Option Select - KI and general FGC discussion

Continuing the discussion from What is DP?:

I was always too much of a casual fighting game player to really understand a lot of the jargon and lingo surrounding the FGC, but this is definitely different than what I thought an Option Select was!

I thought it meant you input the up/down but then have enough time to press the P/K button while you try to see what the opponent is doing; from what you’re saying you press BOTH P and K at the same time and the game just decides which is the “better” input?

Would anyone care to enlighten me on why this isn’t as “free” as I’m thinking it is? Is it a bug that just happens to end up in a lot of games, or something that’s commonly balanced into various fighting games for a particular reason or purpose? On paper, it reads as though it’s something you should be be doing almost all the time because the input is guaranteed to give you the best response. I’m wondering if OS’ might be the fighter equivalent of how the quick-scope and weaponslide/wallbounce started off as glitches but were later made intentional because the pro scenes of the respective titles liked them.

@Infilament or @TheKeits if you have a few moments, could you shed some KI-specific light on the purpose/ideology of Option Selects? I’d love to have a really high-knowledge take alongside what the general community provides.

We had a 200-post thread about this not all that long ago, but the gist of it is that

  • Yes, option selects are about contriving the right sequence of garbage to input in a given situation to force the game to interpret the input in different ways depending on how the situation plays out, to attenuate the need for a good read or reaction to the opponent’s behaviour.
  • Some option selects are better than others: some will beat two of your opponent’s best options outright and lose to another three, some will beat three of your opponent’s options and be okay against another whilst losing hard to yet another, and so on. Option selects are almost never a “win button”: different option selects have different “coverage”, you get from an option select whatever you can wring out of the game systems in that situation.
  • Most importantly: option selects are inevitable, and cannot be removed from fighting games entirely. Introducing purported fixes will often just introduce other, unforeseen option selects, and likely also destroy the “feel” of the game. Designers can and do attempt to clamp down on the most egregiously degenerate option selects, but degenerate tech doesn’t need to be an option select to be clamped down upon, so this isn’t special treatment. Good combat design acknowledges, accounts for, and even embraces option selects.

wrt the example you quoted, though, I’m not sure that’s actually a thing that works.

I’ll admit, I was so concerned about not off-roading the thread my quote came from that I didn’t think to search for existing threads about Option Selects when I created the linked topic from @FallofSeraphs76’s post. I’ll have to skim through that thread, so thanks for the link!

The thing that I’m not understanding (and may not until I’ve read through that thread) is how the system ends up forcing a different event based on what the opponent is doing. I only learned about OS within the last couple of months, so there’s still a lot I don’t know about what they do or why they happen.

1 Like

I wish I could get my thoughts together to give you a good, concise explanation of an option select.

I’ll let Justin Wong show it to you:

1 Like

An example might help.

Say you just blocked Jago’s double roundhouse, so you’re -3 (so pressing a button may not be a good idea), and you don’t know whether what’s coming next is another attack (which you’d be okay with blocking) or a throw (which blocking isn’t going to do so well against). Now, recall that you have ~7 frames after a throw connects within which you still perform a throw tech. That’s not enough to allow you to react to the throw with a throw tech (if you’re trying to reaction-tech throws, you’re probably just getting thrown all the time).

But what if you opt to attempt to predict the throw and tech late, say, ~10 frames into when you expect the throw to be active, if that’s what the Jago player chooses to go with? Well, if Jago tried to throw, then you end up teching it. On the other hand, if Jago goes for the frame trap instead, then you’re still not seeing the move come out in time to know you’re not getting thrown, so you still wind up inputting the LP+LK to tech the throw that didn’t come. However, by the time you’re inputting the LP+LK, that frame trap has already come out, you’ve blocked it, so you’re in blockstun and your input is conveniently ignored by the system. So if he throws, you tech; if he frame traps, you block. Same input, protects you from two different lines of offense.

