FDA reversed its CRL pause in a week. That's not the part worth watching.

A few days ago I wrote that FDA had quietly stopped publishing its rejection letters.

A citizen petition had landed in April. At least one sponsor objected to public release of its Complete Response Letter. The agency paused the program.

Then, on Friday, FDA posted fourteen new CRLs anyway — including a July 9 letter to Hengrui and Elevar for a liver-cancer combination the agency has now turned down three times.

So the pause lasted about a week in practice.

I could write the transparency piece: is it fair to publish a company's rejection, does it help patients, does it arm competitors? Everyone is writing that piece. I think it is the less interesting one.

Here's the thing I'd actually pay attention to, and I hold it even though the policy fight is louder: whether or not FDA keeps publishing tomorrow's rejections, there were 455 Complete Response Letters in the public database when I checked, going back to 2012, and almost no one writing submissions is reading them.

I only looked because of the news. The letters are on openFDA, as an actual API, with redacted, extracted text of each letter — not just a scanned PDF, the text, searchable. You can pull every rejection in a therapeutic area. You can search the language of the deficiencies. I searched "narcolepsy" and got three letters back: the new Tris Pharma one, and before it Bioprojet and Merck. Three prior rejections in one indication, each one FDA telling a sponsor, in FDA's own words, exactly what wasn't good enough.

Think about what that is for someone drafting in the same space. A Complete Response Letter is the most direct thing the agency ever writes about why a submission failed. Not guidance about what they'd like in general — a specific finding about a specific dossier that didn't clear the bar. When a writer sits down to build the argument in a new submission, the questions that matter are: what has this division rejected before, on what grounds, and does my draft answer those grounds before a reviewer has to ask. That answer is partly public and has been for years. It just isn't on anyone's desk.

This is the same thing I keep coming back to. A submission isn't strong because it reads well. It's strong because it's built around the right argument, grounded in what actually happened — and part of "what actually happened" is what the agency has already said it won't accept. A drafting process that ignores the public record of rejections in its own field is leaving the clearest signal there is on the table.

The redactions are real — confidential and trade-secret material is stripped before these go public, so you get the agency's reasoning, not the sponsor's private data. The text is OCR'd off old scans and it's messy in places. It's NDA and BLA letters, not everything. None of that changes the basic point. The reasoning is there, in the agency's voice, and it's citable.

I build a tool in this space, so weigh that.

But the bias runs the boring way here: I would rather the useful resource stay obscure.

It will not.

The record of what FDA has rejected is one of the most useful things a regulatory writer can read. Not because every CRL is directly applicable. Not because redacted letters replace guidance. But because they show, in FDA's own words, what was not good enough.

Enough of that record has been public to matter while we have been arguing about whether it should be public at all.

The policy will keep swinging.

The better teams will already be reading.