Eighty-seven Medicare specialty categories, drawn to true scale. Drag one line across them and count how many you are standing inside at once.
Everyone quotes the first half. Fortes fortuna iuvat — fortune favours the bold. Pliny the Elder said it on a warship in the ash of Vesuvius, with the helmsman telling him to turn back, and then he said the rest of it: Pomponianum pete. Head for Pomponianus. A friend. By name. At Stabiae, across the bay.
The reason the second half matters is not sentiment. It is navigation. A category has no coordinates. Here is what that looks like when you draw it.
Each horizontal bar is one specialty category — its true 10th-to-90th percentile of Medicare beneficiaries seen in the year, computed over all 1,235,757 individual billing clinicians in the country. The dots on each bar are twelve real clinicians drawn at random from that category. Drag the red line.
Put the line at 108 beneficiaries a year and it lands inside 75 of the 87 categories. Not near them — inside their middle eighty percent. A clinician who saw 108 Medicare patients last year is a perfectly ordinary member of nurse practitioner, internal medicine, family practice, emergency medicine, psychiatry, urology, cardiac surgery, hospice, and sixty-seven other things.
At the national median of 128 it is still 71 of 87. You have to push the line past 1,000 before the count drops into single figures, and past 4,000 before any single category is left alone. For almost the entire range where actual clinicians live, the specialty label does not narrow the field — it names a crowd.
Formally: sorting 1.24 million clinicians into 103 specialty categories accounts for 35.8% of the variance in how many patients they see and 50.6% of the variance in how many distinct codes they bill. Two-thirds of what makes a clinician’s year what it is happens inside the category. Picking a specialty is about a third of a decision. Most builders spend it like the whole thing.
The bars are honest; the dots are a sample, and the sample is worst exactly where you would trust it most. Each bar carries twelve randomly drawn clinicians. Take the median of those twelve and compare it to the category’s true median. Medical Toxicology has 36 members nationally — twelve of them land its median within 1.5%. Internal Medicine has 93,627 — twelve of them miss by 45%. Physician Assistant, 112,514 members, misses by 42%. Median error is 27.7% across the 27 largest categories and 19.5% across all 87.
That inverts the instinct. The big category is not easier to sample; it is harder, because being big is what let it get so internally wide in the first place. A dozen customer interviews characterises a tiny specialty well and a huge one badly, and the huge one is the one on the slide. Turn the dots off and on: the bars barely move, and the dots move a lot.
The bars themselves also flatter the categories. p10–p90 discards the outer twenty percent by construction — the true full range of every one of these categories is wider than what is drawn, in some cases by more than an order of magnitude. The overlap is an undercount.
Medicare Part B fee-for-service only. No Medicare Advantage, now roughly half of all beneficiaries; no commercial, no Medicaid. A clinician with a young commercial panel shows up here as a small bar-position because Medicare barely sees them, not because their year was small — and pediatrics, obstetrics and behavioural health are systematically understated for exactly that reason. Counts are distinct NPIs with at least eleven Medicare beneficiaries, so a part-timer and a full-timer are indistinguishable, which widens every bar. Specialty is CMS’s rendering provider type, one label per NPI, so anyone who does two jobs is filed under one. Categories with fewer than 30 clinicians are dropped; 16 in the source did not clear that bar.
The overlap count is also a property of a one-dimensional projection. Add a second axis and some of these categories separate cleanly — chiropractic bills a median of 2 distinct codes and interventional radiology 104, and no line on this chart can see that. The companion piece plots both axes at once, and the boxes still swallow each other.
None of which changes the shape of the answer. You are not going to find a specialty label narrow enough to build against. What you are going to find is a person.
Fortune favours the bold. It does not tell you where to sail. The second half of the sentence does, and the second half of the sentence is a name.