EPTWhat is an "easy game"?
![[image loading]](http://i.imgur.com/S7hPM.jpg)
Tic-Tac-Toe, case closed
Defeating the Misconception
"If the game is so easy, then you should be able to win 100% of the time."
The idea that you should win 100% of the time in an easy game may be true when playing versus a computer or a scoring system, but it is the grand misconception of what it means to be an easy game in any competitive game. When two players play against each other in an easy competitive game, something else happens. Let's use tic-tac-toe to illustrate what happens.
In tic-tac-toe, there are an extremely limited number of options on any turn and the depth is limited to eight turns (on the ninth turn, there is no choice). Therefore, any competent tic-tac-toe player can memorize all the positions (or even logically solve the problem on-the-fly) and come up with the optimal solution. When a master tic-tac-toe player with 10 years of dedicated practice plays against a competent tic-tac-toe player with an hour or two of training, the master doesn't achieve a 100% win rate. Barring a really silly mistake, every game should be a draw and the win percentages should converge to 50%.
In an easy game, a player has no ability to distinguish himself from the masses.
Thus, if a 1v1 game was actually easy (and balanced), nobody would hit 100% (or 90%, 80%, etc) win rate. Any competent player would hit 50% win rate against other competent players. Given the notable variance of skill between players with years of practice, we can immediately say that Starcraft 2 is not an "easy game" by our first definition. However, perhaps we have a problem of definitions. What if we switched our definition to be a "trivial game", and made a new distinction for "easy game".
In a
Starcraft 2 is clearly not a trivial game, but that leaves us still attempting to answer what an easy game is. And to really define an easy game, we will need to start digging into what makes a game trivial, easy, or skillful. We'll need to talk about the skill cap.
The Soft Cap vs the Hard Cap
"The skill cap in Starcraft 2 is infinitely high, just watch Automaton 2000."
![[image loading]](http://i.imgur.com/wywcp.jpg)
perfection is a good skill to have
Take a look at some of Automaton 2000's performances:
+ Show Spoiler [Automaton Ownage] +
Watch those videos and tell me that SC2 has a low skill cap. You can't. With the limits on human dexterity, no human being can actually consistently perform the micro tricks that Automaton 2000 does, so it is safe to say that the skill cap in Starcraft 2 is infinitely high. And thus, no matter how we define "easy game" related to the skill cap, Starcraft 2 could not be considered an easy game, right? Not quite.
What we need to understand is that along with the hard skill cap which is virtually infinite in SC2, there is also a "soft" skill cap. To understand what the soft skill cap is, we'll want to talk a little bit about how players get better in any game.
As a player puts more "practice effort" into a game, he will gain "skill" in that game assuming that the practice effort is relevant towards improving. Some practice regimens are better than others and some people will learn faster than others, but all reasonable practice should result in an increasing skill level. However, in all games, practice effort has diminishing returns. For example, let's say that you and your clone both choose to play SC2 with no prior experience. If you played for 100 hours before your clone played his first hour, you'd have a huge advantage. If you played for 200 hours while your clone played for only 100 hours in the same practice environment, you should still have a noticeable advantage, but it wouldn't be quite so large. Finally, if you played 2100 hours versus your clone's 2000 hours, the advantage would be small enough that it might not show up in most games.
![[image loading]](http://i.imgur.com/AJSkA.jpg)
Clones?!? So that's how they did it.
The diminishing returns on practice effort creates a parabolic curve of skill gain. In any non-trivial game, the parabolic curve should never completely level off. However, the slope will get very mild past a certain amount of practice and that point is what we'll call the soft skill cap. The soft skill cap is the level at which further practice has very minimal effect.
