Comment Text:
The academic research replication crisis is well-documented and structurally persistent. Journals seek novel, positive findings; data models and code are tightly protected; and researchers run multiple trials, tweaking data & variables each time in an effort to publish the most favorable outcome ("p-hacking"). Researchers face little downside for these practices. In fact, these practices are incentivized by the academic community. Fortunately, academic purists are catching on. For instance, the Institute for Replication runs the "Replication Games" in which teams re-run published papers from top economics and political science journals. The Berkeley Initiative for Transparency in the Social Sciences trains researchers in open-science practice. Society is fortunate that a growing number of journals now carry replication sections and pre-registration requirements, but much more can be done.
These efforts matter. But they share a structural limit: they are backward-looking and low-throughput. The published literature generates thousands of new findings in the time it takes Replication Games to formally re-examine a few dozen. Nothing in the current research ecosystem produces what a well-designed nonprofit prediction market would: a forward-looking, continuously updating, market-aggregated probability that a given finding will survive replication.
I spent time at Juul Labs during the development of its Premarket Tobacco Product Application (PMTA). I was not involved in the scientific submission itself, but every employee was well-aware of what a federal regulatory filing of that magnitude actually involves. Juul submitted more than 110 scientific studies across toxicology, clinical pharmacology, behavioral science, and real-world evidence. The FDA denied the application in June 2022, stayed its own denial within two weeks, rescinded the denial in June 2024, and granted marketing authorization for five products in July 2025. This was roughly the same body of evidence, weighed differently at different points.
I take no position on whether the agency ultimately got the outcome right. What I can say is that every commercial counterparty in Juul's orbit also had skin in the PMTA game - and the volatility of the decision directly contributed to the volatility of the counterparty relationships. Vendors signing supply contracts, banks extending credit facilities, investors underwriting the company's equity, even distributors holding inventory exposure were all making decisions contingent on this regulatory outcome. Importantly, this outcome was itself contingent on scientific studies, studies which outsiders had no independent way to evaluate.
Imagine a replication market on the outcomes of these studies. Let me be clear: a market-aggregated probability on whether the key toxicology and behavioral findings would replicate would NOT have resolved the regulatory question. Instead, it would have fed directly into credit decisions, vendor commitments, valuation work, and counterparty risk management. That is § 3(a) in its plain sense: managing price risk, discovering prices, and disseminating pricing information.
My background is in finance and technology. After receiving my undergrad degree at The Wharton School, I held extensive operating roles in CPG and fintech, and I gained experience with investing in both public and private markets. I've watched the digitization of conventional markets flatten spreads, erode information asymmetries between retail and institutional participants, and enable mechanics that pit-era markets could not have supported. Digitization also expanded what can be traded: not just new instruments on existing underlying assets, but entirely new tradable markets on questions that previously had no market-clearing price at all. Research-claim probability markets are a natural extension of that trajectory into a domain where the information output is a public good and the downstream users include every regulator, firm, and investor whose decisions depend on whether a published finding will hold up.
The two-submarket thesis tracks operational reality: a market whose output is a probability on a replication attempt shares almost nothing with a market on the Super Bowl, and the Commission's regulatory framework should reflect that. Research markets and commercial markets are not parallel categories sitting side by side. Rather, research markets are upstream inputs to the commercial markets the Commission already oversees. Treating the nonprofit-research pathway as a narrow carve-out understates what it's actually providing. Codifying that pathway rather than continuing to administer it through discretionary staff letters would let this information infrastructure develop at the scale the evidence environment increasingly requires.