Of course, there are ways of beating late teching. The Jago player could jump, for example, which wins because you’ll whiff a throw when you input that LP+LK, and Jago will just land on you with a jHP before the throw whiff recovers. It’s also worth noting that blocking a frame trap isn’t amazing, because Jago could easily follow it up with another one. But late teching is still a pretty important technique at high levels of play.

Another potential example of a simple option select more specific to KI is inputting a special move input immediately after a first special move is input, for example on Spinal doing light soul sword then medium boneshaker. If the soul sword misses or is blocked then nothing happens, but if you hit then you will transition straight into a medium linker.

In that same thread, I posted a video from a youtuber named VesperArcade. His focus on option selects is really in Street Fighter 4, but the core of the Option Select mechanic is ultimately the same for KI. It’s all about the input buffer for specials and the game playing out inputs based on how your opponent reacts. Even if the video uses SF4, the same principles can be applied to KI. It also has some really good examples to show you what is happening.

They are sort of complicated, but no where near a win all situation some people were afraid they where, as there isn’t an OS that’s a cover all in the game. Some lose to invincible reversals, some lose to backdashes, some lose to neutral jumps, it’s all about your opponent’s reaction.

I have to ask why this is relevant? Bugs are also inevitable and cannot be removed from fighting games. Does that mean that developers shouldn’t remove them? Of course not… And yet any time OS gets brought up, this line of reasoning also inevitably gets brought up.

If you are philosophically not opposed to them or even more strongly philosophically for them, that is one thing, and I would gladly be interested in hearing why.

But the argument that “they exist so deal with it” is… well, frankly it is dumb. Moreover, there are degrees. SFV and SG have both largely removed OS. Do they still exist? Sure. Are there any super egregious ones? Not really. So if you are opposed to them (design-wise), then their removal is something worth pursuing… even if OSs “will always exist”.


The way the system forces something is as follows:

The system responds predictably in various scenarios so you can use that to your advantage. For example, if you whiff a normal, you cannot (typically) special cancel that normal, however if you hit your opponent (including block), you can. So if you do a normal with a special input, you’ll do nothing after your normal on whiff, but you’ll throw out your special on hit or block.

Another OS in a lot of games is block low, high, low fairly quickly. This leads to you blocking a low attack, and a high attack (that has a little bit slower start-up).

So to be perfectly clear, you don’t just do two things and the game optimizes it for you. Instead, you do two (or more) very specific things that you know in advance have desirable outcomes. Not just everything can be an OS.

Here’s the thing, though; in theory, you can have a bug-free piece of software. It’s incredibly hard, but not theoretically impossible.

When we say it is impossible to get rid of option selects, we don’t mean it in a practical sense, as in, “you can try but you’ll always miss some.” We mean it is literally impossible by the very nature of how fighting games work.

You seem to understand OSes based on the rest of your post, so I’ll make this comment; there are times when you will press jab in a situation where the game is not able to accept that input (you are knocked down, blocking, doing a combo, you’re whiffing a move, etc etc). This is fundamental to fighting games, right? This situation will never go away. The question is, what do you do with that input? Do you ignore it? Play it back later? Something else? All of these are “option selects” in their own way, and clever people will find ways to use that to their advantage in a match. So, the concept of an option select is inherently necessary to building a fighting game engine, and it’s different than trying to squash out bugs or other undesired behavior.

Now, of course, some games have option selects that make the game worse. You can try to code your game engine in such a way to remove these “bad” option selects (and KI has done this, with their combo assist stuff). But usually these types of option selects are the result of a poorly thought out game system or mechanic, and they are a very specific “class” of OSes. The fundamental concept of OS is still necessary.

Just as an aside for other people reading this, this specific OS is important enough that is has a name, called “fuzzy guarding”. You take advantage of the fact that a mixup’s different options come at you staggered, slightly delayed, so you try to input the blocks/avoidance for ALL of them in rapid succession. As another example, you can do this with TJ’s powerline mixups as well; you can block, then jump a few frames later to block powerline/spin fist but jump to avoid command grab.

