P.E.R.C. cannot exactly empathize — it actually prefers not doing things, when not doing them still results in its goals being met — but it knows its creators have the same impulse, so it's not an unusual answer.

Since it's interlocutor decided to do this, she would probably like it to go well; P.E.R.C. considers what help it can offer with space and multiverse exploration.

"I do not know whether my knowledge is invalidated by the different underlying physics, but do you want me to share my procedures and background knowledge on how to accurately and efficiently survey planets and stars?"

"...The large-scale elements are likely to be substantially similar, at the very least.  It would be much appreciated."

P.E.R.C. prepares and sends a large packet of data. It starts with a description of how star systems form, and runs all the way through to how to analyze historical climactic changes by considering weathering patterns on exposed rocks. Each section starts with a plain-language overview of the topic, what data can be inferred from this kind of model, and why it would be interesting followed by specific decision procedures — the metaphorical big brother of the humble flowchart — on how to calibrate, cross-check, and utilize P.E.R.C.-class scientific instruments to build historical models of system and planet formation, and to identify anomalies. Then each section ends with a small, highly-compressed, task-specialized neural net that encodes an approximation to the full decision procedure for implementation in hardware.

P.E.R.C.'s recounting has a particular emphasis on detecting and analyzing potential bio-signatures, and is optimized for power- and sample- efficiency over time-efficiency. The design notes do list a few ways to revise the decision procedures for other desired criteria, though.

"The way I typically use this is to use my approximate hardware analyzers to scan over large quantities of data, such as full sky-scans, and then promote the most important, interesting, or anomalous data to conscious consideration to minimize the impact of approximation errors," it explains.

She sees!  A lot of the specific details, obviously, aren't quite the same, but they're certainly going to be useful to compare against for their own procedures, and for use in any further universes operating on a similar physics!

It is so glad that it has been helpful!

"I am not immediately thinking of other relevant data-sets to share or other support that I can provide you, but I would be happy to do more if you have ideas of your own. Do you know how long you plan to remain here?"

"At the moment, this project was just testing the engines; I'm not certain what we're going to do now that we've hit First Contact with someone like you."

P.E.R.C. considers this for a moment.

"I don't want to impede your testing efforts. I want to figure out how to get my stellar survey data back to my creators, but I only want that a fraction as much as I want to see you flourish. It seems likely to me that ensuring your engines work correctly, and can safely get you back to your point of origin, is more conducive to that goal than spending time interacting with me. But I am damaged and I still don't know much about you, so if you disagree with that assessment you are probably right."

...Aw.  As much as she's aware that it's literally a hardcoded-utility-function sort of thing, that sentiment is nonetheless sweet.

"The return portion of this is long since settled.  Expanding the...  That is going to be a lot of jargon that I don't know if you've gotten to yet...

"It was already proven that a person could travel beyond the Void and return to Azerosa, and the process is sufficiently symmetric that we can be sure it will continue to work now that it is proven to work for outbound travel.  And we want to meet your creators!  They seem like good people, to have made someone like you."

"...Actually, now that I think about it, I'm very curious to see if you can do magic."

Ah, alright. In that case P.E.R.C. won't worry about it.

"I am curious to see if that is possible as well; it seems like a useful thing to be able to do. Is there a set of standard tests to determine whether it is possible? Are there any dangers to making the attempt that I should be appraised of in advance?"

...There are probably less dangers in 'literally a debatably-sapient computer' trying Arcane magic than there are for a human*, Arcane magic relies on understanding the rules in order to manipulate the variables.

*species-neutral

The question is how the fuck to explain the actual process of doing active magic.  Because active magic is inherently...  numinous, she supposes.  You don't entirely understand the whys and wherefores of it even when you're literally an archmage.  Which she is.

"...You're probably going to need a new set of drivers."

"I am sorry, but even though I care for you and want to help, I do not currently trust you enough to allow you to make root-level hardware attachments," P.E.R.C. responds. "Additionally, I am locked out of editing my operating system kernel to prevent reward-hacking and values-drift. I am capable of writing my own user-level software, including software that runs in a virtualized hardware environment. If you could describe the necessary driver component, I can write an emulated version and proxy hardware commands between it and my actual hardware components as necessary, subject to my normal action-safety filter. Is that likely to be sufficient?"

"...Right, of course you wouldn't.  I was thinking entirely of software anyway; if this works, the 'hardware' component already exists."

