Who is the Best?
by MizenhauerThe word 'best' has only been in the English language since the 12th century, but in spirit, I imagine we've been arguing about who's 'the best' in sports since time untold. We've gone from a time when the majority of professional sporting events weren't even broadcasted on television to an era in which we have more access to games and data than ever before.
Despite this wealth of information, we seem doomed to revert to our old definitions and axioms when deciding who's the best. Who won the biggest event of the season? Who recorded the highest win-rate over a set period? And, of course, what have you done for me lately?
As we head toward the end of the 2019 season, I felt like it was worth taking a step back, catching our breaths, and thinking about all the inconsistencies and arbitrary judgments that we employ when arguing about which player is 'the best.' Some of the ground I'm covering should feel awfully familiar, but I think it's worth a review anyway. And, perhaps, there's some new perspective I can provide as well.
Discussing the events of 2019 would make things a little too contentious—fortunately 2014 presents us with some great examples to consider with the benefit of hindsight (and I'll espouse a bit on 2019 toward the end).
What's the value of a non-first place finish?
We generally believe championships matter most when deciding who's 'best,' but we also seem to agree that other criteria have some varying amount of importance. The greatest and most famous example of this will always be soO’s championship-less 2014 season where he reached every single Code S final and lost in all of them. Until Maru's 2018 run, three straight Code S finals was an unprecedented, unthinkable accomplishment, and one could have argued that soO was an overall superior player compared to any of the one-time champions of 2014. In the course of reaching every single Code S final that year, soO notched a ridiculous 54 victories in Code S over the course of those three seasons, nine more than Season 1 Champion Zest (45) and approximately twice the number of wins logged by Season 2 Champion Classic (27) and Season 3 Champion INnoVation (28). There's our first complication in determining who the best player is: How do we measure the ability to place consistently high in tournaments compared to winning the tournament outright?
[ByuL’s three final appearances in GSL/SSL events in 2015 is another story in this vein, but we'll leave that for another day.]
The playing field isn't always even.
Another championship-less player from 2014 who deserves a second look at is Maru, who appeared in the quarterfinals and semifinals of Season 1 and 2 of Code S, respectively. These hardly seem like noteworthy results on paper, but it's different once you present some context: only three Terrans even qualified for Code S Season 1 to begin with, only two reached the Round of 16, and only Maru survived to play in the Round of 8. Things didn’t change much in the following Season 2 where Maru was one of four Terrans to qualify for Code S, one of only two Terrans to reach the Round of 16, and the only one to make it to the playoffs. That time, he even won a series on top of that, beating Soulkey to reach the Round of 4. Terrans won a mere 42 games during the first two seasons of Code S, with Maru responsible for over half of them at 23 (in terms of matches, he won nine out of Terran's sixteen).
[Proleague rosters do show that Terran was slightly under-represented in Korea compared to Zerg and Protoss, but not directly proportionate to their poor Code S showings].
History might mark INnoVation down as the only Terran to win a Code S championship in 2014 (Season 3), but Maru actually equaled his win count across all three Code S tournaments. That brings us to our second complication: How do we account for a player's success relative to their faction's strength at the time? Without going into a deep-dive on balance, we can at least say Terran was doing poorly in Code S for the first half of 2014, while Aligulac.com's balance report presents somewhat poor overall win-rates for Terran across all competitions. Does anyone really think Maru was worse than all the Protoss players who won some form of championship that year? (MC, herO, San, HerO, sOs, Classic, StarDust, Pigbaby, Zest). Some might even compare him favorably to fellow Terrans like Bomber, TaeJa (more on him in a second), Polt, or MMA who won championships in weaker competitions outside the GSL. Which brings us to the next point...
Not all championships are made equally.
