◕ The Durability Curve The Instrument Room · Issue 73 · 133 releases tracked
AI is everywhere. The interesting stuff is underneath.
Find which part of your work still has a job when AI gets cheap. No-login instruments you run on your own position, and structural essays you can use the same day.
69 essays, dated · 7 instruments · 600+ readers · Claim Ledger →
Every release adds noise.
Underneath it, value follows a curve.
Output commoditises. Architecture compounds.
Run it on your own position
My advantage runs through
Awaiting a layer
—
Half-life of the advantage, before it has to be rebuilt
Watch for
Pick the layer your advantage actually runs through, then answer the one question that decides which side of it you are on.
Station I
Start Here
One short essay hands you the lens, then routes you by the problem you actually have.
IndexStation II
The Bench
Walkthroughs you can run this afternoon: gates and audits your agent cannot talk its way past.
Index
Station III
The Rack
Runnable instruments, no login. Each one reads only what you give it.
- 01 The Two-Rate Diagnostic Your AI edge against its layer’s clock. LIVE
- 02 The Multi-Agent Decision Whether a flat loop beats the fleet. LIVE
- 03 The Marathon Calculator What a finished job costs once retries are counted. LIVE
- 04 The Metric Validity Audit How your number lies, and what to do. LIVE
- 05 The Potemkin Map What your system only appears to do. LIVE
- 06 The Structure Spotter Whether your belief is a measurement artefact. LIVE
- 07 The Shape Test Whether your curve is really compounding. LIVE
Station IV
The Charts
Structural essays on where the margin migrates, argued so you can check them against the next release cycle.
-
FLAGSHIP, STANDALONE 2026·09·14 You Can Win the Wrong Game for Years You Can Win the Wrong Game for Years -
THE BLUEPRINT 2026·09·09 The Planes That Didn't Come Back Success stories and the antiques that outlasted the rest. Your evidence was filtered by what survived to reach you. -
PROOF & TRUST 2026·09·04 Your Verification Budget Is Backwards You can't check everything your agents produce, so you're already triaging. Most of us spend the effort where checking is easy, which is rarely where a mistake is expensive. -
THE BLUEPRINT 2026·09·02 The Extra Lane Fills Itself Wider roads, faster teams, more appointments. Add room to a queue that turned people away and demand rushes back. -
THE BLUEPRINT 2026·09·01 The Option You Were Never Meant to Buy The weak rival, the overpriced tier, the house shown first. The option you never choose can change what you pick. -
THE RUNBOOK 2026·08·31 The Skill That Never Fired A Claude skill can fail two ways: its instructions are wrong, or Claude never picks it. Anthropic's own tooling scores whether one skill fires. This tests it against the neighbour that can quietly win its requests, with a small routing eval you run yourself. -
WALKTHROUGHS 2026·08·30 The Graph Looks Right. The Merge Is Where It Breaks. A multi-agent graph can lose a verdict with no error and nothing turning red. The bug is one line in the state schema. -
PROOF & TRUST 2026·08·28 Everyone Got Safer. That's the Problem. Safety has two numbers: how often each system fails, and whether they fail together. The field has spent years driving the first one down while almost no dashboard reports the second.
Station V
The Chart
You have been looking at it the whole time. The field behind this page is the chart: the two curves the whole publication argues.
Interactive: trace any lab’s releases along the curve.
Station VI · The Log
Get the next dated read when your layer’s clock moves.
Several essays a week. The week’s main piece lands Tuesday, 10am UK. The instruments need no address.
Join 600+ readers · Unsubscribe anytime
Every essay and all 7 instruments are complete. No opt-in.
Latest You Can Win the Wrong Game for Years·The Planes That Didn't Come Back