Blog — Andrea Moro

When brute force beats HNSW — a DuckDB VSS lesson

▸TL;DR

At small N, DuckDB's vectorised columnar scan beats HNSW because round-trip and planner overhead swamp the O(log N) saving; the index is dead weight until ~10k+ items.
Lesson: a vector index is a property of the query shapes the planner can rewrite, not of the column. Run `EXPLAIN ANALYZE` before trusting the 'index created successfully' message from DuckDB.

AI & LLM · 7 May 2026 · 10 min read

Table of contents

What is HNSW?
Why I added the index in the first place
The three queries
The benchmark
Why each non-winner loses
Why brute force wins this size class
What I changed
The takeaway

SEO & GEO 1 May 2026

The cost of falling in love with technology

Cypher is expressive, graphs fit SEO data, embedded options exist. None of it survived the question 'what happens when the user opens DevTools?'

SEO & GEO 22 Apr 2026

Shadow DOM — the panic room inside your web page

The Shadow DOM is a secret room inside your web page — with its own CSS, its own events, and the quiet ability to vanish from your SEO audits and every AI crawler that ever visits.

AI & LLM 17 Apr 2026

Italian NLP — entity weighting — what didn't work

Determining which entities a document fundamentally discusses is a deceptively simple problem. Here's what I tried for Italian — and why none of it worked.

AI & LLM 16 Apr 2026

A 2019 paper quietly described Topic Augmented Generation — and nobody noticed

A quietly overlooked EMNLP 2019 paper anticipated the idea now driving modern generative search. Six years later, it finally deserves a name.

AI & LLM 10 Apr 2026

Everyone was talking about MCP. Nobody built the right Python bridge

By August 2025, nine months after MCP's release, Ollama still had no native support. I built a working prototype in a weekend.

AI & LLM 10 Apr 2026

Your LLM visibility data is probably lying to you

Ask the same question ten times and you get ten slightly different answers. That's why your AI visibility dashboard might be misleading you.

Digital Marketing 7 Oct 2020

Inbound marketing: how to distill topics by narrowing the 5 Ws queries with Google Data Studio

Search Console queries are full of What, Where, When, How, Why — and every one is a content idea in disguise. Five steps in Google Data Studio, a parameter and a regex filter, and the 5Ws surface themselves.

SEO & GEO 10 Apr 2020

Enhance Screaming Frog's 404 errors report

Screaming Frog's 404 export is useful, but it's missing a summary and a way to subset. A small Python notebook with Pandas picks up the slack — and turns the report into something your content team will actually read.

Nix 2 Apr 2020

Logrotate for a Web Server: overview and examples

Server logs are priceless for audits — right up until they eat the disk. A minimal logrotate config, a dry-run test, and the four permission mistakes that will bite you on first roll.

Blog & notes