User:Semicolon/Treatise on the Existence of Tiers

The printable version is no longer supported and may have rendering errors. Please update your browser bookmarks and please use the default browser print function instead.

This is an essay, a page containing the advice or opinions of one or more SmashWiki contributors. You may heed it or not, at your discretion.

:By the SLAPAHO Crew, coauthored by Foghorn and Ax.

It seems that controversy over the actual existence of tiers is far more contentious since the release of Brawl, and since the last tier wars fizzled to a standstill, and as such, we here at SLAPAHO have decided to compile this volume. Its intent is to put away the idea that tiers do not exist. However, we're not asking you to like that tiers exist; we’re asking that you accept it. In other words, this treatise is designed to answer a qualitative question, not a normative one.

Anecdotes where a player playing with a lower tier beats a player playing with a higher tier do not disprove the existence of tiers, but they do illustrate a far more important point. Skill is oftentimes greater than tier. Occasionally there will be matches that you lose or win because of the relative tiers, but the tier difference is not insurmountable. A smarter, better player can always win out over tiers. We are not suggesting that you should abandon playing lower tiered characters. Smash is about fun; play whoever you want. We play characters in all tiers, and it helps us become better players. If you’re that serious, play only higher tiered characters in serious matches, but picking a higher tiered character will not overcome the skill difference between a lower tiered main and an unfamiliar higher tiered character.

Tiers are empirically decided. There are very few statistics, facts, or otherwise rationalistic data to make tier decisions. It takes a long time to get things right, and in the end there may never be a consensus. The original Smash Brothers tier list still has controversies for just this reason.

Summary of arguments

Outlined reasons are the following and will be elaborated on:

Argument outline

It is statistically highly improbable (rather, nigh impossible) that a game as complex in variable (meaning, in this case, the myriad of variable abilities including air speed, priority, weight etc.), medium (meaning, in this case and hereby referring to, a character), and construct (meaning, in this case, the metagame, strategies, mindgames, and techs) could possibly be balanced under almost any set of assumptions.

A corollary to this argument involves the unpredictability of the constructs. Since it is impossible for the game creators to predict the ultimate utility of the various constructs, it is impossible to balance them.

Even minute differences can cause imbalance.
There is a near consensus among professionals that tiers exist, and tiers have developed in many other fighting games that are character selectable.

Important concepts

First, before we get to explaining of the arguments and what they mean for tiers, we have to examine a critical idea: what is a tier? Simply put, a tier is a ranking of how a character (or a set of characters) is expected to perform under tournament conditions based on that character’s metagame. The practical offshoot of this is the idea that the higher the tier of a character, the better that character is in relevant aspects of the game. Tiers are malleable, and change over time (even if these changes are marginal). They are based on tournament placing, so they are empirically decided, and disputes over tiers cannot be decided with logical number crunching (and before you say it, the existence of tiers CAN be logically proven, as we will see later). Instead, they must be decided by a very large sample of data under very controlled tournament conditions, and from there medium placing can be determined through inductive reasoning.

Another idea important to this discussion is what is behind a tier, or in simpler terms what causes the existence of tiers. Clearly, differences in the abilities of characters are what tiers are based on, say they can be said to cause the existence of tiers. But these differences have a quality referred to in gaming as imbalance. Particularly in real time strategy games, the idea of imbalance features prominently, but it is common in fighting games as well. There is no game that has ever been said to have perfect balance, with the possible exception of Starcraft: Broodwar. So, with the conditionality of Broodwar’s balance in dispute, it can be said that the set of all balanced games is the null set, that is, there is no such thing as a balanced game.

Now we will move to the discussion of the evidence for the existence of tiers.

Specific favorable arguments

From statistics

The first argument, the statistical argument, is as stated above: “It is statistically highly improbable (rather, nigh impossible) that a game as complex in variable, medium, and construct could possibly be balanced under almost any set of assumptions.”

Here are some facts: in Super Smash Brothers Brawl, there are 39 media, at least 13 variables, and nearly unlimited constructs. In fact, the documented constructs are too numerous and variable to be enumerated here, though reference to the relevant SmashBoards thread may be enlightening on just a few. With all this variation, how could it be possible that it all could fall neatly so that every medium, played to its highest, be balanced? Simply put, the chance of this is so infinitesimal as to be absurd. Even in the most balanced game in existence, Starcraft: Broodwar, there are only 3 media and about five meaningful variables (there are some parts that are not analogous, but these do not invalidate the analogy). In the case of Broodwar and the professional scene, even the constructs are limited (in the form of build orders) to about seven to ten useful or efficient. These limitations allow for a more even playing field. The sheer number of these arguments (here meant in the sense of concrete items, referring to the number of media, variables and constructs) make balance impossible.

