On an alternate stream for ESL Pro League, Christopher "GeT_RiGhT" Alesund, Duncan "Thorin" Shields, and Tomi "lurppis" Kovanen had an argument about which player on Astralis was "the best" in Counter-Strike: Global Offensive.
Thorin and lurppis argued for the consensus best player, Nicolai "dev1ce" Reedtz, while GeT_RiGhT argued contentiously for Andreas "Xyp9x" Højsleth, considered the support player of Astralis. This suggests a vast difference in what both parties consider to compose the "best" player, and it speaks to the polar-opposite opinions in the never-ending debate.
The term "best" refers to a concept, essentially meaning "the player who contributes most to reaching an objective." In Counter-Strike, the goal is to win tournaments. The aforementioned argument demonstrates a difference in measures of this concept. To figure out what makes a player best is to ask how we should measure "bestness."
The obvious argument for dev1ce is the statistical one. The sheer volume of kills that dev1ce gets is staggering and unparalleled in his team. Using the HLTV rating, he is the highest-performing player on his team on LAN. His opening kill ratio is remarkable. He's the best statistical player on Astralis in basically every measure we use to determine the best.
However, statistical analysis of this kind can be challenged. Data points like average damage per round (ADR) and kills per round (KPR) have a star-player bias, as they generally are put in situations where they are more likely than their teammates to find kills, which end up padding their statistics. So if Xyp9x and dev1ce both had an 85 ADR, it is quite likely that Xyp9x was the better performer of the two, as his kills were likely to come from clutch situations and more difficult positions. So if we want to prove that dev1ce is the better player, statistics won't do the trick.
If roles bring out a flaw in statistical measures, it blurs other measures of the "best" player. Specifically, roles force us to take into consideration the burden of expectation that should be placed on each player. For dev1ce's performance to be considered average, it must be significantly more impressive than Xyp9x's as a consequence of the sacrifices his team makes to set him up for success. The economy is an example of one such sacrifice. Since dev1ce is a dedicated awper, his team will purchase something that has massive economic repercussions for him. If someone gets only a single kill in three purchases of an AK-47, it is much less costly for the team than it would be if the exact same stat line was with an awp instead. With an awp rather than an AK, $6,350 more was sacrificed to get that one kill. As a consequence of this added cost, dev1ce is expected to have a significantly bigger impact on the game than he would with the AK.
And yet, there's more to dev1ce's performance than his statistics and the burden of his role. In fact, his role itself is not merely a constant pressure -- it's also a pedestal. Of course, Xyp9x's performance relative to his role's limitations may be more impressive, but is it more important? Remember the definition of "best" given: the player who contributes most to winning tournaments. Does Xyp9x contribute more to winning tournaments than dev1ce?
Dev1ce is put into a role with a higher burden because he is more capable than Xyp9x. If the latter could carry a game better than the former, the positions would be reversed and Xyp9x would be the one set up up for success. This is critical because the star player's role contributes more to winning than the support's. After all, the support's role is make room for the star to dominate. So if a star is more skilled than a support, and his role is more important within the game, it's hard to argue that dev1ce isn't the best player on Astralis.
This argument has its own difficulties. The assumption made is that "best" means most skilled. Obviously, Counter-Strike is in large part about aim. Yet, for a support player, very often the most useful skill is not aim. For example, when Xyp9x plays solo B site of Train, his fitness for the role is not due to his ability to kill multiple people entering the site -- the position is much too vulnerable to reliably succeed in doing so, regardless of aiming ability. Rather, the spot demands the ability to navigate angles effectively to stay alive as long as possible. Hopefully, he'll be able to get a kill or two, but the setup doesn't require him to do so. It does, however, require him to stay alive long enough to enable a quick rotation from his teammates to create an effective retake situation. In fact, most situations Xyp9x finds himself in requires a different kind of ability. Simply implying aim-based positions are more important than other equally demanding positions is unsatisfactory as a justification of dev1ce's superiority.
Since arguments on both sides fall short, we need a new angle with which we can tackle this question. A player's performance must be placed not simply by the facticity of his role, but by its context. That is to say that the player must not only be compared to his teammates, but also, perhaps even more so, to those who share his role. For example, dev1ce should not be compared to Xyp9x. Rather, his impact should be compared with other awpers. If we look at Xyp9x's role, his performance relative to his peers is not only good -- it is the best.
Here's a thought experiment that would help integrate this added variable: If you had to replace either dev1ce or Xyp9x for any other player in the game, would you do it? This question forces us to compare each player to their peers. However, it does more for us than that. It makes us really consider who's most valuable on the team. While we can find Xyp9x more valuable relative to other support players, we can reject his claim to being the best by considering who's more valuable to the team. True superstars, like dev1ce, s1mple or coldzera, offer more to the team than mere skill. They have unique playstyles that their teams build themselves around. Swapping them with other equally skilled players would be, in most cases, ineffective. You wouldn't be able to put s1mple on Astralis, even without language barriers. Astralis is strategically constructed to match dev1ce's pattern of play, and no other player can replace him.
Despite the final dull conclusion, there's an insight to be found in this process. It would seem that being the "best" on your team doesn't simply mean being the most skilled, or the star or even the most important. At least in part, in means being irreplaceable.