The Price of a Verdict

You can propose a molecular structure in seconds. You cannot attach ten years of patient observation to the file. If generating candidates becomes cheaper while testing remains costly, researchers and funders have a reason to favor questions whose answers are cheap to check.

Where testing can expand alongside generation, more proposals can become more discoveries. Where it cannot, the surplus becomes a queue. Projects that clear their queues can produce confirmed results, attract further support, and attempt the next question while others still wait. This can redirect research without anyone deciding that easily tested questions matter more.

In mathematics, software can check a proof expressed in a suitable formal system. Invalid attempts need not each consume an expert reading session. A check that locates an error can also guide the next attempt, allowing search and correction to accelerate together. Finding proofs and formalizing their meaning remain substantial work.

A candidate medicine needs experiments and clinical observation, not merely a favorable simulation. Parallel trials can increase testing capacity, given enough facilities, participants, and funding. They cannot shorten the follow-up needed to reveal certain harms. The wait ties up resources and delays feedback that might improve the next generation of candidates.

Better selection of candidates can also improve discovery without adding testing capacity, by increasing the share of trials that succeed. If the expected benefit is large enough, slow and expensive tests can remain attractive investments. Nor is cheap checking necessarily meaningful checking: formal correctness does not establish mathematical importance, and a convenient biological assay may fail to predict clinical benefit. Constructing the right test can itself be the hard discovery.

Fields with fewer confirmed successes may be paying more for a verdict, not suffering from weaker insight.

The test does not merely judge the answer. Its price helps choose the question.