I have been thinking about the difference between human users and computational agents. There is a widespread pattern of enforcing that only human users may interact with a service, not computational agents. Running a user-installed extension is one example; additional examples are CAPTCHAs, account verification by phone number, or services which refuse to provide API access such as banks.

Sometimes the reason is to protect users (e.g. banks), sometimes to protect the service from abuse. But often the reason is to ensure value can be extracted from users (e.g. serving ad impressions) or the power that can be exerted over users. Regardless, I believe this comes down to power: computer scientists understand that acting through a computational agent is fundamentally more powerful than acting directly as a human user. Computational agents can operate at scale, they provide virtualization (e.g. throwaway identities), and thereby conserve the ultimate scarce resource: our attention. Many systems seem to have computational power asymmetry as their fundamental principle: as little human contact on the system's side as possible; as much human contact on the user's side as possible.

If there's any path ahead that avoids a totalitarian nightmare resulting from asymptotic concentration of power, it depends on cultivating a more widespread understanding of computational power and developing new social norms and policies. As a CS educator, this is one of my top priorities.