A: “Hello, and welcome back to our show! Ari here, with–”

L: “Lex”

A: “… So you were just going to say your name?”

L: “Yes.”

A: “And nothing else?”

L: “Yep.”

A: “And we’re here once again, this time with the one topic that absolutely everyone who plays Arrows will deal with at some point, and that nobody understands!”

L: “Nobody is pretty strong language. Unfortunately, it’s also true. So remember that what you’re getting here isn’t actually perfect information, and is instead the best, most effective theories in a lot of cases. We may eventually come back and update this episode, or put out a new one correcting misconceptions if more information comes out.”

A: “Which does occasionally happen, though not often. We’ve had multiple years to go through all of the abilities that came out with the game, but there’s still fairly limited categorization and not every ability is accessible for every person, wildly increasing the difficulty of cataloging them.”

L: “Part of which is a self-selection bias that’s difficult to quantify. We know, for example, that not every passive ability, of which there must be thousands, is available to every person. Instead, each player only has access to about a thousand.”

A: “But. And this is a big ‘but’ too. If you go looking for any specific passive, it will be in the list. So it’s more like each player has access to every passive, but is only presented with about a thousand that they’d be fairly likely to use.”

L: “We’re fairly certain there’s a tag system with them tied to a person’s personality and stats, but that’s dystopian enough that we try not to think about it.”

A: “They’ve got the right of it, there. It’s not like a company as well-funded as Granduon couldn’t find out basically all that information anyways, but the fact that they don’t bother with a quiz or anything and just pull it from my brain? Heebie-jeebies.”

L: “I don’t really care, personally.”

A: “You just said you try not to think about it.”

L: “Because it’s a waste of time. Plus, it makes the game more fun.”

A: “A-ny-ways. The reason that they brought up self-selection bias has to do with that personality selection. If we know that a passive exists–”

L: “It’s either because someone appeared using it, or, more often, some massive nerd copied down their entire list and added it to the community’s data table.”

A: “Both of us have done that, by the way. We’re both massive nerds.”

L: “You. Say. That. With. Pride.”

A: “But that also means that the majority of our data comes from people who are the type to copy down a thousand abilities and enter them into a sheet. We already know it biases the data, from appearance rates, but we aren’t sure by how much.”

L: “That said, it’s not like someone’s likely to appear out of the woodwork with something that the general population isn’t prepared for or that carries them to victory because they’re just that edg-[cough] I mean awesome. Awesome.”

A: “Granduon’s AI wouldn’t let that happen. We think. It’s hard to say, because we only have second-order information on that particular beast.”

L: “But that’s for later.”

A: “Yep. Because after you choose a passive, next up is your four abilities.”

L: “And the mess that entails.”

A: “Lex is being dramatic, but for a reason: unlike with the passive, Granduon doesn’t bother to protect you much. Instead, they provide a fairly robust categorization and search system and throw you to the wolves.”

L: “The name of the game here is ‘Types’ and ‘Effects’”

A: “‘Types’ here refer to the general styling of whatever the base effect is. This is where you get your ‘Fire’, ‘Water’, ‘Wind’, ‘Void’, ‘Creation’, and similar, as well as subcategorizations like ‘True Creation’ and ‘Psychic Creation’.

L: “As far as we’re aware, the largest number of Types on an ability right now is five.”

A: “On the other end of things, we have the ‘Effects’ categories, which mostly refer to how whatever the Type is, is used. Here’s where you find ‘Damage’, ‘Movement’, ‘Teleportation’, ‘Reconstitution’, ‘Spell’, and their subcategories like ‘Direct-cast Spell’ and ‘Passive Generation Spell’.”

L: “Passives are technically also categorized this way, or at least we have reason to believe as much, but they’re unlisted. You can do somewhat of a reverse search for the text of the passive, but you can’t just outright look at the tags. Because Granduon are monsters who ruin my fun.”

A: “That’s… probably not why, Lex.”

