A forecast is a promise. A paper is proof that the promise can be kept. We took our own forecasts, turned them into concrete questions, tested them with rules written in advance, and wrote up what we found. Papers came out, and we keep working on new ones — in general relativity, computing and chemistry from the arXiv generations, and in energy engineering, animal science, computational mechanics, cognitive psychology and dermatology from the flagship v3, which reads all of science.
From a forecast to a result
The same chain, eight times. Every step is public: the forecast with its fingerprint, the question, the test fixed in advance, the numbers, the paper.
Read the papers
Each study has its own page, written for everyone, and the full paper — as a PDF, or as the full text with its LaTeX source and code.
Why it matters
A list you can work on tomorrow.
Each of these studies started from a single line on the Frontier: a method and a field. No hidden instructions — the forecast was enough to find a real question.
Better than the state of the art, where it was.
One formula beats the established fits for black-hole overtones; one approximation beats the best-known ones for parallel queues; one benchmark shows a widely used AI tool picks the wrong structure two times out of three.
Studies with no lab. Imagine with one.
These were done in days, on a laptop, with open data and simulations. A research group, a company or a funding agency working from the same list can go much further — and much faster than by waiting for chance.