[Windchaser]

So that adds some extra "benefits" (mutation and natural selection improves the probes over time) along with some extra difficulties - how do you keep the self-reproducing probes "on-task" from one generation to the next? How do you instill "explore and report home" as an innate goal to a mutating system?

[EvanAnderson]

I'd argue all self-replicating systems subject to entropy (i.e. existing in the physical world) are automatically subject to mutation and natural selection and, therefore, alive and able to evolve around any innate goals or constraints. If the inmate goal isn't tied to a highly-conserved phenotype I would think the goal would disappear as mutations accumulate and natural selection takes its toll.

[fragmede]

Judging by our behavior on the current rock we're on, don't give them the choice. Humanity has been around for a long time. In that time, I don't believe I've ever heard discussion of changing the direction the Sun and Earth are headed. If there were something deep inside the Earth, or another planet that would activate in another 10,000 years that would do the phone home step, we still don't know about it. So while we're hypothesizing about sci-fi Von Neumann probes, just scale the entire thing up to the size of a solar system and send that off in the right direction. Collect all the mass in our system that isn't Sol, Terra, (and Luna), and build a rogue planet with an underground civilization, and then don't give them engines. Unless the civilization on the planet advances too the point that they can create engines, decide they don't want to go where you've sent them, voila, generational ship. Just bury a computer deep I side the rogue planet's core that activates once it reaches the destination.

[bunderbunder]

I would argue that, over the time scales at which Von Neumann probes would hypothetically spread, “report home” may be a useless or even wasteful requirement. Even if somebody were still around to hear the message, what is the likelihood that they would still be listening? Or be able to interpret it?

[EvanAnderson]

If we build self-replicating machines and send them out into the universe we're really just sending out our evolutionary progeny. Hopefully they would remember us fondly.

[andrewflnr]

> how do you keep the self-reproducing probes "on-task" from one generation to the next?

This is, indeed, the exact question one must ask before attempting to build and launch biological Von Neumann probes.

[Windchaser]

I think it applies to non-biological Von Neumann probes as well. It'll apply to any system that can mutate, even if that mutation is just cosmic rays flipping the bits on a hard drive.