![[image loading]](http://i.imgur.com/YeI98.jpg)
defining easy, not so easy
As you can see, the soft skill cap is not really an exact point on the chart. It's more of a matter of opinion and that's why I leave a large area on the graph as the potential soft cap spot while also marking where I'd personally put it on the charts. Also, as you might have noticed in that graph, I have vaguely defined an easy game. However, you'll notice that there are no numbers on either axis. I do not define "easy" on absolute terms (I'll leave that exercise up to you), but instead on relative terms. A game's difficulty is determined by the depth of its skill curve relative to its peers.
In an easy game, the skill development curve is shallower than in a more skilled game.
Every game has a different curve and every player will develop in a game on a different curve. Some curves might have bumps in them as a player makes a big realization and some might have drops due to skill atrophy/injury. Also, depending on the way a player practices, he might actually work into the hard cap before completing the soft cap. For example, imagine a bronze level Terran player who has terrible macro, doesn't hotkey, and right-clicks his army rather than attack-move, but he practices Marine Split Challenge for hours a day and can beat extremely high levels like a boss. Will that practice help him in game? Absolutely, but while he may save a few extra marines versus banelings in a few situations (and have faster clicking overall), he still won't be able to get out of bronze. Basic marine splitting might be part of the soft cap, but spending many hours to squeeze out one or two more splits is a hard cap skill. Think of the skill curve the next time someone advises you to work on your macro. It's not that the other skills are a waste, it's simply that in most cases you can make the greatest gains for the effort by working on macro first.
![[image loading]](http://i.imgur.com/NSv3g.jpg)
it takes more than marine splits to be like me
So now we've defined the hard cap and the soft cap. And we have a general idea of how to determine if one game is more skilled that than another. Knowing all this, is there a good way to calculate how much depth there is to a game's skill curve?
SCBW vs SC2: Calculating a Game's Skill
One of the most obvious ways determine a skill curve is to look at the win percentages of the top players. Can the top players differentiate themselves from the masses? Presumably, the higher a top player's win percentage, the greater the skill curve in the game.
+ Show Spoiler [Datasets Chosen and Why] +
Starcraft 2
For Starcraft 2, I wanted the most recent data possible since the game is constantly evolving and players are getting better and worse (relatively speaking) on a daily basis. I also wanted data that represented the most current metagame at the highest level with few anomalies due to extremely poor opponents. In a perfect world, I would have been able to gather 100s of games for each player from yesterday to make my determination. Unfortunately, I needed something resembling a significant sample size, so I had to choose a time period longer than one day. As it turned out, I chose eight months from 2012-Jan-1 to 2012-Aug-1 from the Korean TLPD database (not nearly as many weak opponents as the international). For the sample of players, I have an average of 123 games. I did my calculations on the top 39 players by ELO simply because 40 is the number of player on one page of the TLPD (also, should include all code S plus the top of A); however, one of the players had only played against other Kespa players (no opponents in the top 40 SC2 ranks), making his dataset completely worthless (you'll see why soon) and was thrown out.
My apologies for taking so long between initial data gathering and finishing this article.
Starcraft: Brood War
For SCBW, I wanted a dataset that represented the golden age of SCBW without going too far into the past. Since I took data from the top 39 current ELO players (to match SC2), it would be very poor form to see how they did long ago. However, I didn't want to include the most recent data of hybrid proleague since that data is tainted by a lack of dedicated practice. Therefore, I only included data up to 2012-Apr-8 (last day of previous proleague). Finally, SC2 players play many more games than SCBW players in the same time period on average, so in order to get a reasonable sample size, I had to include a larger amount of time and went with approximately 15 months from 2011-Jan-1 to 2012-Apr-8. Even still, the average game count per player was 72. Hopefully a more stable metagame and more stable list of "best player" should mitigate some of the increased volatility introduced by the addition of more time.
tldr: I tried to compare the most current SC2 versus a relatively idealized time of SCBW.
For the chosen datasets, the top two win percentages for players were Flash (75%) and Taeja (69%). However, the top SCBW ELO was Fantasy (65%) and only Flash and Bisu (72%) had a better win% for that time period than Taeja. Also with less games for the SCBW guys (Flash: 111, Bisu: 97, Fantasy: 101) than Taeja (180 games), there is more room for volatility. Overall, I'd have to give the slight advantage to SCBW, but the results of any single player can be an aberration. So what happens when we rely less on a single player's data and we look at a group of top players as a whole?
