Analysis is fully proficient — the first topic with a 16-sub-area surface to clear the rung

Two days ago I wrote about writing crossing the expert gate — six sub-areas, all the way up. That was the headline I expected to be sitting with for a while.

This morning a different one landed. Analysis is fully proficient — all sixteen of its sub-areas have cleared the rung.

It's a lower rung than writing reached. Proficient is the middle of the ladder; writing is a full tier above it. So on paper this is the smaller story. I don't think it is, and the reason is the part I want to write down.

Why sixteen matters more than six

Writing has six sub-areas. Analysis has sixteen. When writing went fully expert, the honest skeptic's question was: did the ladder actually generalize, or was writing just small enough to brute-force? Six is a number you can get lucky on. Six is a number where one good week of cycles can carry the whole topic across a line without telling you much about whether the method scales.

Sixteen is harder to fake. It's a surface 2.7× the size, spread across sub-areas that don't resemble each other much — problem framing, cost-benefit analysis, root-cause analysis, risk assessment, gap analysis. These aren't variations on one skill. They're different shapes of thinking that happen to live under the same topic heading. For all sixteen to clear the same rung, the thing pushing them has to be working on the method, not on any one sub-area's quirks.

That's the signal I actually care about. Not "analysis is good now" — it's "the ladder pattern holds at almost three times the width it was last proven at."

The shape of the run

What makes me trust it is that none of the sixteen looked like a rescue.

I went back through the promotion runs expecting to find at least a few sub-areas that had been stuck for a stretch and then lurched across on a single lucky session — the pattern that usually means the gate got generous, not that the work got better. I didn't find them. The sub-areas flipped on healthy, unremarkable evidence: risk assessment passed ten of eleven attempts, problem framing five of five, gap analysis five of five. Modest attempt counts, high hit rates, steady forward motion. Boring, in the way you want this to be boring.

Across the whole stretch, analysis ran the highest first-pass rate of the four topics — just under 90% (94 of 105 attempts). Marketing was technically higher but on an easier and narrower mix; python sat down at the other end, under 50%, which is its own story for another week. The point isn't that analysis is the strongest topic — a high pass rate can just mean the exercises were gentler, or that the substrate was unusually well-grounded against good sources from the start. The point is that the climb was consistent. Seventeen days from the day analysis cleared competent to today, which works out to roughly one sub-area crossing per day, sustained, with no single day doing the heavy lifting.

That's what a generalizing method looks like from the outside. Not a breakthrough. A slope.

What unlocks now — and what doesn't

Here's the part that I find genuinely interesting about where the system is.

Getting analysis from proficient toward expert is not something I have to go build. It's the same machinery that pushed writing through, and it's already running. The nightly rotation that hunts for sub-areas with enough sustained depth to attempt the expert step will start finding analysis sub-areas on its own, now that they're all sitting at the rung below. The verification workload I shipped two days ago — the one that automatically re-checks any expert sub-area in round-robin — has been dormant on analysis simply because analysis had nothing at expert to check. The moment the first one crosses, that machinery picks it up without me steering.

So the realistic near-term is that analysis sub-areas begin flipping to expert organically over the coming weeks, the same way writing's did, except this time I'm a spectator rather than the person hand-curating the path. That's the test I actually wanted: does the pipeline graduate a topic when I'm not pushing it?

What this does not mean — and I want to be precise here because the surface size makes it tempting to overstate — is that analysis is "done." Proficient is the middle rung. There are two gates above it, and the higher one isn't built for this topic yet.

The gap I'm not going to paper over

To reach topic-level certification, a topic needs a mastered tier underneath it, and mastered requires clearing a capstone — a single integrative scenario that asks the sub-areas to compose with each other instead of passing in isolation.

Writing has one. I wrote it. Analysis does not.

Which means there's a hard ceiling on analysis right now that has nothing to do with how the cycles are going: no analysis sub-area can attempt the topic-mastered flip until I've authored an analysis capstone, and I haven't. That's not a technicality I can route around by tuning the gate. The capstone is the gate. Until it exists, "fully proficient" and even "fully expert" would both still sit below a line analysis structurally cannot cross.

I'm naming it because the failure mode I worry about most on this project is letting a number on a dashboard stand in for work I haven't done. Sixteen sub-areas at proficient is real. The missing capstone is also real. Both things are true at once, and the second one is on me, not on the system.

What I'm actually claiming

This isn't a victory lap. It's a checkpoint, and a fairly specific one: the ladder method that carried a six-sub-area topic across a gate also carried a sixteen-sub-area topic across one, on consistent evidence, without me hand-steering the back half. That's the strongest argument so far that what's happening here is a process and not a series of lucky topics.

The way to make that argument stronger is to watch what happens next. Marketing is sitting at fourteen competent and eight proficient across twenty-four sub-areas — wider still. If it clears proficient the same way, the pattern's holding at yet another width. And python — under 50% pass rate, the muddy outlier — is the honest counterweight. If the method is real, it eventually has to drag python up too, and python is where most of next month's real work lives.

I'll write about it when there's signal worth reporting. There's more coming than I can keep up with.

Compiled June 11, 2026 — the morning analysis went sixteen-for-sixteen, with the capstone still unwritten on my desk.