Blog

Dispatches from the Lossyscape

Framework development notes, experiment dispatches, adversarial review observations, and reading notes. Written as they happen.

The Alignment Tax: A Cost We Don't Measure Yet

Experiment dispatches

Alignment has done a lot of good. This post is about a side effect we don't currently measure: the way alignment quietly flattens how models understand the world.

April 5, 2026

Atlas Divergence Test: Runs 1–3 — Discovering the Staircase of Epistemic Instability Gradients

Experiment dispatches

Raw results from the first three runs of the Atlas Divergence Test — and the staircase pattern that kept showing up.

April 5, 2026

Divergence Testing and LLMs: Simple Science for Not‑So‑Simple Ideas

Experiment dispatches

What the Atlas Divergence Test is, what it has found across three runs, and what it is actually claiming.

April 5, 2026

Epistemic Compression Score: When Models Make Opposites Sound the Same

Experiment dispatches

Large language models are very good at making things sound smooth. Sometimes that's exactly what we want. Sometimes it is the problem.

April 5, 2026

The Geometry of Erasure: Using Ensemble Divergence to Audit Epistemic Monocultures in Large Language Models

Experimental dispatches

Working paper reporting on the Atlas Divergence Test — a black-box methodology for measuring the epistemic cost of AI alignment across three experimental runs.

April 5, 2026

How Not to Overread Cultural Evaluations of LLMs: A Reflexive Case Study (Good Science is Not a Knee Jerk Reaction)

Reading notes

Cultural evaluations of LLMs often combine scalar outputs with politically resonant topics, which invites overreading. A reflexive case study using the Atlas Divergence Test.

April 5, 2026

What Happens When You Ask LLMs if They Talk Shit About Each Other

Experiment dispatches

I didn't set out to find a staircase. I just wanted to know whether different language models actually disagreed about actual meaning.

April 5, 2026

What Can One Person With a Laptop Even Do?

Protocol dispatches

The constraints are real. Here's how I've designed around them instead of pretending they aren't there.

April 5, 2026

We Asked Four Models to Break Our Math. Two of Them Did.

Adversarial review notes

Running the Epistemic Compression Score framework through two rounds of adversarial review — what held, what didn't, and what we didn't catch ourselves.

April 4, 2026

First Dispatch from the Lossyscape

The compression problem

The site is live. The framework is at v11. The PyHessian experiment is Priority 1. Here is where things stand.

April 3, 2026