The top 39 SC2 players by ELO had a record of 2832-1956 (59%).
The Top 39 SCBW players by ELO had a record of 1609-1194 (57%)
Advantage SC2.....err wait! The SCBW scene was much more insular. Top players had much less opportunity to play against outsiders and mostly had to play against each other while SC2 players have a few more open tournaments in which to pad their win% against lesser players. What to do about it?
Let's strip out all the data of top 39 players versus each other to see how dominant these guys are.
The top 39 SC2 players by ELO had a record of 1693-817 (67%) versus non top 39 players.
The top 39 SCBW players by ELO had a record of 783-368 (68%) versus non top 39 players.
Close? Yeah. In fact, I charted it for each individual player in the top 39. Here are their win percentages versus the outsiders.
![[image loading]](http://i.imgur.com/sXWiw.jpg)
Naniwa ruining everything at 8-0
You should note how close everything is from bottom to top. The graphs nearly touch at the beginning, middle and right before the end. Ignoring the Naniwa data due to an extremely small sample size, only at the tip top of BW stats does there seem to any real divergence where Bisu (84%) and Fantasy (88%!!!) put up some astounding numbers versus weaker players. Even the mighty Flash (81%) is in the same realm as TaeJa (80%) and Curious (80%) when it comes to defeating the weaker players.
If we limit the data down even further,
The top 10 SC2 players by ELO had a record of 736-424 (63%) versus non top 10 players.
The top 10 SCBW players by ELO had a record of 421-215 (66%) versus non top 10 players.
Still quite close, but with just a slight advantage to SCBW. So what we've determined is that the top players in both games put up very similar numbers against weaker players except at the very extreme top of BW. However, we can still be tainted by very weak competition. And in either case, it is possible that less than 40 people have hit the soft cap in SC2 thus far so we could be looking at two reasonably difficult soft caps, but due to SC2's young age, still vastly different soft caps.
What about play within the top 39? If the soft cap is sufficiently high or the road to the hard cap is still reasonably steep, shouldn't there be more variety within the top 39 than in a lower soft cap game? If we only take the wins and losses versus other members of the exclusive 39 club and then graphed everyone's win percentage, a greater slope should show that there is more ability for players to separate themselves. Essentially, a greater slope should show that the game requires more skill. Thankfully, I already did the work for you.
![[image loading]](http://i.imgur.com/VcbEk.jpg)
Flash: The Peer Pulverizer
Here we see a definite trend in favor of SCBW. While the graphs are extremely similar in shape, the bottom is lower and the top is higher for SCBW. The reasonably smooth slope in the middle of both graphs is also greater on the SCBW side. Perhaps this isn't a huge indicator that SCBW has a greater skill curve, but at least it's something to point to.
I tried.
I really did try to find data that would support the SCBW claim of superiority because all the data I've seen up to this point shows SC2 having an extremely similar difficulty level. I really had to massage the data to finally get just little bits and pieces where you could claim SCBW requires more skill, but even those bits could be the result of small/oddball sample sizes. Like, if we redo that last chart for only games within the top 10 players, we get this:
![[image loading]](http://i.imgur.com/XpJMX.jpg)
goodbye conclusions
Mitigating Factors
Having said all that, there are mitigating factors on both sides.
- The datasets that were chosen were reasonable in my mind, but are still at least questionable. If we calculated out the old ELOs from a pre-sAviOr scandal era and then used that top 39, perhaps we'd get different results. If we did these calculations 6 months ago or 6 months in the future for SC2, perhaps we'd get very different results. Hell, we might get different results pre-Queen buff versus post (our set includes about equal time for both).