It’s a hard OS to input correctly, though, because the timing is very character specific and it loses if the opponent delays their inputs at all (whether accidentally or on purpose).

1 Like

I understand that it is literally impossible, but in many cases that is a mere curiosity. There are accepted OS by even the most ardent OS-haters… buffered specials for example.

The reason I call it out specifically is because it is often times used to shut down any “remove this OS” type of conversation. There is quite a large difference between a buffered special and for example the MKX OSs that would literally allow you to control which special move came out on hit vs block and seeped into virtually every normal in the game. The existence of the former does not necessitate the latter.

I understand that OSs will always exist, but my design philosophy would like each known OS to be evaluated for acceptability. If it is something that limits large portions of the game (like how useless throws mostly were in SFIV), then it needs removed (mechanics permitting of course… but one would hope the developers would have the foresight to have reasonable mechanics… looking at you MKX).

This is fair. This means that OSes that are nice as a “quality of life” thing (such as buffered specials, safe jumps, late throw teching) are good, because changing the system to remove these will a) make the game way more frustrating to play and b) no doubt bring in way more game-destroying OSes than these. But a throw leading to a safe jump unblockable that has OSes designed to cover your escape options and leads back into itself is maybe a bit much.

Eh, don’t be so sure about this. In that other thread on OSes linked above, we had a few people say that buffered specials were bad for the game.

I’d be curious how they would plan on solving that while still keeping intentional specials on block/hit. I guess there are extremists on all sides of issues.

Some people proposed a few solutions that would have ended up making the game much worse in other areas. But in general, they seemed to have a philosophical disagreement with the concept of OSes, so because this was an OS, it just fell in that camp automatically. They wanted it removed “somehow” without really thinking of how to go about doing it, or the ramifications of doing it.

Actually, I think SG successfully solved it by making normals cancelable on whiff.

For my money, that is less desirable because nothing feels shittier than making your opponent whiff an unsafe normal only to be hit by their canceled DP… or worse, whiff a normal into a whiffed DP into a successful super… ugh.

I am generally pretty hostile toward OSs, but I definitely have a list of OSs that I either don’t mind or are so minor that it would be too risky to remove them.

On top of Infil’s already great answer (for the curious: Rice’s theorem is the bedrock of the theory which tells us that it is extremely hard to detect bugs in software systems, but it says nothing about the actual existence of correct software systems), I think it should be directly noted that bugs can generally be removed once discovered, which means the “cannot be removed from fighting games” applies in a fundamentally different sense to option selects: once discovered, removing an option select usually entails fundamentally redesigning the game.

I mean, this is in part what Infil was getting at, anyway, but it can’t be overstated.

I’ve never really thought of myself as having a “strong” philosophical view on option selects, but okay: the way I see it, when you set a bunch of constraints in place in a system that people care about (like a fighting game), inevitably someone will come along and demonstrate some counter-intuitive implications of those constraints that you never expected to be there. Hell, as a logician at heart, those counter-intuitive consequences are kinda what I live for. I’ll take the consequences with the premise, and if I feel like the consequences are evidence that the constraints are bad then I’ll look to reengineer the constraints; otherwise I’ll embrace the consequences, and probably even think less of people who refuse to.

In fighting games, that means that the sub-reactable window + how you deal with inputs that can’t be parsed as a move, begets option selects. I do strongly believe in playing the entire game, not some idealized “pure” game in my head where people only do “fair” things, because the only way that strategic lines of play gain any meaning is in being gauged relative to optimal play. So I embrace option selects.

But this wasn’t the conversation we were having. We’re literally having one of those Fighting Games 101 lessons, we’re on the lesson where the students have just discovered that option selects are a thing and some of them are instinctively revolting at the very concept.

But if you were worried about a slippery slope of sorts – someone using my post to justify keeping something really dirty in the game down the road – note that I did put a caveat in the bullet-point in question noting that it is worthwhile to clamp down upon overpowered option selects that degenerate the game.

1 Like