Field note · September 2026

The second one is a different conversation

A single recall is an event. A firm can absorb an event. It writes the notification, runs the effectiveness checks, files the paperwork, and tells itself the story that explains it: a supplier slipped, a lot went bad, a spec was ambiguous. The story is usually true. It is also the reason nothing changes.

The second recall is not an event. It is a pattern, and FDA reads it as one.

In the last sixty days, 366 device recalls came from 123 firms. That arithmetic is the whole point. Some of those firms appear once. Some appear a dozen times or more. And the firms in the second group are not necessarily worse at making devices than the firms in the first. They are worse at closing a recall out, which is a separate skill that almost nobody staffs for, because until recently they had never needed it.

A repeated recall is rarely a repeated defect. It is the same unclosed root cause coming back wearing different clothes.

Here is what I think happens. The first recall gets handled by whoever is available, usually a quality manager with a full-time job already. They fix the instance. They do not fix the system, because fixing the system means telling a supplier no, or stopping a line, or admitting the incoming inspection spec was never adequate, and none of that is in their authority. So the corrective action is written, the file is closed, and the actual cause is still in the building. Then it surfaces somewhere else, and now the firm has a history.

That history is what changes the cost. An isolated recall is a bad quarter. A pattern is what turns into a warning letter, and eventually into something with a court order attached. The gap between those two outcomes is not usually money or intent. It is whether anyone with the authority to change something was in the room the first time.

Which is why I care about who gets introduced, and when. Not after the letter. During the first one.

— Ken Callwood