We'll leave off on 2014 with one final example: TL.net's 'Best Korean Player' award (de facto best in the world—TL.net awarded foreigners a seperate, participation prize until 2018). TL.net writers ended up voting for Zest, who won three tournaments in Korea that year. TaeJa also won three tournaments that year, just not in Korea (HomeStory Cup 9, Dreamhack Summer and IEM Shenzhen). Perusing the comments, it seems that this decision was not especially contentious (by TL.net standards, anyway). It seemed fans agreed that TaeJa's three did not quite equal Zest's three. But somewhat paradoxically, part of the reason Zest won his Player of the Year award ahead of soO was precisely because three championships was greater than zero (to Zest's credit, he also placed high in tournaments he didn't win). Along the same lines, TL.net awarded TaeJa 'Best Korean Terran' for his three foreign title wins, ahead of INnoVation who had won just GSL Code S that year.
This presents us with the most familiar quandary of all—when players DO win championships, which ones are more valuable? Consciously or unconsciously, we all do this mental math: factoring an event’s prestige, prize pool, magnitude, the overall level of competition in the event, and the level of the competitors the winner actually faced, and trying to come up with a conclusion. There's no absolute truth or mathematical formula here—we do most of these determinations through pure feel and conjecture.
Statistics lie.
Let's bring this back to 2019. Back in August, Stats lifted the trophy at Assembly Summer, even defeating Serral along the way. Despite this, there didn't seem to be any rush to immediately crown Stats as 'the best' player in the world. Code S Season 2 champion Dark retained the top Korean spot on the TL.net Power Rank, despite the fact that he hadn't competed in in Finland, and despite the fact that his Code S title had come over a month prior.
Intrigued by this debate (or non-debate), I tried to make the arguments for either player's superiority. Subjectively, either one could have been judged to have the better tournament resume (Dark: top four in Code S Season 1, top four at IEM Katowice, and a championship in Code S Season 2. Stats: Round of 32 elimination in Code S Season 1, finals loss at IEM Katowice, and Round of 16 elimination in Code S Season 2, top eight HomeStory Cup, 1st place Assembly Summer).
But what about match records and statistics? Those should be less subjective, right? However, they proved to be just as malleable. Dark's win percentage during the period from IEM Katowice to Assembly was five percentage points higher than Stats (80% to 75%). However, one only needs to change the filter the results during the same period we just examined to exclude online matches we find that Stats’ win percentage jumps to nearly 83%. That seems to support the argument that Stats was actually slightly better than Dark in offline matches which ostensibly 'matter' more.
But if one filters the stats a second time to remove results against supposedly inferior non-Korean players (who both Dark and Stats faced a glut of at WESG and HSC, respectively) Dark comes out slightly ahead with a 77.27% win-rate to Stats' 74.07%. And if one filters the data yet again to re-add only matches against Serral (creating a 'versus Koreans+Serral' stat), Stats retakes the lead with a 75.86% (22-7) win-rate compared to Dark's 73.91% (17-6). The point is: Stats can be tricky, and I don't mean just the player.
Of course, I could have begun this all by pointing out how arbitrary this selected date range (IEM to Assembly) is to begin with, given that it has no inherent significance. Why didn't I just start it on February 2nd, the start of Code S Season 1? Or on January 25th, the date of the Code S qualifiers? Or on November 4th, the day after BlizzCon 2018?
At the end of the day, determining who 'the best' is might not even be all that important to anyone except the fans, considering how players like INnoVation seem to be more focused on things like making money. But the pursuit of determining who's 'the best' is, at worst, a pleasant distraction, something that keeps us attached to this game, and plain old fun.
To that end, I hope I've given you some more perspectives to consider the next time you're arguing on TL.net, help you understand that everyone is just collecting different pieces of an unsolvable puzzle, and maybe even convince you to use those Aligulac.com stats more honestly.
Credits and acknowledgements
Written by: Mizenhauer
Editor: Wax
Images: GomTV, Liqupedia, Blizzard
Blanket credit for existing at all: Aligulac.com
Written by: Mizenhauer
Editor: Wax
Images: GomTV, Liqupedia, Blizzard
Blanket credit for existing at all: Aligulac.com