Defensive strategies

A long greenhouse in low afternoon sun, worked by many people at once. In the foreground a woman in an apron repots a single pink orchid, a watering can beside her lettered SAME PLANTS DIFFERENT DAYS and a crate at her knee reading One garden, one pot, one planet. Behind her a line of gardeners works along a shared bench of seed trays and labelled boxes, TOMATO and BASIL among them, under hanging signs reading DIFFERENT HANDS SAME HARVEST and SMALL SEEDS BIG FUTURES. A water channel runs the length of the floor, ranked young plants receding on either side of it, misting nozzles playing over the far benches and a supervisor walking the aisle with a clipboard. High in the roof a figure in a hard hat and harness turns a valve wheel on the pipe run, beneath a board listing WEATHER, LIGHT, HEAT, HUMIDITY and WIND. On the right a gardener kneels with a notebook and a soil probe by a sign reading SOIL MOISTURE, PH, EC, TEMP, a curtained bay glows magenta under its own lamp behind a card reading SOME THINGS GROW BETTER IN PRIVATE, and a cage trolley of potted plants waits on castors by the open door, labelled READY TO ROLL.

Defence at every level, whether the defender is a citizen, an NGO, a company, a research institution or a state, sits downstream of the same structural conditions: what the surrounding system rewards, who benefits from leaving it as it is, and why technically sound counsel keeps losing to the status quo. Name those first, or any advice is only motion.

Where the models stop fitting

The threats described in these models do not affect everyone equally. A civil society organisation operating in a country with functional judicial oversight faces a different landscape than one operating where the courts cannot be relied upon. A researcher at a well-resourced university has options that a freelance journalist does not. A citizen who is already a known activist is under different pressure than one who has no current profile but may acquire one.

Generalising across these situations produces recommendations that are accurate for some readers and irrelevant or even misleading for others. The alternative, producing bespoke guidance for every possible context, is not feasible.

The models were built with specific, recurring threat patterns in mind: commercial data extraction, nation-state surveillance, and partner abuse. The defensive guidance follows from those patterns. Where a situation diverges significantly from the framing, because the adversary is different, the jurisdiction unusual, or the threat model more acute, the framing bends to the situation, not the situation to the framing.

A general model read as precise guidance for a specific context is its own category of error.

Last reviewed: 2026-07-19.