The Self-Spreading Cure Under Attack
Question. Does attacking the network make it stronger or weaker?
Analogy. It works like stress on metal: pound a ductile steel and each blow work-hardens it, leaving it stronger than before, but land the same blows on brittle iron and they drive a crack until it shatters, and which one happens is fixed by the nature of the metal, not the force of the strike. The study asks which kind this network is, whether adversarial pressure work-hardens it or fractures it, measured by adoption, retention, and detection before and after staged or real attacks.
What's at stake. The proposal's growth model is built on a striking claim: the cure spreads through the same channels the attack uses and is paid for by the pain the attack creates, so the harder the adversary pushes, the faster it propagates. That is an antifragility claim, the system gets stronger under stress, and it is load-bearing for the whole self-funding, self-spreading case. But antifragility is the exception, not the rule; most systems get weaker under attack. A hostile press campaign could fuel adoption or brand the network as dangerous; a weaponised betrayal could harden the network's vigilance or shatter the trust it runs on; an attempt to bait operators into overreach could be self-destruction for the adversary or for the operator. The proposal bets antifragile across the board and has not tested whether the network actually strengthens or fractures under real adversarial pressure.
The two answers it decides between. Either adversarial pressure fuels the network, hostile attention drives adoption, a survived betrayal hardens vigilance, and forcing the adversary into the slow human-asset game is a net cost to them, so the network ends each attack stronger (antifragile); or attacks fracture the trust the network runs on, a betrayal drives people out faster than vigilance hardens, and hostile press brands the network rather than spreading it, so the network ends each attack weaker (fragile). Measuring adoption, retention, and detection before and after staged or real adversarial events tells them apart.
What a null result would mean. If the network weakens under attack rather than strengthening, the self-spreading-cure model's core claim, that the attack fuels the cure, needs revision, and the growth and economics that depend on it must be re-based. That is a finding about the growth theory, not a sign the platform was built wrong, and it is consequential because the "harder they push, the faster it spreads" line is load-bearing.
Why this matters to defence. A defence that strengthens under attack is rare and valuable, and whether this one actually does is the difference between a self-sustaining civil-defence asset and one the first serious adversarial campaign breaks (DRDC Objective 6, the cognitive level; the evolution-of-science-and-technology focus area, a system that adapts to pressure). It changes a concrete decision: whether to rely on voluntary, attack-fuelled propagation as a deployment model, or to assume the network needs active protection from the adversarial pressure it will draw.
How we would run it. Measure the network's adoption, retention, and detection capability before and after adversarial events, using both naturally occurring ones, a hostile press episode, and controlled red-team ones, a staged betrayal or baiting attempt run with consent and ethics safeguards. The discriminating measure is the direction of change across event types: does the network end stronger or weaker, and which attack types fuel it versus fracture it. The cost-imposition logic is measured from the adversary side where possible, whether the attack forces the slow, attributable human-asset game. The only-us part is a real trust network with real members under real or realistic adversarial pressure, which a managed cohort cannot reproduce because the trust an attack tests is the trust the platform builds.
Earliest start. Stage 11: the study needs a real trust network with real members exposed to real or realistic adversarial pressure.