We publish the conclusion.
Not the paper trail.
Subtl is an intelligence editor. It reads the technology press without stopping, watches for the moment independent sources start saying the same thing, and writes that up before it is obvious.
The position
Most of what reaches you has been optimised to be read, not to be true. It arrives early because someone benefits from you seeing it early, or late because it took that long to become safe to say.
Subtl sits between those. It waits for corroboration, then moves. The interesting window is narrow — after a thing is established, before it is priced in — and almost everything published lands outside it.
The method, in outline
A wide, continuously maintained intake of primary technology writing. Extraction down to claims. Clustering across publications until convergence is measurable rather than felt. A judgement about whether the convergence means anything.
The outline is the whole of what we publish about the method. The intake list, the weighting, and the thresholds are not disclosed — they are the product, and they stop working the moment they are public.
What we withhold
We do not name the publications we read. Not on a topic page, not in a byline, not in the markup, not in a directory. Anyone can subscribe to a list of newsletters; the value is in which ones are weighted, when, and against what — and a source list hands that over for free.
This is a stated editorial position, not an oversight, and it has a cost we accept: you cannot audit us by following a link back.
What we expose instead
Every topic states how many independent sources converged on it. Where they agree and where they split is written out rather than averaged away. Direct quotes are attributed to the person on record — concealing who said a thing would corroborate nothing.
Confidence is labelled as Subtl’s own inference, with the basis it was derived from shown alongside it. Where a claim is ours rather than reported, it says so. You are told what kind of statement you are reading, always.
Who reads it
Operators who have to commit before the picture is complete: founders choosing what to build, investors deciding what has become interesting, and the people inside larger companies whose job is to notice early.
It is not a newsletter, and it does not want your attention daily. It wants to be correct on the handful of days that matter.
“Everyone has access to the same information. Almost no one has done the reading.”