From Here to There

Changing the incentive structure of all of science would require a great deal of money, and it is not the sort of thing the government ought to do without strong empirical evidence behind it. It is also better phased in over years, or even decades. But it should be possible for private organizations to prove the concept first.

Suppose a large foundation dedicated to scientific research wanted to try. It could start a journal in a relatively small or new subfield, such as the economics of science itself, or AI alignment, using ordinary peer review or something better. It would draft and publish a legal template for the ownership of scientific papers, and then pay the works cited in its journal, on the condition that those works are registered under its ownership template and were published after the journal began. Cited articles that are not freely accessible to the public might receive half as much as public ones. And articles would earn a weighted score, counting citations in this journal, or in others that pay their citations, more heavily than the rest.

If the journal is reasonably well funded, the established leaders of the field, who are already cited often, would have an immediate financial incentive to register the ownership of their papers, even if they keep publishing elsewhere. Having the most prominent scientists in a field register with the journal should be enough to establish it as prestigious. A foundation that preferred not to start from scratch could instead buy an existing journal at the top of its field and keep its editors and processes intact. Because publishing elsewhere would not earn the same rewards, the journal would soon become the first choice for authors looking to benefit from this new revenue stream, and if its editors are any good it would settle at the top of its field. At that point it could taper off the weighted score, having served its purpose.

Once the model is proven in one field, private funders in other fields could do the same, creating new funded journals or buying existing ones and adding citation grants. The government could then run a trial program for ten years in a sample of fields it wants to support. Without cutting its ordinary funding budget, it would begin paying the citations of new articles, starting from a fixed date, in existing journals.

If the trial succeeds, the existing grant-making bodies, universities among them, would gradually shift from donors into investors and incubators. With enough funding flowing through journals, major research universities could become less dependent on tuition and unreliable state budgets, and new research centers and universities could succeed or fail on the quality and significance of what they produce rather than on their ability to court donors.

What it does not fix

This would not cure every ill. It rewards influence, and influence is not always the same as truth, so a fashionable mistake could be well paid for a while before the field corrected it. It would favor fields where impact arrives quickly over those where the important work takes decades to be recognized, which the per-field pools can only partly offset. And every measure invites someone to game it. These are real costs. The claim is only that they are smaller, and easier to fix, than the problem we have now, of trying to decide what science is worth before it has been done.

Read the draft legislation: The Marketplace of Ideas Act