A corollary to this is further enhancing the idea of the constructs. Constructs affect tier placement heavily. Look to the Melee Ice Climbers before Chu Dat, or now to Lucario before Azen. Before the constructs of these media were developed, their tournament placement (and by extension, tier placement) were not what they would become. These strategies cannot be predicted by the game developers. So, if this is reasoned to its limit, constructs affect balance, and balance affects tiers, then constructs affect tiers. If constructs affect tiers, and if constructs are unpredictable by developers, then it is clear that developers cannot program balance. Again, imbalance causes tiers.

Now we move on to the argument of difference...

From difference

Imagine a football game where there are only two teams. Both teams are identical, and played by players of identical skill. The ONLY difference is that one team has white jerseys and the other black. This ought not affect the outcome, you might think, but imagine this: the two players have selected a perfectly balanced football stadium, except that the weather conditions are snowing. Imagine if the tournament rules for this game are that it must be played in this stadium, and that any weather condition is acceptable, or even that snowing is the only acceptable weather condition. In this situation, it is reasonable to assume that the white team will win this game, because it will be difficult for the black player to spot the white players in the snowy graphics. Even if this affects a single play, it affects balance, and balance affects tiers. Even a difference as small as having a team with white jerseys can win a game, all other conditions controlled for. The point is not to illustrate a crazy scenario that seems vaguely racist, but to demonstrate that even small differences can cause imbalance and cause tiers.

Something similar can be accounted for in Smash Bros. Even if there is a small difference in a singular game mechanic, the balance can fall out, and this is assuming that the game is perfectly balanced at the start. Now, apply this to knowledge of Smash Bros. There are dozens of differences, even in cloned characters in Melee. Doubtless these differences are often large, influencing balance severely. Even in several instances, including simultaneous deaths in Sudden Death mode and co-incident grabs, the game decides based on controller ports. Small differences in variables can make huge differences in constructs, and thus it can be argued that big differences in variables make even larger differences in constructs (though this is not necessarily the case per the fallacious argument not enumerated here, still, in Smash Brothers such a comparison seems valid). Differences, as we have seen, cause imbalance, and imbalance causes tiers. Therefore, small differences help create tiers.

From consensus

Finally, taking a page from the play book of the global warming activists, there is argument by consensus, which is actually a valid argument when empiricism is the only possible epistemological method.

It is a near consensus among professionals and many casual players that tiers exist. It is intuitive; when you picked up the first game, did you not make judgments about which characters were good and which were bad? Additionally, it has been said that most other games with analogous mechanisms possess tiers.

Arguments in opposition

Now there are some opposing arguments that must be examined. These will be dealt with in order:

Tiers weren’t intended.
Tiers ruin fun.
Individual matchups are the most important determinant in tier placement.
Tiers shouldn’t exist, therefore they don’t.
Tiers cause the focus of play to go to high tier characters, causing them to be the only ones played and thereby reinforcing the tier list.

We shall examine these thusly.

From intent

The first argument, the argument of intent, is a fallacy. To start, just because they were not intended does not mean they do not exist. In fact, it is very logical to assume that if tiers were not intended that they should form (because then the opposite is also lacking in developmental consideration). It seems reasonable to think that the developers gave almost no intent towards the balance of the game, making tiers unintentional consequences. Indeed, to actually balance the game must be an intended result, and the lack of intent on the part of the developers under any case would result in tiers.

From 'fun'

Of the second argument, the argument of fun, nothing needs to be said other than to read the opening ‘Keep in mind’ section, and then say that this argument is irrelevant. It is also important to remember that tiers are rarely applicable to casual play, in fact, a large constituency of play breaks standard tournament rule sets.

From match-up

Of the third, doubtless the most substantiate, it must be remembered that match-up conditions fluctuate and that match-ups are not the only way tiers are decided. Play at its highest level is meant to control for individual match-ups and skill with volume. So, no, individual match-ups are not the most important factor because it is (a) already considered through tournament placements and (b) it is one of many things that must be considered. Additionally, a bottom tier medium may have a good match-up against top tier media (see DK circa Melee) but have poor match-ups against others. This does not make the low tier medium top tier, though the property is certainly unique and intriguing.

