I had a habit, on the longhaul message circles catering to navs, programmers, and the like, of claiming that there are two steps to solving a problem using code. One of them’s easy, and one of them’s hard, and almost everyone flips them around and gets into trouble by doing so.

The first step is the hard step: figure out exactly what you’re trying to do. The second step? Making the computer or, in my case, magical computational substrate, do the thing you want to do? If you’ve figured out what you’re trying to do exactly enough, this step is easy.

What do I want to do? Most of all, I want to identify the Skills of the opponents we’ll be facing. It’s possible, but unlikely, that they’ll hold Skills in reserve, so I actually need to create a holistic model of the System connections, something that can identify even skills I’ve never seen used based on the interface between the soul and the structure that empowers them. It’s certain that some of the Skills are passive, and likely that some of those will not come into play in the two matches I’ll get to see, which adds another layer of challenge.

It takes a bit for me to explain my logic to Vonne, but I make the time. Where my conclusions differ from hers is mostly because of my looking at the situation through the lens of my companions; the ability to get an illusory overlay might be nice, but it will also be distracting, and there isn’t time for the others to retrain themselves to get used to that, much less actually make use of the information. All of that makes having some sort of live overlay trying to predict or even alert on peoples’ Skill usage somewhere between a dubious benefit and an active malus. Knowing what someone else’s Skills are, on the other hand, is the kind of cerebral evaluation that has been informing their engagements since they were children, though often their knowledge was limited to vague extrapolations and assumptions rather than knowing exactly what the very System had to say about those Skills.

Omniglot coming in with the last-second debris deflection into the capture net, yet again.

If it weren’t for us being in the Temple, this would be impossible, now that I can’t fork out an unlimited number of threads to, for example, recursively iterate over every possible set of phonemes and conceptual equivalents, defining that through Omniglot. I come up with seventeen different ways of using that basic strategy to solve the problem before I manage to stop thinking about it; I don’t regret what I did with that particular piece of bullshit, because figuring out a non-coercive party binding was both necessary at the time and a spectacularly important long-term investment going forwards, and I knew going into it that it was probably going to cost me that particular ability.

But we are in the Temple, and these opponents of ours are Imprints, people reduced to and inscribed into vast multi-dimensional duodeca-scale runic frameworks. I don’t have the slightest bit of access to those, of course, but the very fact that they exist means that instead of having a squishy organic soul interfacing into the System, it’s something with more structure, and whose complexity is about three orders of magnitude lower at a minimum, not accounting for any compression algorithms.

It’s a small edge, but it’s what I’ve got, so I’ll have to make it be enough.

Well, it’s all I thought I’d have, but I pretty quickly figure out another edge. There’s some sort of fit happening between my Visor’s computational capabilities and the System connections on the Imprint souls. My working hypothesis is that it’s a reflection of the fact that the Visor is going through the Earring, a reward from the Temple, or maybe a reflection of the fact that Sara rerouted so many of my own System connections into or onto or just to the Earring, and so the Temple-System and System-Temple relationship that I’m trying to figure out for these others just so happens to be something that the Visor-and-Earring, or Earring-empowered Visor, speaks natively. So to speak.

It’s a tenuous hypothesis, but any road you chart to starshine is the right road.

I’m on the clock, which I loathe. You’d think that I’d be used to it, given my profession, but you’d be wrong. I tended, in a typical jump, to take it easy, which was in hindsight absolutely unfair to the women who’d been my partners during those jumps even if I was still completing the jumps faster than most other navs would have. But it’s not something entirely unfamiliar to me, and I have always been decent enough at dealing with the pressure, so I… cope, really.

Tooling and utility functions practically dream themselves into existence. As with everything else, I can’t actually introspect with regards to the data transmission or storage formats that I’m trying to manipulate, but I don’t technically need to. The Visor can tell when I’m not trying to do that, which I can only chalk up to the usual inadequate handwave of it’s magic, even by comparison to the basically-magical Volitional Interfaces of my home, but at least it means that I can store the data, cross-check, munge, manipulate, and then do more checks, and ta-da, I’ve confirmed that—you know what, let me retrace a little.

People have styles and skills and preferences, leaving aside the Skills and whatnot that they are so known for. Sages has a guy who absolutely loves, and apparently excels phenomenally at, an archetype people call Stormlord, and his big thing is these flashy leaps of something lightning-related that let him reposition both defensively and offensively. He works with… not his brother, his brother actually has his own totally different team and they’ve run into each other in the arena before and gone after each other with a vicious glee, but a guy who people joke is his brother or his lover, though nobody makes both jokes at the same time, which seems polite. His partner tends to go for some sort of aggressive mobility, combined with something that can keep people away for long enough for the two of them to double-team whoever they both jump on, and then as long as the rest of the team can survive, it’s a 4-on-5.

They’re pretty known for being one of the teams that’s genuinely trying, when not everyone is, so suffice it to say that it’s not exactly a sure shot; time and again, something goes wrong, they can’t win the two-on-one, the team can’t hold three-on-four, or they can’t win the 5-on-4 after committing too much to the strategy. But it’s brought them close enough to victory time and again that they keep working on it, keep iterating, and keep hoping.

It’s not a shock at this point for me to realize that not everyone here wants to leave. There are a hundred thousand people who have lived longer lives in the Temple than they ever would have on the surface, and for a lot of them it’s not entirely objectionable; they work as background material or behind the scenes of a variety of scenarios, Lily’s Tournament in particular, and they live and love and run riddling contests or compete with no intention of winning and occasionally have children, which is possibly the strangest thing.

Anyway, Sages, also previously known as Wisdom and Joyful Children and a few other names, has been trying for a win pretty seriously, and they’ve more or less settled on a strategy. They mostly all have, except for Flight, who’re absolutely notorious for never doing the same thing twice and absolutely not my problem since if they make it through to the quarterfinals, which they’re favored to do over Epiphany, they’ll be fighting against Rei. So: keyword searches.

There’s no attempt, in the storage/data representations of the System connections and Skills, to obfuscate or hide or encrypt. There’s some sort of hashing going on at the actual soul interface, for anyone who’s got an actual one of those, yeah, but not for an Imprint. So, a keyword search, which is hilariously basic and yet the tooling around it takes me almost more time than we have available.

Storm, lightning, bolt, electric. Jaunt, surge, strike, maneuver, motion. I come up with a couple dozen of these for the Sages’ Stormlord, feed them all into Omniglot, and what comes out of it is more or less a gigantic set of string arrays, containing dozens and dozens of data representations of the abstract System- or Temple-mediated notion of language, some sort of bitwise encoding of the concepts themselves in all of their possibilities.

I can’t test them, though, because I haven’t seen Sages fight, and that should have been obvious at the start, but at least now I have all of the tooling and translation layers and whatnot in place, so it’s not like it was wasted work. I test on Houndmaster instead. Getting a hit on [Largest] on a whim was kind of shocking, but machete only gives me a partial match. I go through thirty variations on throw looking for the companion word and wind up picking up a complete match on [Hurl], which means I’m not looking for another throwing-keyword to go with machete.

[Triple Machete] winds up being Vonne’s suggestion, after she gets exasperated enough to make it, having clearly thought of it and then waited for about a kilosecond to see if I would get it, and then the next matches are starting up and it’s time to jump back into actually watching them, or at least to churning out lists of words that are plausible matches for what they’re packing.

So we settle ourselves down, stomachs mostly full of food I barely noticed and brains buzzing, to watch the combat begin anew.