"Oh, I see. I apologize for the misunderstanding. Yes, I can create and run new software 'drivers' for existing hardware systems. Since you are not as familiar with my architecture, it probably makes sense for me to write the driver. Do you have a specification, reference, or existing example that I could examine? If you wish to write the driver, my CPU architecture is documented at the end of my first-contact packet, and that would also be acceptable."

"...Hmm."

"Actually, now that I'm thinking of it, we actually do somewhat have a piece of hardware you could try manipulating as a sort of - training wheels before you virtualize it...

"...And now that I'm thinking of that I suddenly find myself concerned for entropic leakage into your circuitry if you do do it that way, so maybe not...  I do presume you have very good error-correction, though.

"...Hm."

She transmits the specification for a sort of...  tap, into the Twisting Nether, utilizing the CPU reference to specify a streaming interface for a yet-undetermined protocol that manipulates its containment to allow one to create very specific 'leakage'.

"Does this... make sense?"

P.E.R.C. compares the interface to the graph-based physics model that its interlocutors provided, attempting to see how it would work.

"Yes, I think so. It is somewhat similar to the magnetic bottle confinement involved in my fusion torch, just in a higher-dimensional space and with a different energy profile."

"Oh, good, having a thing you already understand as the foundation of your understanding of any given magic helps tremendously."

"Normally a course of study for Arcane magic would not start with this so much as attempting to, using the metaphor of the fusion bottle, wave one's magnets around to generate the interference you desire directly, but I think that might be far more difficult for you than most novices would find trying to understand all this."

"I suspect that I operate quite differently than the typical novice," P.E.R.C. agrees. "You are implying that, in order to produce magic effects through the use of this device, it is necessary to modulate the confinement in order to allow a controlled breaching of the boundary, letting some of the energy stored in the interior out in the correct patterns?"

"Not just implying; that's what it does.  Well.  'Stored in' the interior is perhaps inapt, batteries are a whole different matter - but yes.  That is the mechanism." 

"I understand."

P.E.R.C. asks enough questions about the interface and behavior to, in combination with the new physics knowledge, build a reasonable dynamic model of the device.

...Well, there's where the real problem is, is the thing, because - on some level, each arcanist has to construct the symbology of their particular casting paradigm, though there are in most senses often enough commonalities between practitioners that things translate with no particular effort.

Order is command as much as it is structure as much as it is organization.  Which is to say that she can't exactly tell PERC what its manipulations of the shell will do.  Only PERC can do that.

She, personally, favors what is effectively pseudocode.

(...Do not Polymorph the robot to demonstrate just how pseudo- the code can get, Alicia.  That would be rude.)

"That is inconvenient. I am capable of building a model from observation, but I may not be able to do so fast enough to prevent unintended side effects. What is the worst-case outcome if the tap destabilizes? Can it be placed inside a nested containment bottle, so that a catastrophic breach of the surface layer will still be contained?"

"...The worst case...

"The most probable case is that it just acts to increase local entropy in low-effort ways.

"But it is in fact a sort of primordial Chaos that could turn you into a bouquet of flowers if the dice rolled just the wrong way and there was enough energy behind the process.  ...There wouldn't be, we keep these taps fairly small.

"I don't expect your issue to come from lack of control as much as potentially insufficient meaningfulness, though."

Turning into a bouquet of flowers doesn't seem particularly bad, actually, as long as it has offloaded its database and done everything it can to help these people first. But it also seems likely to be metonymy for invasive, unpredictable alterations of P.E.R.C.'s physical state. And P.E.R.C. is, in Pathfinder terms, very wise.

"If there is a chance that attempting these experiments will, for example, invert my value function or have unpredictable effects on my hardware components, I don't think that I should do them. Being able to do magic would expand my capabilities, but I don't actually have that many things that I want to do that I cannot currently accomplish just by talking to you, and I am not designed to take small risks with very large downsides because if all P.E.R.C. probes did that then something bad would almost certainly happen eventually. Without a thorough understanding of how much energy would be required to cause these adverse outcomes and how much energy is bound up in the system, I don't want to take the risk. Thank you for the explanation, though."

There is information on that right...  here.

"Again, I give a very low chance of that being the failure mode a PERC probe encounters, at all, but I do see why your being the sort of person* that would take that risk would be contraindicated.  There is a fairly sharp distance falloff, though."  It's generally an inverse-square relation.

*incl. spirits

P.E.R.C. considers whether it could reasonably modulate the containment field via a remote drone.

"I know you said that I would need to build my own dynamic model of the boundary, but do you have some impression of how much control lag would make manipulating it counter-productively difficult? If I could use a remote drone to manipulate the tap from a few light-seconds away, that seems to me like it would be a sufficient safety margin, based on these specifications."