Find any trial
Search by drug, disease, sponsor or NCT number to see whether its forecast is sealed.
How it works.
We forecast trials, keep score, and turn every miss into the next experiment.
Forecast first, then wait for the result.
Each forecast draws on earlier trials, the drug's biology and how often trials like it succeed. It is sealed the day it is made.
A miss shows where the model is wrong.
Every miss becomes an unknown: a link nobody has measured, a number that is only a guess, or studies that disagree.
The next experiment is the one that teaches most.
Experiments are ranked by how much each would settle, per dollar. Each result updates the model and re-ranks the rest.
What changes with a world model
Eight steps of science, as they work today and as a shared model would run them.
| Predicting what works | Most drugs that reach human trials fail, often because the biology was misunderstood. | Forecast every trial before it reads out, and keep score in public. |
| Reading the literature | Over a million biomedical papers a year. Nobody can read them all. | Every claim, number and method from every paper, trial and dataset, read and linked. |
| Knowing what's known | Each lab keeps its own mental model of a disease. | One shared model per disease, every number pooled from all studies, with honest uncertainty. |
| Picking the next experiment | Driven by what a lab can already do, what gets funded and what gets published. | Ranked by how much each experiment would teach, per dollar. |
| Finding what's unknown | Spotted informally, one scientist at a time. | Found automatically: links nobody has measured, numbers poorly pinned down, studies that conflict. |
| Running the experiment | Every lab has its own recipe, so results rarely combine. | Standard, machine-readable protocols that any qualified lab can run. |
| Updating what we know | Years, through papers, reviews and textbooks. | The model updates within minutes of a result and re-ranks what to test next. |
| Negative results | Mostly never published, so failures get repeated. | Every result counts. The model needs “ruled out” as much as “found”. |
Try the loop with a loaf of bread
The recipe promised a tall loaf. It came out dense. Test your way to the reason in a few clicks.
The mystery
The recipe promised the loaf would rise 2.0×. It rose 1.3×.
So 0.7× of rise is missing. Something is wrong, and we don't know what.
Missing rise explained so far0.0× of 0.7×
Next test
Check that the yeast is alive
Stir a spoonful of yeast into warm sugar water and see if it foams within 10 minutes.
Why this one next: If the answer is "yes, that's the problem", it could explain up to 0.8× of the missing rise, and it's the best value for the effort (free · 10 minutes).
Free · 10 minutesThe loop
Every test goes all the way round. Run the first one and watch each step fill in.
- 1. What we don't knowIs the yeast still alive?
- 2. The testCheck that the yeast is alive
- 3. The result
What the test shows.
- 4. The picture updates
Suspects are ruled out or confirmed; the gap is re-measured.
- 5. What to test next
Every remaining test is re-ranked.
What we don't know yet
Every possible cause, sorted by why it's uncertain. Each has a test that would settle it.
Never checked
Things that could matter but nobody has looked at.
- Test #1Is the yeast still alive?
The packet was opened 8 months ago, and nobody has checked it since.
Check that the yeast is alive · Free · 10 minutes - Test #5Does chlorine in tap water slow the yeast?
People argue about it, but nobody has checked with this water.
Bake two mini loaves: tap water vs bottled · An afternoon
Only a rough guess
Things we have a number for, but a shaky one, or from a single source.
- Test #2How hot does the oven really get?
The dial says 450°F, but home ovens often run 25–50° off. It's only a guess.
Check the oven with a thermometer · $10 · 1 day - Test #3Does this flour need more water?
A single blog post says so. No one else has tested it.
Try a slightly wetter dough · An afternoon
Sources disagree
Places where the advice points in different directions.
- Test #4Do you have to knead for 10 minutes?
One forum says knead for 10 minutes; another says no-knead bread rises just as well.
Knead one loaf, not the other · An afternoon
The picture: how bread rises
Everything bakers know, in one diagram. Side inputs are the suspects; the last one that changed is ringed.
suspect, untested ruled out a cause new idea
From bread to biology
OpenQuestion runs this same loop on open questions in human disease.
| The dense loaf | An open question | Why fat builds up in the liver in type 2 diabetes |
| Cookbooks, forums, other bakers' notes | Papers, trials, datasets | 29 papers and 7 dataset records |
| How bread rises | The causal model | Fat supply, fat made in the liver, fat export, body weight |
| “The loaf should rise 2×” | A prediction | “Pioglitazone should cut liver fat about 10%” |
| It rose 1.3× | A trial that disagrees | A trial measured 54% |
| Suspects | Knowledge gaps | Missing links, numbers poorly pinned down, studies that conflict |
| Yeast in sugar water, an oven thermometer | Lab experiments | Cell and worm experiments at qualified labs |
| “The kitchen is too cold” (new idea) | A proposed missing mechanism | Adiponectin makes the liver burn fat |
| A baker in Denver shares a result | Labs worldwide feeding one shared model | Every lab's result updates the same model |
| Bake again and check | Confirm a new prediction | Predict a trial before it reports |
Tips
Check a forecast yourself
Every forecast has a fingerprint. Recompute it when the trial reads out.Read the rules
Which trials count, what meeting the goal means, and what keeps the model honest.