Health moves at the speed of its slowest step.

I work where biology, data, and health systems meet, because the problems that interest me have never sat inside a single discipline. Medicine teaches you to think one patient at a time. Public health teaches you to think one population at a time. Management and economics describe the machinery meant to connect the two. Each is indispensable, and none is sufficient alone; scaling health takes all of them at once.

Health reaches people through a long chain of actions: discovery, evidence, approval, financing, supply, and the human friction at the end of it. Call the whole of that the health chain. The clinic and the pharmacy are where all of it finally meets a person, and most of what is possible there has been settled well before anyone arrives: by which questions were funded, what evidence was gathered, what was approved, and what a system can afford to offer.

Like any chain, it is governed by its rate-limiting step rather than by its strongest link. Adding capacity where things already move quickly changes little; the system improves only when the constraint does. So the useful question is not which part of health is most advanced, but which part is holding the rest back, and that can only be answered by looking at the whole.

One way to sort work is effort against effect. High effort for high effect is where the field rightly concentrates; low effort for low effect is maintenance; high effort for low effect is where projects go to die. The quadrant that interests me is the fourth, high effect for low effort. It is the hardest of the four to see, because attention tends to follow what is already visible.

Effort also tends to show diminishing returns. The more work an area has already absorbed, the less each additional unit tends to add. Not everywhere, and not always, but often enough to plan around. That has a consequence for the grid. A question that has already absorbed a great deal of effort needs more of it again to reach the same level of effect. An area still early on its curve returns more effect for less. The quadrant is not a fixed map; diminishing returns keeps redrawing it, and part of the work is noticing where things have moved.

This matters more now than it did. As capability arrives faster in biology and computation, the binding constraint shifts from what can be built to what we choose to point it at. Defining the problem is becoming the more important half of the work.

This blog is where I will bring up the parts I think need more attention, in the fields I have some experience in.