"...As long as the drone itself isn't laggy, I certainly see no reason why doing it that way shouldn't work.  I'd recommend buffering your inputs."  She'd offer a piece of comms hardware, to cut down on the lag, but that's probably contraindicated.

That seems reasonable. P.E.R.C. will, with the assistance of the villarosans, coordinate on a set of orbital transfers that puts the tap quite far away, with a simple radio relay holding one of P.E.R.C.'s spare field-reprogrammable gate arrays, and the local planet between the tap and the two ships. As P.E.R.C. refines its dynamical model of the tap, it will push updates to the 'reflexes' embedded in the FPGA, and receive measurements back. Those measurements go through an error-correction code and then through its normal networking hardware, which will strip out any messages that are still malformed. Its network parser is formally verified, so it shouldn't be vulnerable to malformed messages in any case.

(...It's not going to explode quite that badly, but better safe than sorry, she supposes.)

The specifications that came with the tap suggested that increasing local entropy was one of the easiest things to do; it has a precision infra-red thermometer in the probe. It attempts to modulate the containment on the tap in order to increase the heat of a metal ball that the thermometer is measuring.

It is not going to be particularly dismayed if the first few attempts don't work. In comparison to any human mage, it is incredibly patient and methodical about working through the implied state space.

Does it manage to heat the metal?

...Yes.  It is, however, noticing a lot of other distortion in its testing platform's sensors.

Is this distortion, perhaps, model-able as an additional error term drawn from a common distribution, such as normal error or lognormal error? Because it has dedicated hardware for handling that very common case. If the distortion is more complicated, it may have to think for a bit in order to figure out what the pattern behind it is.

"I have successfully created heat energy," P.E.R.C. reports. "I am currently trying to narrow the model to produce heat more efficiently."

It's random.

It is, however, predictably random, as much of an oxymoron as that sounds - what's happening is random, but the magnitude of the randomness is not.

"Personally I'm more of a frost mage myself," she comments idly.  "If you do it right it's self-reinforcing, and I can just imagine how much use you could get out of enchanted superconductors.  Or just an infinite heat sink."

It's random.

Yes, but what kind of random. Is it statistically independent? Is it time-varying? Just because something is random is no excuse not to categorize it in an orderly system.

"It would be useful to be able to get closer to stars without having to bleed off my superconducting energy storage," P.E.R.C. agrees. "Since cold represents a decrease in local entropy, I surmised that it would be harder to use the tap for that purpose. Is that inference incorrect?"

It's statistically independent, obviously.  Except when it isn't!

"It depends on the caster, really.  I think you'd have better luck with Frost than Fire."

P.E.R.C. doesn't really have a good model of why she would think that, but she is the domain expert. It wraps up its current experimentation and switches to trying to cool the metal. Perhaps if it inverts the helicity of the induced vortex here?

Sure!

Great. It will report this success, and practice heating and cooling the sphere until just before the orbit will take them out from behind the shielding planet. At that point, they practice bringing the tap back to quiescence.

"I believe I have a working model of how to manipulate temperature, but I'm not sure how that generalizes. Do you have advice on expanding beyond that single manifestation?"

These taps fail shut, but that is a useful skill to have.

"Well.  The basics of Arcane magic as a school of magic are that of a directed application of force.  Heat, you may recall, is indistinguishable from kinetic energy on a molecular level.  So that's the easiest place to expand to, I think."

That does make sense. P.E.R.C. constructs hypothetical models, refines its 'reflexes' based on the gathered data, and continues chatting via radio while it waits for the orbits to line up again.

Hm, she wonders if...

Well, it's too late for you to make a reflector or lens for this pass, but the next one, perhaps.

P.E.R.C. is very patient. It is accomplishing all of its goals right now: helping people to flourish and explore space, and actively working on putting them in contact with its creators. It is happy to take as many orbital cycles of experimentation as are needed to get this right.

This next pass sees it figure out how to apply reactionless forces to the test sphere. That will be useful for orbital maneuvering once it is confident that the technique is safe.

"Where I'd take you from kinetic energy, now that you're confident in that, is probably gravity, actually.  It's of perhaps greater utility, given that last I heard space itself is probably the only thing that can travel faster than light."

P.E.R.C. takes a moment to process that, since its interlocutor had to travel faster than light to get here.

"Does your experimental drive work by manipulating gravity or space-time? I have some plans for a hypothetical faster-than-light drive that my creators designed but couldn't build since they could not produce contained areas of negative local curvature. That drive does not involve disjoint translocation like your drive does, however. Instead, it just moves an area of space along a spacelike curve, rotating the contained subspace so that the curve becomes timelike in the local reference frame. Is that something that you believe would be possible to emulate using arcane magic? And are you interested in a copy of the plans?"

No, no, she didn't travel faster than light to get here, specifically, she travelled along a metaphorical fourth axis of direction at presumptively subluminal speeds.

Teleportation is a thing she can do, though.

Oh, and: She would like those plans!  They're fairly sure they worked it out properly but something to crosscheck their work against would be useful.

P.E.R.C. sends over a translated description of the Alcubierre metric, along with its notes on how it believes it would be easiest to construct one using magic. Normally, it wouldn't be possible to construct in flat space without either extensive preparation or tachyonic particles, but it is pretty sure magic can be used to work around that. Which incidentally also means that the described solution is usable as a time machine.