Biology as engineering

Every advanced biological capability is worth less per unit of effort than the things this site covers in tier 0. A society that skips to the sophisticated end has misallocated badly, and the misallocation is invisible because the sophisticated end is more interesting.

The floor, before this step gets resources: water and sanitation for everyone, vaccination coverage above 90% with a working cold chain, attended childbirth, and adequate nutrition. Measure them; do not assume them. For measles, the gap between 80% and 95% coverage is the difference between recurring outbreaks and none.

Every biological capability is downstream of being able to see what is there, and the order is fixed because each layer calibrates the next: microscopy → culture → staining → biochemical assay → immunoassay → sequencing.

Do not skip to the end. A sequencer with no culture capability behind it produces data nobody can check, and an assay with no reference material produces numbers that drift. Keep reference strains, reference sera and reference standards, and re-run them regularly — a measurement you cannot reproduce next year is not a measurement.

Knowing what someone has, quickly and cheaply, determines whether the treatment you already have gets used correctly. A settlement with good diagnostics and three drugs outperforms one with poor diagnostics and thirty.

Prioritise tests that are cheap, fast, usable without power, and change a decision: is this fever malaria, is this diarrhoea cholera, is this pneumonia bacterial, is this water contaminated. Pair every test with a written action for each result — a test whose positive and negative lead to the same action is not worth running.

Marker-assisted selection and, later, direct editing compress a decade of crop improvement into a season — drought tolerance, disease resistance, nutrition.

Keep it anchored to field trials rather than to the laboratory claim: replicated plots, multiple sites, multiple seasons, and comparison against the best local variety rather than against an average. A variety that wins in one field in one year has demonstrated nothing. And keep the parent material alive: every new variety is a narrowing, and the landrace it came from is the only way back. See securing seed.

Synthesis, purification, animal and then human trials, manufacture to a standard, cold chain, distribution and regulation — a therapeutic programme fails at whichever of those is weakest, and usually it is manufacture or trials rather than discovery.

Trials before deployment, always: controlled, with a comparison group, pre-registered as to what counts as success, and reported whether the result is favourable or not. See controlled comparison. Batch manufacture to a written specification, with a retained sample and traceable numbering, so a bad batch is recallable. Publish the harms.

The same tools that make a vaccine make a pathogen, and the barrier to misuse falls every year as the technique gets cheaper. Capability outrunning governance is the historically standard failure and the one with the worst tail.

In place before the work starts: containment standards matched to the agent, with physical separation, controlled access and trained staff; screening of every synthesis order against a list of sequences of concern; publication norms that weigh method detail against misuse; and independent oversight with the authority to stop work, funded separately from the work it oversees and reporting publicly. Oversight that can be overruled by the researcher is decoration.

Uniformity raises yield and raises correlated risk at the same time: a monoculture is one pathogen away from a famine, and the historical examples are not marginal — the potato blight of the 1840s and the 1970 southern corn leaf blight both ran through genetically near-identical crops.

So: seed banks in at least two locations, landraces and wild relatives maintained as living populations rather than only as frozen samples, rare livestock breeds kept in working herds, and a written register of what is held where. Regenerate stored seed on a schedule before viability falls — see securing seed.

Biology locked behind patents and trade secrets advances more slowly and distributes worse. This is a choice, and the first time round it was frequently made badly: essential drugs priced beyond the people who needed them, and diagnostic sequences owned to the point where a second laboratory could not verify a test.

So decide it deliberately and write it down: methods, sequences, reference strains and trial data published, with a stated exception for the narrow set of sequences the governance rules restrict on safety grounds. Publish negative results and failed approaches, which is where most of the wasted effort is. The rule that survives contact with money is the one written before there is any.