My question is, what methods of Infinite Prevention are in games that you explicitly disapprove of, which ones do you think are acceptable, and are there any other ways that you could maybe see a different IPS working (one that's totally different from your system)? Also, you mentioned that restands are usually not allowed cause they can cause other resets. What are some of the problem moves that tend to lead to infinites, if there are any?
Long story short - I hella disapprove of BB's time-based junk because then combos with long bounces or you being off by a frame or two can make it drop. Same-move-proration is bad because it's arbitrarily designer decided, and they can be wrong or make poor choices (same with Fatals).
Restands are often not allowed because designers seem to think that if you don't have restands then all you have to do is limit juggles and ensure that every ground combo ends fairly early with either a knockdown or a juggle. In practice this doesn't really work, though.
Acceptable-to-me infinite prevention methods are ones that:
1 - Actually work to prevent practical infinite combos. This should go without saying but lots of them fail the basic test. Guilty Gear has infinites even with extremely heavy-handed hitstun/pushback/gravity scaling, so does CvS2, etc etc. This doesn't even mean "prevent 100% combos", which is stricter, just "prevent a combo from going on forever given infinite opponent health".
2 - Are based around the specific combo you're doing, rather than a set limit that applies to all combos, so you are not just trying to find the easiest combo that reaches the same limit as all other combos (e.g. not Maximum Damage or time-based).
3 - Are based around WHICH hits happen rather than WHEN they happen, so that being late by a frame or two at a spot where there is still hitstun to spare doesn't break your combo later (e.g. not time-based at ALL).
4 - Do not have exceptions designed into them, because exceptions mean infinites (e.g. not MvC3 TACs or moves that just ignore hitstun scaling entirely, GG ignores ground combos, etc).
5 - Cover every possible hit scenario, not just some of them (OMF2097's pretty great juggle limit system completely ignores projectiles).
Bonus points for these features, which IMO really should be required:
6 - Not being dependent on the combo starter, whatever you determine to be the "starter" (i.e. not giving different results depending on whether you start with LP or LPx2 or j.H into LPx2).
7 - Not breaking multihit moves or canned followups (e.g. specials or supers not comboing into their final hits later in combos in GG).
8 - Not making the same chained or cancelled series of hits behave differently later on in a combo.
9 - PROVABLY preventing infinites in some sort of formalized way rather than hand-wavy "Yeah combos will eventually end" mumbo-jumbo.
More bonus points for varying levels of transparency to the people who are gonna play the game:
10 - Being understandable at all by dedicated players, rather than "just try the combo and see if it works".
11 - Being explainable at all without having to play the game and find out when it will stop your combo. (Not necessarily the same thing as "being understandable by players" - you can understand that MvC2 has undizzy and basically how it works, or that 3s has set hitstun values per juggle number, but you can't figure out when it will stop your combo since it is partially timing-based.)
12 - Being
easily explainable, at least in basic principles. "Don't start a chain with a move you ever hit with before", for example. Sure, you can stretch combos beyond that, but if you adhere to that in SG your combo will never be stopped.
13 - Displaying relevant info so that players can figure out if a combo
should work if they had better execution. (In most games with hitstun deterioration nothing tells you how long the hitstun of the move really was when you used it in your combo just now, so you can't really
know if the next move would connect if you were just a LITTLE faster...then again, most developers are against displaying frame data of any sort so this is unlikely to happen.)
And finally,
extreme bonus points for:
14 -
Enforcing some sort of fun frequency for two-player interactions, cuz that's sort of the goal of stopping infinites, just more focused: If you're gonna stop combos arbitrarily you might as well use that power for the benefit of the players. (For example MvC2's system prevents most practical forever-loop infinites but does nothing about repeating a would-be-stopped-later-on meterless loop combo for 100% damage, and MvC3's system does nothing at all about this point. GG's system is very good for this, minus the "practical infinites exist" part. And yes, SG's system did not initially do this at all.)
I tried very hard to make sure SG's system meets all this, and I initially failed at #14 because I didn't think it was important enough.
That alone was enough to make the game pretty derpy out of the gate.
(I think it's pretty decent now, but that's just my view. :^)