From ignorance

The fourth argument, called the argument by plugging the ears and saying ‘lalalalalalalalala’ in response to arguments in favor, should have its deficiencies apparent to anyone. If they are not, please contact us, and we will arrange for private tutoring in logic. Not free of charge. On a more serious discussion however, the idea of should-and-do are completely different contentions. 'Should' implies a normative contention, whereas 'do' implies a qualitative contention. Many things in the world should not exist if society was to be perfect, but this does nothing to diminish the fact that they do exist. We should not live in a world where teachers are paid a significant amount less than those they trained, yet we do. The purpose of this work is to answer the quantitative question, "do tiers exist." Whether or not they should exist is completely orthogonal to the content of this work (N.B. we will respond to this point later in the work).

Recursive argument

The fifth argument, called the recursive argument, is invalid because it does not account for the large variation among professionals. At least for Melee, while the tiers were in place, professionals still played their best medium, and not always who was top tier. Some players are dedicated low tier professionals. Still, through all this, the clear majority of tournament victories in Melee went to Fox/Falco, Marth and Sheik. The recursive argument may in fact have some validity but its reach is insufficient to disprove the existence of tiers, it merely says that the existence of a tier list affects metagame, leaving unproved the follow up that characters are balanced and the tier list corrupts that balance. Indeed, to derive such an assumption from the argument would be a fallacy.

Normative arguments

While the previous purpose of this paper has been to answer the qualitative question of if tiers exist, continued debate on the subject will prove it necessary to examine two normative arguments against tiers. Again, it should be noted that these two points do not in any way argue for or against the existence of tiers (cf. point four in the previous section), but are simply arguments that must be addressed (to some extent regrettably) when discussing tiers.

We shouldn't form a tier list and/or believe in tiers
Tiers ruin the game

Again, the following will discuss these arguments.

Formation of tier lists

On the first point, this has most often been accompanied by the quote from designer Sakurai about how he did not want Smash Bros. to be competitive. First, it should be noted that his quote was taken completely out of context. All Sakurai said was that he himself was not a competitive person and did not want to design a game where the same person could win over and over simply by continuing to do the same moves. Yes, Sakurai probably would never play in a tournament. But that does not mean that others in the community cannot. I reference any other form of art: just because the artist meant for a singular concrete interpretation does not invalidate other interpretations. To Sakurai, the fun in Smash comes from the randomness of stages like Spear Pillar and the plethora of items. To others, it comes from heated competition, where it is simply your choices against another persons, without game-changing events getting in the way. Is one way better than another? No, of course not. We may prefer the latter, but that is no reason to reject the former or vice versa. Quite simply, people are free to choose the way they want to play the game. Therefore, there is nothing wrong with playing Smash Bros. under a tournament rule set just as there is nothing wrong with playing crazy games with lots of random things happening.

Given that tournament play is acceptable, what then becomes unacceptable about tiers? The general argument is that tier lists should not exist because they only exist for tournament play or that they simply continue to reinforce themselves. For the first of those points, read the above paragraph again for the answer. Yes, they only exist for tournament play because that is what they are intended for. As for the point on them reinforcing themselves, there is more to be said. Yes, there exist players who switched to Meta Knight the moment the Back Room announced him as top tier and now refuse to play any other character. But they are the small minority. Given the inherent neutrality of a mirror match-up, top players will always be looking for the counter to the top tier characters. Take, for example, Young Link in Melee. He was a low tier character, but he also developed into the best Peach match-up in the game. Players that consistently try to win are always looking for the best answer to each problem facing them. Picking a main and secondary characters is not a simple dart board toss or a look at the tier list for top players. They spend hours testing each character, discovering what that character is capable of and how they play them. No player would ever say, well Meta Knight is bad against Diddy Kong (this a purely hypothetical match-up description), so I am just going to have to lose all my Diddy match-ups. Instead, s/he would find another character to counter Diddy. Maybe s/he would discover a new technique(s) with that character and vault them up the tier list, just like Chu Dat and the Ice Climbers. By the nature of the fact that no character has perfect match-ups across the board, players will always being trying new things with other characters and constantly affecting the tier list.

Deleterious properties of tier formation

For the second point, what evidence is there that tiers ruin the game? And more over, tiers are, to a certain degree, not in control of the players. Yes, we are able to develop a metagame for each character, but we are limited by the programming of the game. If it turns out that Meta Knight is so much better than any other character as to be broken, what are we as players supposed to do? We could ban Meta Knight, but that is a subject for a different paper. If Meta Knight is still allowed and he is as good as I am describing, why would any competitive player not play Meta Knight? Even if tiers do ruin the game (and I will contend that they do not), competitive players cannot concern themselves with anything other than winning.

