Editorial writing

Posts

Long-form pieces from the Infrahuman Laboratory. Where a subject is worth an essay, it lives here. Weekly cadence when full.

  1. Post · 1 October 2026 · 9 min read

    The install button

    How Chrome's PWA install icon can download a native Windows or macOS app installer - and why it beats a plain download page. Our pattern for turning the browser's own install affordance into the primary download channel for a desktop LLM app.

  2. Post · 23 September 2026 · 9 min read

    The loopback bridge

    How abliteration.org tells you which models will actually run on your machine, without sending a byte to us. A note on the local HTTP endpoint the Windows app opens on 127.0.0.1, and the CORS-scoped catalog that reads it.

  3. Post · 19 September 2026 · 9 min read

    The year abliteration stopped being a scalpel

    In 2024 the field thought abliteration was surgical. Wollschläger's cones, Piras's SOM manifold, Winninger's RFM-AGOP, and the 2026 Not-a-Scalpel study each moved the picture in the same direction. Two years on, the operation is not the clean cut the community described.

  4. Post · 18 September 2026 · 5 min read

    A magazine, and why we are opening one

    The wiki holds the reference, the reports carry the year. The magazine is where the field is worked out in real time, in our voice, on the record.

Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Open in Abliteration