As NightWatchman product manager I spend a lot of time thinking about putting computers to sleep and waking them up.

Sleep saves a lot of electricity and its costs, and that’s good for the environment. But they need to be woken for updates, maintenance, and to be ready for users to work on them. All of that naturally leads to thoughts of computers dreaming.
Actually, the direct inspiration for those thoughts was my dog. I know she dreams because sometimes when she sleeps her legs twitch like she’s running and she’ll make muffled barks and growls. How endearing. It’s easy to believe dogs’ brains and consciousness are somewhat like ours given that they clearly dream. It’s a common observation amongst dog owners.

We could probably spend a lot of time talking about what dreaming is. But for now, we can hopefully agree it’s a state between sleeping and being awake.

Not fully off or fully on. It’s hard to say that computers as we now know they have such a state. The network interface card (NIC) listens for magic packets when the computer is otherwise sleeping, and the Intel AMT management capabilities are ready when we need them at any time. System-on-chip components can temporarily resume activities without waking the whole computer. So we do have some intermediate states on computers but it would be a stretch to call those dreaming.

What about future computers?

The growth in use of artificial intelligence would seem potentially related but AI, as it’s currently being promoted, is largely a cloud service so any related ‘dreaming’ would not be apparent to you and I. If some AI gets delegated to local computers AI sleeping might be more relevant, if there is such a thing. “Always Connected” computers seem more likely to have a sleeping mode of some sort – they’re not necessarily always powered on but they’re always available, so maybe there could be an intermediate ‘dreaming’ state.

What would dreaming mean for computers?

That’s a tricky question. People even debate what dreaming does for humans. One possibility is memory “garbage collection”. More prosaic possibilities would be more interesting but much more speculative. Would a ‘subconscious’ review of activities and memories be beneficial? Wild flights of fancy might inspire AI to have useful insights. There are likely lots of possibilities.
A future NightWatchman roadmap may well have dream management features. Maybe we’ll call it “Dream Police” in deference to the Cheap Trick album.