machine address while keeping biometric data private (in confidential environments), so decentralized applications can check a “verified human” status rather than raw biometrics.
None of these are perfect. Some create friction. Some risk exclusion. Some rely on community processes that can be gamed. But the alternative is worse: Markets where influence is a function of how many identities can be spun up cheaply.
Science markets cannot afford that. Their whole purpose is to turn uncertainty into a signal that tracks reality, not a signal that tracks coordination.
Integrity becomes a competitive edge
If Sybil-resistant science markets work, they create something new: an information derivative tied to measurable outcomes. That matters for biotech, where uncertainty is expensive, and timelines are long. It also matters for AI-for-science, where claims about models, benchmarks and experimental performance compete for credibility in real time.
The upside is not “better gambling.” It is better allocation: a continuously updating public signal that can guide what gets replicated, what gets funded and what gets treated as credible under uncertainty.
The downside of ignoring identity is already visible in the category’s regulatory attention. When legitimacy is contested, enforcement rarely arrives with nuance.
Prediction markets do not become credible because they are decentralized. They become credible when participation maps to humans rather than to cheap accounts. Science is the domain where that distinction will decide whether markets produce knowledge signals or just produce charts.
Opinion by: Sasha Shilina, founder of Episteme and researcher at Paradigm Research Institute.
This opinion article presents the contributor’s expert view and it may not reflect the views of Cointelegraph.com. This content has undergone editorial review to ensure clarity and relevance, Cointelegraph remains committed to transparent reporting and upholding the highest standards of journalism. Readers are encouraged to conduct their own research before taking any actions related to the company.
This opinion article presents the contributor’s expert view and it may not reflect the views of Cointelegraph.com. This content has undergone editorial review to ensure clarity and relevance, Cointelegraph remains committed to transparent reporting and upholding the highest standards of journalism. Readers are encouraged to conduct their own research before taking any actions related to the company.
