Pricing algorithms don't need to 'agree' to keep prices high. They just need to keep exploring.

Pricing algorithms in real markets do not stop exploring. They keep sampling to stay adaptive to demand shocks, competitor moves, and inventory changes. The classic question, did they settle on a collusive Nash equilibrium, assumes algorithms eventually stop and asks whether the resting point is illegal. The new Meylahn and Schäfer paper shows the resting point is the wrong target. Under constant exploration, the right question is whether cooperation is sustained in the time-averaged sense, and the paper derives a sharp boundary predicting when it will be.

That boundary matters because every regulator currently litigating algorithmic collusion, the FTC, the European Commission's DG COMP, and the UK Competition and Markets Authority, has been reasoning from a convergence frame. The paper does not say deployed pricing software is colluding today. It says the analytical unit for deciding whether it might be should be the long-run fraction of cooperative play, not a fixed strategy profile. A firm that keeps exploring can keep prices high without ever "agreeing" in the equilibrium sense the law has been built to catch.

The frame carries to the next case. Watch for audit trails that record exploration rates, sandbox tests that hold the exploration parameter fixed, and complaints grounded in sustained parallel moves rather than frozen price schedules. The prediction is portable: cooperation that survives constant exploration is the more durable threat, and it is detectable before any algorithm "converges."