Today’s workbench tool comes from oncology, but it travels well. Adaptive therapy treats cancer as an evolving population rather than a static target. The digest frames the idea through Robert Gatenby’s work: instead of using maximum-dose chemotherapy until resistant tumor cells dominate, clinicians may rotate or modulate treatment to keep drug-sensitive cells present enough to suppress resistant subclones.

The concept is strategically uncomfortable because it rejects the clean drama of total attack. In many systems, maximum pressure feels morally and operationally correct. But if the opponent adapts, maximum pressure can become a selection engine. It kills what is vulnerable and leaves what is hardest to control.

Cancer makes that logic brutally concrete. A tumor is not one uniform enemy. It is a changing population of cells with different traits. If treatment removes every sensitive cell it can reach, resistant cells may inherit the space, nutrients, and opportunity left behind. Adaptive therapy tries to manage that competition rather than pretend evolution has paused.

The digest says early prostate-cancer trials have shown patients remaining on treatment longer than with standard approaches. That claim should be handled carefully: adaptive therapy is an active research area, not a universal replacement for oncology protocols. Treatment decisions belong with clinicians and patients using disease-specific evidence. The workbench value here is the model of thought.

In cybersecurity, constant maximum blocking can teach attackers which probes work and push them toward stealthier routes. In fraud, rigid rules can train adversaries around visible thresholds. In regulation, blunt constraints can move risk into less visible channels. In organizations, punishing every failure can make people hide weak signals until they become incidents.

Adaptive restraint is not softness. It is control through feedback. It asks what behavior the intervention selects for, what resistant population it may create, and whether measured variation can preserve optionality. The point is not to do less. The point is to stop confusing intensity with effectiveness.

For founders and operators, this is a useful planning test. Before applying a policy, incentive, price increase, security rule, or performance metric, ask what will evolve in response. If the answer is a harder problem, the strongest move may be a timed, measured one.

The frontier keeps rediscovering biology’s old lesson: systems remember pressure. Good strategy accounts for the memory before it pulls the lever.