L: “I’m choosing to believe that’s why until they tell me otherwise.”

A: “But they won’t. You’re just one random person.”

L: “So I still believe it.”

A: “I… Sure. Fine. Okay. You know what? That’s fine.”

L: “So, Penultimates and Ultimates. The biggest determiner to how some ability sets play, and the only part of an ability set that you don’t get to directly decide for yourself.”

A: “The place where you put yourself at the mercy of the AI’s whims.”

L: “It’s usually annoyingly accurate, though. People have been trying to break it, but haven’t been able to manage it.”

A: “There are a couple of guarantees with what gets selected but they’re fairly minimal.”

L: “First: except in edge cases that get particularly strange ultimates, like our sister’s, every tag that’s represented in your basic ability set will show up at least once in the combination of the two abilities. That doesn’t mean that you’ll get exactly what you want, though. If you took water, wind, fire, and sand ‘bolts’ with a movement ability that lets you choose to transport yourself a quarter of the distance travelled by spells, it’s just as likely that the penultimate eats every one of the element tags involved there and also gets the ‘Movement’ and ‘Reconstitution’ tags, while the ultimate is just one, big, completely unaligned, bolt-styled spell as you are to get something more useful.”

A: “That specifically isn’t always useful, but the exceptions basically all come from something we like to call ‘pre-centralized’ sets. In the case of our sister’s, for example, there’s an extremely strong through-line of self- and attack-buffing abilities, which got represented in the penultimate and ultimate instead of the tag-cloud that makes up the abilities more directly.”

L: “Pre-centralized sets aren’t particularly well-known, and basically everyone wants to ignore them. I love them. They’re so messy and I live for that weirdness.”

A: “Interestingly, they also tend to be… more powerful than usual, but with conditionals, and massive downsides. If you need half of the structures or more on your side to be destroyed before you can use your ultimate, I can basically guarantee that that’s the ultimate of a pre-centralized set.”

L: “If you need an ally’s permission to use the ability, it’s a pre-centralized set.”

A: “Assume that these rules are somewhat relaxed for the pre-centralized sets in general, except this next one. Second: The ultimate and penultimate are the AI’s primary balancing force for any ability set– a less powerful, less synergistic ability set will get significantly stronger ultimates and penultimates. It’s almost never worth it.”

L: “She’s putting that kindly. Let me do the opposite. If you’re an overconfident idiot who makes a willfully bad ability set, you’re locking yourself into a ridiculously high-risk, moderate-high reward playstyle. The AI doesn’t care that you’re awful and only hit half of the death-bolts you can cast in a game. It only cares about how many death-bolts you can cast. If you want strong ultimates that are good, you want to build sets that are on the weaker end, but synergistic.”

A: “The ultimates you get will be somewhat weaker, but also are guaranteed to scale in a predictable way that’s likely to work with the rest of your kit. That’s more of an items discussion, though, so for now suffice it to say that that’s very important.”

L: “Thirdly and lastly: You penultimate and ultimate will have fairly similar effect ranges to your basic abilities. There’s an exception here with teleports, which occasionally have ranges that are instead proportional in the distance per cooldown second, usually half or a third depending on its power.”

A: “These rules also make it pretty hard to really mess up, even for newbies, luckily. So if you want to get out there and test something, just do it! You only need to wait three months to reset an ability set, and you have three of them to play around with. So, Lex, sum up the advice?”

L: “Pay attention to your ability strength and balance your abilities with what you’d like to get for ultimates. Build something synergistic, if you can, because that will tend to make your set more effective overall than one that’s not by balancing ease of use with theoretical maximum effectiveness, which usually, but not always, has more of an effect on the way the AI balances. If you want something truly wild for an ultimate, put together a set that has an obvious binding throughline that isn’t a direct tag. Don’t take abilities that are difficult or impossible to use with your statline.”

A: “So that’s what we’ve got for today! Thanks for sticking with us, and hope to see you next time!”