Okay, comeback mechanics are contentious in a lot of games because they have weird effects on the game.
Comeback mechanics are basically when, as you get closer to losing, you’re given more power to hurt the other player. Some examples of this are Ultras in SF4 (gain ultra meter as you lose health), V-Gauge in SFV (same), X-Factor in Marvel Versus Capcom 3 (more powerful and longer X-factor as you lose more characters), Rage in Smash 4 (deal higher damage and knockback as your percent is higher), and item drops in Mario Kart (get better items the further behind you are). A more subtle version of this used on game shows is to have points increase in value as the game goes on, so the end of the game ultimately decides who wins more than the early game, they just put you through the low value early rounds to pace out the show.
The idea behind comeback factors is to avoid lame duck situations, situations where the game is already over, but the players are still playing, prolonging the inevitable.
The trouble with comeback factors is, they tend to have unintended consequences. Rather, they tend to make it so that losing puts you closer to winning. Another way of looking at it is, across the whole game state, the guy who appears to be winning, is not actually winning yet, because his opponent is sitting on a credit for a ton of points and hasn’t cashed it yet. You might be in first place, but there’s always a blue shell coming.
The intent with comeback factors is frequently to make the game easier for beginners, let people who are losing have a chance to come back, but it frequently means low effort upsets, which are frustrating. It means that people who are good at the game will win with less consistency, because in a game with comebacks, getting ahead also means you’re getting closer to losing.
In some games however there’s an equally troublesome mechanic, the slippery slope. Slippery slopes are when winning makes it easier to win, such as by increasing your own power, or by decreasing your opponent’s. Common slippery slope mechanics are when your power is related to how many people you have on your team, and team members are knocked out over the course of the game. This is seen in Counter Strike, Marvel Versus Capcom, and every RTS game. When you kill your opponent’s troops, the opposing side loses its capability to fight back progressively, ultimately leading to lame duck scenarios. In these cases, comeback mechanics can help counteract the slippery slope effect when implemented well, but they’re still dangerous. Marvel 3 implemented X-Factor to help compensate for the loss of team members, but level 3 X-factor can let you destroy your entire opponent’s team in seconds with the right anchor character.
Slippery slope and comeback mechanics can also be referred to as positive and negative feedback. Positive feedback is when a feedback loop magnifies disturbances over time, and negative feedback is when disturbances are counteracted over time. Negative feedback helps keep the state stable, positive feedback helps make the state fluctuate wildly. In this way you can see negative feedback isn’t always a good thing because the match gets closer to being decided, but neither player is technically ahead, even though the player that’s been winning all the exchanges looks like they’re ahead, which can result in confusion or frustration for the players when the match is called. If players grow stronger as they take more damage, then quickly the winning strategy becomes to damage your opponent, but at a slightly slower rate than they damage you, then knock them out when the potential amount of damage you can deal to them is higher than their ability to knock you out with their current damage potential. This strategy is confusing as hell to actually execute and partially demands players lose exchanges on purpose, which is counter-intuitive and counter-competitive.
Positive feedback or slippery slopes have the downside of creating lame duck scenarios more commonly. When winning exchanges also increases your power to win exchanges, you become radically more likely to win overall. In Starcraft, the vast majority of matches are ended with the losing player surrendering to avoid wasting the other player’s time. Creating lame duck scenarios isn’t ideal, so the solution to this is to either limit the effects of the slippery slope, or actually accelerate the effects of the slippery slope once it’s reached a certain threshold. The idea behind accelerating the slippery slope is to make it so that if either player ends up in a lame duck scenario, the game is instantly won right there, rather than playing the rest of the game out. Obviously a compounding problem of this approach is that lame duck scenarios may emerge earlier into the game, which is why it pays to be cautious with accelerating the slippery slope, and make the acceleration contingent on already having a significant lead, rather than accelerating the slippery slope at all phases of the game, so hopefully the actual lame-duck phase will be very short. One simple idea is just to say that if you’re this far ahead, you win right then and there, long before the ultimate victory condition at the end-game.
Minimizing the effect of the slippery slope can be done in a couple ways. I’ve already discussed using negative feedback, comeback mechanics, to negate the effect of the slippery slope, provide compensation for the loss of fighting power, this can be dangerous because the amount of compensation can easily go too far beyond what was originally lost if you’re not careful. Another interesting example of this type of comeback factor in RTS games is the attention tax and defender’s advantage. As you control more units in RTS, you gain more power than your opponent, but you also pay with the amount of attention you need to devote to controlling so many units, reducing your effectiveness. The defender’s advantage is that when attacking an opponent’s base, the defender’s new units will be spawned directly at the site of the battle, but your unit reinforcements will take time to arrive from your own base as your units are depleted.
The other way slippery slope effects can be minimized is by making them temporary. Team FPS games do this by having team members respawn after being killed, so killing an opponent can reduce the effectiveness of their team temporarily, but not permanently reduce their ability to win exchanges. In fighting games, you can be temporarily disadvantaged with a knockdown, but this expires as you get up and play the wakeup game. Smash Bros is particularly complex for almost never being completely neutral. There’s almost always some type of positional advantage or disadvantage with regards to stage positioning, platforms, the ledge, offstage, or being in the air, even when you’re not knocked down or being combo’d. This can also mean disconnecting the win condition from metrics for efficiency in winning exchanges. For example, the win condition in Starcraft is eliminating every single enemy unit. The less enemy units there are, the harder it is to even get closer to achieving this win condition due to Lanchester’s Laws. By disconnecting these, such as by having an additional statistic for “victory points” or some such, you can award points for winning exchanges without those points then being able to influence whether you win future exchanges (of course in the case of a game like starcraft, this wouldn’t have much effect in reducing the slippery slope, but it’s a start).
The last way slippery slopes can be avoided is by not having them in the first place. Fighting games are notable for this. Losing health in a fighting game doesn’t decrease your ability to win exchanges, so it’s always possible to come back if you have a sliver of health, but it may be difficult, requiring you to win exchanges many times in a row. Part of the way they accomplish this is also by allowing players to negate attrition. You cannot damage the other player without winning an exchange. In many other games like RTS and First Person Shooters, damage is automatic, and winning exchanges is about dealing more damage than your opponent, rather than choosing who will be the only one to deal damage. This can also prevent lame duck scenarios where it’s impossible to negate enough damage to come back, such as where you’re in an FPS and you have no health, your opponent has a bunch of health and you both point at each other and shoot without missing. Though it is also possible in fighting games to knock someone down, then hit them with a super or other high chip damage attack to force them to lose in a small slippery slope moment. Tactically, most fighting games are also designed where health bars are short enough where a player can never get an advantage on their opponent that takes longer than 30 seconds to equalize, so it’s impossible to get a big enough lead that it takes many many won exchanges to come back (an exception to this is Smash Bros, where comebacks are technically possible, but can be difficult and take a long time). So basically, don’t make progression towards the victory condition inevitable and governed by an uncontested efficiency race.
So the gist is, be very very careful with both slippery slopes and comeback factors, consider limiting their effect by either making their effects weak or temporary so they don’t result in either the match see-sawing between powerful comeback factors, or leaving lame-duck players. Doing this can help people more clearly understand who’s actually in the lead, and not need to forfeit to avoid wasting time.