Paul Nurse’s What Is Life? is 160 pages, five ideas, and one contradiction that runs through the whole thing. I’m reading it now, and the useful part is not the survey. It’s watching a Nobel laureate state the orthodox position so cleanly that the seams show.

The five ideas: the cell, the gene, evolution by natural selection, life as chemistry, life as information. If you work in this field, none of it is new. That’s exactly why it’s worth reading. A short book by someone who built the consensus is the cheapest way to see where the consensus is thin.

The definition has three rules. Two of them exclude viruses. The other one includes them.

In the final chapter Nurse rejects MRS GREN — movement, respiration, sensitivity, growth, reproduction, excretion, nutrition — as a checklist that describes life without explaining it. He builds his own: something is alive if it can evolve (reproduces, has heredity, shows variation), if it is a bounded physical entity communicating with its environment, and if it is a chemical, physical, and informational machine that makes its own metabolism.

Then he gets to viruses and calls them “a gray area.” That’s the tell.

Viruses vary, they are selected, they evolve. Condition one puts them in. They have no metabolism of their own; they are bounded particles with no interior economy. Conditions two and three put them out. The definition does not settle the case he raises — it picks a side by weighting metabolism over evolution, while insisting evolution is the spine.

Both camps in the literature agree the question is definitional, which is the more interesting finding. Moreira and López-García, in Nature Reviews Microbiology, argue for exclusion and are precise about why the question keeps reopening: the idea of viruses as living had a renaissance “owing to the discovery of unusually large and complex viruses that possess typical cellular genes.” Their counter is that giant-virus genes resolve to multiple, unrelated origins — chimeric acquisition, not shared ancestry.

Koonin and Starokadomskyy go further and call the whole thing “effectively without substance because the answer depends entirely on the definition of life.” Their move is to replace alive with replicator, which admits viruses, plasmids and transposons on equal footing.

Both are right. Nurse’s book raises the hardest case for its own definition, then declines to defend it. That’s not a scandal. It’s a demonstration that “life” is a term of convenience with no natural kind behind it — and that a smart person can choose a boundary without noticing they chose it.

Five ideas is a genealogy, not a description.

Here’s the structural problem. The five are the ideas that made molecular biology possible. They are not the ideas that describe life. Nothing in the list explains a fertilised egg becoming an organism.

No Waddington. No canalisation, no epigenetic landscape, no evo-devo. Development is the most impressive thing biology explains and it is entirely absent — not by oversight, but because it is the domain where the gene-centric story is weakest. A chapter on development would have broken the book’s frame, so there is no chapter on development.

Ecology is out too, along with symbiosis and holobionts. So is the organism — the level Nurse actually worked at for fifty years. Genes and cells appear. The thing that constrains both appears only as a recipient.

And the chapter that breaks the frame is his own.

Chapter five is titled “Life as Information: Working as a Whole.” Read that title again. The whole is the finding. Information is the metaphor.

Nurse is too good a physiologist to write a pure blueprint chapter, so he keeps pulling toward systems integration while retaining the storage metaphor wholesale. That tension is the most productive thing in the book.

The rigorous version of the argument he is circling belongs to Denis Noble: biological relativity, no privileged level of causation, the organism constraining its chemistry — including its genes — to serve the whole. Philip Ball’s How Life Works is the popular demolition of the same target, and it lands the point flatly: DNA is neither a code nor a blueprint, whatever we teach students.

The strongest argument in the book is the one that got older.

Nurse’s best idea is buried near the end. Everything alive shares the same chemistry, the same genetic code, the same energy currency, the same chirality. Four billion years of selection should have eroded that uniformity. It didn’t. So life arose here once, and the sample size is one.

The 2021 evidence was Weiss et al.: 355 protein families tracing to LUCA, an anaerobic, H₂-dependent, thermophilic organism with a Wood–Ljungdahl pathway, living in an iron-rich hydrothermal setting. A competent microbe, not a soup.

Then in 2024 Moody et al. dated LUCA to roughly 4.2 billion years ago, reconstructed a genome of about 2,500 protein-coding genes across 2.5 megabases, and described it as “a complex anaerobic acetogen that lived within a pre-existing ecosystem.”

That last clause is the sharpest thing in the recent literature, and the book predates it. LUCA was not the first life. It had contemporaries and a food web. So:

  • The uniformity argument gets stronger — life had already been diversifying before LUCA and still only one lineage survived to be seen.
  • But we are not observing the origin of life. We are observing the survivor of an event that had already produced ecology.
  • And 4.2 billion years is a lower bound on the origin, which puts it uncomfortably close to the end of heavy bombardment. A short window between “habitable” and “inhabited” argues that life is chemically easy — the opposite of the cosmic-rarity conclusion the n=1 argument is usually deployed to support.

Nurse’s version predates the evidence that made the question sharp again. Both readings are live. He doesn’t get to have just one.

Nurse vs. Nurse.

The same year as the book, he published a Nature column arguing that biology is drowning in data and starving for knowledge — Sydney Brenner’s line, which Nurse then reuses as his own complaint about research talks where speakers “unleash a tsunami of data.” His conclusion: data collection is necessary but insufficient, and ideas that might turn out wrong are worth more than another dataset.

Then he wrote a book organised around five settled ideas.

That’s a genre constraint, not hypocrisy. But put the two documents side by side and you have a cleaner statement of the field’s problem than either alone: the shortage isn’t ideas. It’s people willing to be wrong in public. The column says so. The book demonstrates the opposite.

Bottom line.

Read it as the consensus stated well, by a participant, with the hedges left in. Then read Ball and Noble next to it, while the framing is still warm. The five ideas are right about what made molecular biology work. They are incomplete about what life is, and the gap is exactly where the interesting work now sits — development, the organism, and the question of whether the virtual-cell programme inheriting the information metaphor is building on a blueprint that was never there.

I’ll take the book’s own test and apply it: if it can’t account for an organism assembling itself, it hasn’t answered its title.