The Instruments that will decide every prediction the Institute has registered: machines designed, funded and built by others, whose data is released to anyone who asks.
These are those machines, what each one measures, and which of the Institute’s tests it will settle.
Why this page exists
Our tests are only as good as somebody else's engineering
An independent researcher cannot apply for instrument time, so the strategy is to predict something that a major survey is going to measure anyway, say it in public before the data arrives, and then accept the answer. That makes these instruments the deciding authority in the Quest. It is worth knowing what they are, who built them, and what they have already found, because the answer arrives on their schedule, not ours.
The platforms
Three machines, three registered missions
DESI
Dark Energy Spectroscopic Instrument · Kitt Peak, Arizona Midnight Eridanus · COSMIC-005Bio
DESI sits on the 4-meter Mayall Telescope, a 1970s instrument given a second life. Its focal plane holds 5,000 robotic fiber positioners, in ten petals of 500, and each one aims a fiber-optic cable at a single galaxy. The whole plane re-aims itself between observations, roughly every twenty minutes, which is what makes the survey's scale possible. Over its run it will take spectra of about 35 million galaxies and quasars across 14,000 square degrees of sky. Lawrence Berkeley National Laboratory leads the construction and operations, with a collaboration of several hundred researchers.
A spectrum gives a redshift, a redshift gives a distance, and millions of distances give a three-dimensional map of how fast the universe has been expanding at different times. That is the measurement.
Position in the Quest
DESI's third data release decides Midnight Eridanus, our first pre-registered mission. The prediction was deposited before the data exists, and we have committed to publishing the outcome either way.
What it has already found
Its 2024 and 2025 results reported a hint that dark energy may not be constant after all, worded carefully by the collaboration and still contested. Those results are what made this test worth registering.
- Deciding release DR3 cosmology, expected 2027, month not yet announced
- Data access Public releases, free to anyone
Vera C. Rubin Observatory
Simonyi Survey Telescope · Cerro Pachón, Chile Midnight Orion · COSMIC-007Bio
An 8.4-meter telescope built for one job: photographing the entire southern sky, over and over, for ten years. Its LSST Camera is the largest digital camera ever built, at 3,200 megapixels, and each exposure covers more than forty times the area of the full moon. The camera was lifted into place in March 2025 and the first images were released that June. Over the survey, the sky will be imaged roughly 800 times.
The result is not a picture but a movie, and a public alert stream of everything that moves or changes brightness. Where DESI measures expansion history, Rubin measures how structure grew, by tracking billions of galaxies.
Position in the Quest
Rubin's first data release decides Midnight Orion, our second pre-registered mission.
What it has already found
It is only beginning. Its first images already contained millions of previously uncatalogued objects, including thousands of new asteroids in a few nights of observation, which is a fair measure of how much of the sky nobody had looked at carefully.
- Deciding release DR1, estimated mid-2028
- Data access Public alert stream now, staged data releases after
Simons Observatory
Cerro Toco, Atacama Desert, Chile · 5,190 m Midnight Phoenix · COSMIC-008Bio
Four telescopes on a ridge three and a half miles above sea level, where the air is dry enough to see microwaves from the early universe: one 6-meter Large Aperture Telescope and three half-meter Small Aperture Telescopes, all with detectors cooled to near absolute zero. Two of the small telescopes began observing in late 2023 and the third in August 2024. The large telescope reached first light in February 2025, on an image of Mars. Further telescopes funded by Japan and the United Kingdom are being added.
What it measures is the polarization of the cosmic microwave background, the oldest light there is, to a precision nobody has reached before.
Position in the Quest
Simons Observatory results would decide Midnight Phoenix, and this is the one place on this page where the work is ours, not theirs. COSMIC-008 is not yet stated sharply enough to be settled cleanly either way. We have given ourselves until 31 December 2026 to fix that or withdraw the mission. If the instrument is ready before the prediction is, the prediction does not count.
- Deciding release First results, estimated early 2027
- Our deadline Sharpen or withdraw COSMIC-008 by 31 December 2026
Keeping it honest
What the institutional path is worth
The Quest is blunt about how hard the institutional path is on an independent researcher, and about what funding can do to a conclusion. None of that is an argument that the path does not work. It plainly does, and the three instruments above are the proof: no independent researcher was ever going to build them. Here is the ledger with both columns showing.
What it delivers
- The Higgs boson was proposed in 1964 and found at CERN in 2012. Forty-eight years of sustained institutional effort, and a prediction that could have been proven wrong.
- Einstein predicted gravitational waves in 1916. LIGO detected them in 2015. A century of patience, and then a signal matching the prediction almost exactly.
- These three instruments exist at all, and they publish their data to anyone, including people with no affiliation. That policy is what makes our missions possible.
