I’m Chris — a software developer based in London, currently CTO at Searchable. This is my personal blog, where I typically write about programming, search systems and technology.
A correction to what I wrote about TurboQuant in May. A new paper from Mariano Tepper and Ted Willke makes the case that the rotation in front of a scalar quantiser should be learned from your data, and that learning it is cheap enough that I no longer think calibration cost belongs in the argument.
AWS cut S3 Vectors query pricing by up to 80% at scale. It is a real win for your bill and says nothing about your moat - because the price cut lands on the one slice of retrieval where quality barely lives.
Hornet held the retriever, model and corpus fixed and swapped the harness from one tool call per turn to one Python program per turn. Recall went up 65%, tokens down 51%. Strip the new vocabulary and the win isn't Python - it's that retrieval was finally allowed to fan out.
Dot product, cosine, Euclidean, Manhattan and Hamming - what each one actually measures, why most of them collapse into the same ranking once your vectors are normalised, and the handful of mistakes that bite in practice.
A small, self-contained experiment on why single-vector retrieval breaks on compound, high-intent queries - and how late interaction keeps two facets intact where a dense embedding averages them away.
