The future of biology.

Everything below it either settles or repeats. Life compounds.

Rocks erode, stars burn out, and chemical reactions run until they reach balance. Even complex physical patterns, like crystals or hurricanes, form and dissolve without keeping anything they learned. Life is different: it copies itself, varies, and keeps what works, so each generation starts from the last one’s best. Technology and culture compound too, but only because living things make them.

Physics explains how matter behaves. Biology explains how matter learns.

Biology is often treated as a catalog of special cases, with physics as the home of deep theory. But the principles that govern how living systems adapt, regulate and evolve are as fundamental as any law of motion, and far less worked out. What has held biology back isn’t a lack of theory; it’s the cost of testing it. Cheap, fast experiments will make biology the most interesting theoretical science there is.

Everything else got a factory. Biology still has a pipette.

Semiconductors, cars and software moved from craft to automated production decades ago. Most biology experiments are still run by trained people moving small volumes of liquid between tubes. That makes experiments slow, costly and hard to reproduce, because no two pairs of hands work the same way. The next leap in biology will look less like a single discovery and more like a factory.

The next decade belongs to whoever makes testing cheap.

Sequencing, DNA synthesis and computational design have made it easy to propose what to try next. Testing those ideas still takes weeks of bench work per round. When ideas outnumber experiments, the experiment sets the pace of progress. Whoever cuts the cost and time of a single experiment speeds up everything that depends on it.

Truth still comes from the bench.

A model trained on published data can rank a million candidates, but it can’t tell you which ones are real. Biology is noisy, and much of what matters never made it into the literature. Models get better when they are wired to a lab that tests their guesses and feeds the results back. The winning systems will close that loop, not just widen the top of the funnel.

Four billion years of search, running in parallel for free.

Every population of cells is searching for solutions, all at once, every generation. That search tries possibilities no designer would think of, and keeps what works without needing to know why. Rational design asks us to understand a system before we can improve it. Evolution improves it first and leaves the understanding for later.

They’ll be made by machines, not scraped from papers.

Most biological data was collected for one paper, in one lab, under conditions no one else can repeat. It is scattered and inconsistent, and the failed experiments are usually missing. The data that will train the next generation of models has to be generated on purpose, at scale, by machines that run the same protocol every time. Whoever builds that pipeline will hold the most valuable data in the field.

Cells copy themselves, repair themselves and adapt. Factories don’t.

Factories need supply chains, tooling and constant maintenance. Cells assemble complex molecules from simple inputs, correct their own errors and make more of themselves. We already grow insulin, vaccines and industrial enzymes this way. Over time, more of what we manufacture, from medicine to materials, will be grown rather than built.

Small, automated and everywhere beats big, staffed and central.

The traditional lab is a building full of people, equipment that sits idle overnight, and results that depend on who was on shift. Compact automated instruments can run continuously, identically and wherever they are needed. Ten small machines in ten places beat one large facility that everything has to be shipped to. Distributed labs will do for biology what personal computers did for computing.

Code needs a programmer to improve. Life improves itself. Engineer that, and you get technology that compounds.

Software changed the world because it could be copied and improved at almost no cost. But code only gets better when someone rewrites it. Living systems improve themselves, grow from simple inputs and repair their own damage. Once we can engineer biology as reliably as we write code, we get technology that keeps getting better on its own.