Dogs know who to obey. Household robot appliances, so far, don’t. Several schemes have been proposed for sending signals between such agents but I have not heard a non-gullible scheme yet. How should suitably suspicious appliances be convinced to obey?

I am thinking of stationary agents like hot tub thermometers, TV controls, PCs, light switches and coffee pots. I will assume that most of these have no significant user interface, perhaps none. I also assume tamper-proof appliances.

In one scheme I imagine an easily carried wand with dominion over a set of appliances. An agent fresh from the factory is susceptible to the first wand that attracts its attention. If the wand gets close enough to tap the appliance then it bonds to the wand. Afterwards the new appliance is under the wand’s dominion. If the wand breaks or is misplaced then a new fresh wand is bought at the store and attached to a trusted computer with a secret protected like a private key. The new wand then supplants the old. I have no good story yet for a stolen wand.

Part of the appliance behavior is that it will not obey a new master until it has been released by the old master. (There is a “release” button on the wand.) The wand might have a simple indicator lamp indicating acceptance or rejection by the appliance. This is subject to confusion about which appliance the indication is for. After release it would be wise to keep the appliance in a potato chip bag to protect it from other wands, a bit like protecting RAM chips.

This scheme may be part of a solution.
The WPS button addresses this question.