So, the question becomes why it does not ruin the game to have tiers. Here I simply reference almost every other competitive game there is. For the sake of a specific example, I will discuss Magic: the Gathering. At any point in Standard (Type II) Magic tournaments, there are over a thousand cards that are tournament legal. Is there any way to balance each of these cards perfectly? Of course not (cf. the statistical arguments above). However, does this ruin the game? Except in very specific periods, no (for those Magic buffs out there I'm referencing Urza's block and to a lesser extent Mirrodin). The reason it does not ruin the game is that there is a certain Rock-Paper-Scissors element to the metagame. On the one hand, there is deck X that is first seen as the top deck. Then deck Y comes along and beats deck X, but does not do much else. So now we have deck X that beats deck Z, but Z beats Y, which beats X. The point here is that because there are answers to even the top deck, there is an element of balance to the system. In Smash Bros. the same things happen. Yes, characters exist that are better than others, but they also have bad match-ups against low tier characters. This creates the dynamic described above. Even in the top tiers this exists. Here I reference Melee. In Melee, it was seen that Fox was a good match-up on Sheik, Sheik was a good match-up on Marth, and Marth was a good Match-up on Fox. Falco also factored into the mix as a slightly good match-up on Fox and Sheik, and less poor Marth match-up. That's balance even in the face of tiers.

Addendum as of March 2010

It has become necessary to make an addendum to this treatise. Much of the dissent regarding this treatise and indeed tiers themselves comes from a deficit in the understanding of their nature and function. While this was treated at some length at the introduction to this treatise, further detail is required.

Nature

The nature of tiers is a dramatically misunderstood concept. As enumerated in the introduction to this treatise, a tier is a ranking of sorts. It is does not contain information about matchups, player skill, stage balance, the effects of items, timed matches, or special brawls. It is relevant only in a single context: tournaments run by Smash Back Room rules. Tiers are derived from a body of data from these very same tournaments, based on the performance of characters at these events. Thus, tiers are rankings of which characters are most effective in the aforementioned tournament context relative to other characters. Here are some important points:

Tiers are graded. A character at place 12 is more effective than a character at place 21, but the difference in effectiveness, theoretically, is less by relative position; that is, the difference in effectiveness is greater the larger the margin of separation between two relative places.
Tiers divisions are distinguished by performance. Tier divisions (S, A, B, etc.) are meant to reflect the absolute performance of certain characters. That is, characters is S belong to the most successful characters who have largely similar tournament track records, A for the next most distinct group, B for the next most, etc.
Divisions are absolute, positions are relative. The particular division selected for a character is based on a cutoff for tournament success, but the placement within those cutoffs is considerably more fluid. Divisions are meant to group characters with similar successes, positions laterally may be used to compare characters.

Function

The function of a tier list is to group characters. The practical offshoot, as discussed, is that because the characters grouped together tend to be similar in their success in tournaments, certain characters are better than others. The function is to discriminate characters into categories, nothing more. The following are things that tier lists, as mentioned above, are not intended for:

Tiers are not meant to predict which player will win a given match in any context. They predict on the macro, not the micro. There are many factors contributing to who wins matches. Tier lists are typically accurate in determining which character (note, not player) will be the eventual victor in a tournament but that is not their function.
Tiers are not meant as a matchup chart. Tiers do not contain information about specific matchups; that information is not incorporated into what data constitutes the body of data from which tier conclusions are derived. However, in certain regards, matchups influence by nature which players have more success.
Tiers are not meant to tell you which character to play. Play who you want. Who is best should be irrelevant to your experience with the game, unless your intent is tournament success.
Tiers don't define strategic roles. Tiers are not a subsititute for strategy, nor do they account for the execution or potential of strategy.

Conclusion

To conclude, several arguments both in favor of and opposed to the idea and existence of tiers have been discussed, and arguments opposed to have clearly been shown to be deficient. Doubtless opposition to tiers will continue for these reasons and beyond the extension of this discussion, but consider this a handbook to use against opponents of tiers, who have already been pushed from the mainstream, but ought to be obfuscated entirely from discussions of competitive play. Again, we are not asking for people to like tiers, but asking for the intellectual honesty to acknowledge the overwhelming case for their existence. Should this discussion ever require amendment for some new argument presented by opponents of tiers, it is resolved that this discussion will continue to be the definitive volume for the case in favor of the existence of tiers.