Where it slips
- When 100 psychology studies were repeated in 2015, 97 of the originals had reported statistically significant results, and 36 percent of the replications did.
- In preclinical cancer biology, a project that repeated 50 experiments from 23 papers found 46 percent of effects replicated on more criteria than not, and replication effect sizes averaged 85 percent smaller than the originals.
- Industry-funded drug and device research is about twice as likely to report favorable results, and roughly two and a half times as likely to reach favorable conclusions.
Those two columns are not in contradiction. The difference between them is mostly the difference between a shared instrument and a private one. Where the equipment is expensive, the data is public and hundreds of people are looking at the same numbers, a wrong result gets caught. Where a study is cheap, small, run once and never repeated, it does not.
That is the useful lesson for us, and it cuts against our own convenience. Our advantage is that we test against public data from shared instruments, so anyone can check our arithmetic. Our exposure is that we are small, unfunded and want our own framework to be right. The protection is the same one the good column relies on: state the prediction before the data, publish the outcome either way, and leave the record where we cannot quietly edit it.
Why we work in the open
And sometimes the data is simply false
Everything above is about honest error and quiet bias. There is a third thing, and it has to be said plainly, because anyone who believes the process is self-correcting by nature will not understand why we bother publishing our predictions before we have the answers. Some published data was never measured at all.
The volume
More than 10,000 papers were retracted in 2023, a record. Over 8,000 of those came from journals owned by a single publisher, withdrawn over what it called systematic manipulation of the publication and peer-review process.
Behind much of it are paper mills: businesses that sell authorship on fabricated research. Estimates of how many submitted papers come from them run from 2 to 20 percent, which is a wide range because nobody knows the true figure.
Physics is not exempt
Bell Labs, 2002. An independent committee led by the physicist Malcolm Beasley examined 24 allegations against Jan Hendrik Schön, a rising star in semiconductor physics, and found falsification or fabrication in 16 of them. Twenty-eight papers were retracted. His doctorate was revoked in 2004.
Superconductivity, 2020 to 2024. Two papers claiming room-temperature superconductivity were published in Nature and both were retracted. An investigation commissioned by the University of Rochester at the request of the National Science Foundation, which Nature's news team found in court documents, reported data fabrication, falsification and plagiarism. The lead author denies any misconduct.
Notice what did not catch either case. Both sets of papers passed peer review at the most selective journals in the field, repeatedly, for years. What caught them was other groups trying to reproduce the results and being unable to, and independent researchers examining the published figures closely enough to see that they did not add up. The correction came from outside the gate, not from the gatekeeper.
That is our motivation, stated exactly. We are not working in public because we are unusually virtuous. We are working in public because it is the only check that has a record of working. So the prediction goes on a public, dated deposit before the data exists, the derivation is published so anyone can follow the arithmetic, and the outcome gets reported whichever way it falls. If we ever fabricate or fudge anything, the material for catching us is already in your hands.
It cuts the other way too, and this is the part worth holding on to: a field that publishes 10,000 retractions is a field that is looking. The fraud gets found because science keeps the receipts. A path that never retracts anything is not more honest. It is less examined.
Where to go next
Follow the trail
These instruments sit at the end of the road. The Quest Map shows the whole road, including the five paths an idea can travel and which of them is actually paid to find the truth. The Testing Schedule lists the registered missions and their dates, and Follow the Quest is how to get the result when it lands.
Sources for this page
- DESI instrument and survey design: NOIRLab and Lawrence Berkeley National Laboratory releases on the 5,000 fiber positioners and the Mayall Telescope; DESI survey operations papers on arXiv.
- Rubin Observatory: NSF-DOE Vera C. Rubin Observatory, LSST Camera installation March 2025 and First Look, June 2025.
- Simons Observatory: Simons Foundation, Large Aperture Telescope first light, March 2025; observatory status reports on the Small Aperture Telescopes.
- Open Science Collaboration, "Estimating the reproducibility of psychological science," Science 349 (2015). doi:10.1126/science.aac4716
- Errington, T. M. et al., "Investigating the replicability of preclinical cancer biology," eLife 10 (2021) e71601.
- Lundh, A. et al., "Industry sponsorship and research outcome," Cochrane Database of Systematic Reviews (2017). doi:10.1002/14651858.MR000033.pub3
- Van Noorden, R., "More than 10,000 research papers were retracted in 2023, a new record," Nature (2023).
- Schön case: Beasley committee report, Bell Labs, September 2002, reported in Physics Today and Science.
- Superconductivity case: Nature news investigation, 2024, on the University of Rochester research misconduct report, and the retraction notices for the 2020 and 2023 papers.