EPTThis quote has always baffled me. I have a great deal of respect for Day[9], having learned a good portion of what I know about Starcraft by watching his dailies. He's repeated this line multiple times, often as the key argument of a reasonably long segment in a prepared daily, so it's hardly something he's saying off the cuff without thinking through what he is saying. And he is, in general, a methodical and intelligent thinker on game-related subjects, as exemplified by his article "The Marginal Advantage" (see here: http://www.teamliquid.net/forum/viewmessage.php?topic_id=64514).
When Day[9] says this quote, he's specifically discussing the league (platinum, diamond, etc.) in which individuals might find themselves. If one individual is in bronze, and another is in masters, most people would be inclined to say this is evidence that the former player is less skilled than the ladder. Day[9] is asserting that this is not the case: ladder rank measures a player's progress, not skill. So the latter player may be more skilled than the former, and there certainly may be a correlation between skill and progress, but ultimately ladder doesn't address the issue. The correct description would be "Player B has progressed further at Starcraft than Player A."
![[image loading]](http://i.imgur.com/RvOD6zx.jpg)
A t-shirt celebrating a complete lack of progress.
But this is somewhat unintuitive; people do, after all, consistently use ladder ranks as a description of skill. Someone in Grandmasters might appropriately describe themselves as "pretty good at Starcraft II," while someone in bronze might claim to be somewhat worse. Isn't skill the ability to win games more consistently, thus earning a higher position on the ladder? And if not, then what exactly is skill, and how can it be measured?
Skill
Competitive games are environments in which a set of rules are imposed on any and all players, and a win condition for those players is set. The players then act within those rules to try and achieve their win condition. The concept of 'skill' ultimately derives from the observation that some players seem to win a lot more than others, and that it is easier to win against some players than others. So whatever measure of skill we choose, it should meet a few criteria:
1. Every player has a quantifiable skill.
2. In any given game, the most skilled player should usually have the highest chance of winning, the next most skilled player the next highest chance, etc. (In a 1v1 game, this can be shortened to "The more skilled player has a higher chance of winning."
3. In a matchup between players of equal skill, wins should be split approximately evenly. (In a 1v1 game, this split is obviously 50%-50%.)
These criteria can help us rule out a few popular measures of skill. Suppose someone likes to discuss the relative skills of players by measuring their APM: the first test is passed, as APM is most certainly quantifiable, but players of greater APM don't necessarily win more often than players of lesser APM, and players of similar APM don't necessarily have anything close to 50% win rates. There may be a correlation between APM and skill, but APM is definitely not a measure of skill.
Similarly, someone might like to use pure win rate as a measure of skill. Tournaments will sometimes display a player's win rates in certain match-ups to show how skilled they are at different match-ups. This measure plainly passes the first test, as it is certainly quantifiable. But it fails the second and third criteria; if my TvZ win rate is 75% and Bomber's is only 53%, I'm not a better player than Bomber; he's just facing more difficult Zergs than I am.
Does a matchmaking system meet our criteria for measuring skill? In short: yes. An MMR is quantifiable, players with similar MMR's tend to have about 50% win rates against each other, and a player with higher MMR tends to win against a player of lower MMR. It's a relative measure of skill, not an absolute one, but so far it seems to be the most reliable measure of skill available to us.
Are there alternative definitions of skill which MMR might not measure effectively? Certainly. One essential problem with the above criteria is that players' skill can change over time; my friend might consistently beat me when we play 1v1's, but if I have 8-hour ladder sessions every day for a month, then at the end I'll probably beat him consistently. So the skill of a player is not a constant quantity, like the atomic mass of mercury or the charge of a single electron; it's a variable, like the velocity of my car or the volume of liquid in my water bottle.
In theory, someone might like to discuss the skill of a player, not as a variable, but as a constant. In this case we are not discussing their ability to win at this exact moment, since that changes over time; instead, we are considering some abstract quantity like a person's maximum potential skill — the absolute highest ability to win that they could achieve given enough study of the game.
![[image loading]](http://i.imgur.com/aN5znLS.jpg)
"Pff, he's not more skilled than me. He's just practiced a lot more than I have."
MMR is, of course, not a measure of this quantity in the slightest. But then, nothing is. About the only way to know what someone's maximum potential skill might be is to have them practice professionally until their skill level seems to stop rising, which is, of course, an impractical test for most players to perform. Even that test assumes that the player's skill will cap out eventually, and learning curves are not necessarily that predictable. Some players may asymptotically approach some skill level, which you would be inclined to call their theoretical maximum even if they never reach it. But what if their skill level just keeps rising linearly (or at least logarithmically)? Then the maximum potential skill simply doesn't exist.
Progress
The term "progress" implies a path. More specifically, it implies a sequence of requirements that must be met in their proper order to get from a bad place to a good one. In the context of Starcraft, this implies a linear curriculum of skills that must be learned in order to progress as a player.
Linearized curricula are not uncommon in education; kids have to learn something like math in some order, so an order is chosen and all students learn math in that order (e.g. 8th grade Algebra I, 9th grade Geometry, 10th grade Algebra II, etc.). One can imagine such a curriculum for Starcraft II:
![[image loading]](http://i.imgur.com/KVH79nM.png)
This picture could be the millionth thing titled "From Bronze to Masters".
The problem is that there's no universal linear curriculum for learning SC2. One player might learn advanced army control while still in bronze league, and only learn the higher reaches of the tech tree later; another might start with scouting in silver and only later learn even basic army control; a third might start learning to macro on roaches at bronze, and never learn anything else until platinum. Another might play the game only to six-pool badly, and never reach platinum at all.
This makes it so that MMR isn't exactly a measure of progress. In fact, it's difficult to measure progress at all, or even to define it. Inevitably in comparing the progress of one player to that of another, one must make judgments about which elements of learning are most important. If this player can consistently macro on their command centers and never miss a moment of production, but this player can 2-rax very effectively and control their marines to great effect, which has progressed further? The former probably has a better base of mechanics to build on, but the latter easily might have a higher MMR.
Conclusions
Contrary to the starting quote, league placement and MMR are measures of skill, not progress. That is, however, qualified by what is meant by the term "skill." Skill means capability of winning at Starcraft II at this very instant. Not how smart or talented you are as a person, not how difficult it was to get this level of skill, not even how capable of winning you will be tomorrow. If the underlying point of the quote is that your league placement today is no reason to believe that you are more or less able to become great at Starcraft II in the future, then the underlying point is correct.
And progress, while not effectively measured by MMR, is something worth considering as a player. If skill is a measure of how capable you are of winning today, progress is how you will reach a higher level of skill tomorrow. A commentator, observer, or game designer might like to talk about evaluating the skill of various players, but from a player perspective, evaluating skill isn't so important. For a player, the important question is not "How skilled am I?" but "How can I progress toward being a better player tomorrow?" Consider what improvements you have made in the past that have helped you earn what league placement you have, and try to identify weaknesses in your play that might be holding you back from a higher league.
But at the end of the day, when you see your league placement or calculate your MMR, don't think it's anything other than a measure of exactly how skilled you are.