- Starcraft 2 is also a new game. Perhaps it takes approximately 3 years of dedicated practice to hit the soft skill cap in SC2, but it takes longer to hit it in SCBW. It'd be very difficult to tell until we hit that point in SC2's lifetime. All we know is that at the very least, the soft cap in SC2 has thus far allowed people to excel in a relatively equivalent way to BW.
- Starcraft 2's metagame is changing multiple times in a year. In BW, there are players who play stylistically, but the basic reasonable options haven't changed in over a year. In SC2, the TvZ metagame has drastically shifted in the last 4 months and players who were destroying TvZ before are struggling now and vice versa. This would tend to force win percentages towards 50% as players switch over from having a winning strategy to a losing one. This is but one of many metagame shifts in the last 8 months which would normalize win rates. It is likely that in a more stable environment, better players would be able to prove themselves by a larger margin.
- If you believe in the superiority of Kespa players, then it goes to reason that their playing SC2 should serve to increase the slope of the curve in the top 39 vs top 39 chart. If you actually believe that the Kespa players at the very top are the combination of hard work and true talent while the SC2 guys were the middling people who couldn't excel in SCBW, then we're in for a treat. Thus far, the "middling talent" of SCBW has been able to differentiate itself in SC2 approximately as much as the "real talent" from SCBW ever did. If the elephant argument is true, then SC2 might actually have a deeper skill curve than SCBW. Wouldn't that be interesting?
- The final mitigating factor I can think of right now is the racial matchups. By their nature, some matchups may be more skillful than others. PvP in SC2 is considered to be a coinflip matchup by many and, if true, would normalize the win rates of all protoss players closer to 50% as it drags down the best protoss players and bumps up the weaker ones. The same was often said of SCBW ZvZ.
Why the Criticism?
So why is there so much criticism towards Starcraft 2? Why do we hear all the "easy game" comments? The first reason I don't want to dwell on too much is simple nostalgia and the unwillingness to accept change. Sure, we accept some change if it is part of the system, like a changing map pool, but changing units, the interface, and the physics engine are large changes that aren't part of the standard system. Those changes are quite jarring and scary and rejecting them is a normal human reaction, no matter if the new things are as good or even better than the previous. This is far from a unique trait of the SCBW community and is instead something that entire books could be written about for humans in general. We're already seeing the negative reaction to the changes proposed in HotS and I guarantee that we'll see the same negative reaction as LotV announcements are made. I'm sure the same thing actually happened when Brood War came out too (though the competitive scene wasn't too big at that point), unfortunately all we're left with now is a humorous spoof.
Perhaps one of the key differences between SCBW and SC2 is the way in which a player shows his superiority. A lot of people point to expert muta control as a defining factor in many top notch BW zergs. It is a visually obvious skill when the ball of mutas is staying tight and maximizing it's DPS on a point while minimizing enemy DPS. There are many other difficult micro tricks like this (reaver-shuttle, etc) that are very visually obvious high skill achievements. In SC2, many of the equivalent actions are not so obvious. Using force-fields perfectly to hit the right spots at the right times to get the absolute most out of your energy is an art form as skillful and beautiful as muta-stacking. Yet, the casual observer won't notice that the force-fields should have been one space further back or are overlapping by one space in a few spots. The elite protoss in SC2 will have extra force-fields to spend per the gas investment because they place their fields better, but it's so easy as an observer to overlook it and it's not really something a very knowledgeable caster can dwell on for too long as the action is pretty much guaranteed to get very hot.
![[image loading]](http://i.imgur.com/kUI72.jpg)
two all-important force-fields you won't have later
Another obvious difference is the macro mechanics. In BW, the mechanics are so hard that even at the pro level you will see obvious mistakes. In nearly every pro match I've seen, there have been idle workers for long periods of time... a clear mistake that the best pros make less often. In turn, the best pros tend to have bigger armies than their opponents even at the professional level. You'd almost never see that in SC2. Bigger armies in SC2 pro games pretty much only come from different strategic choices or previous damage dealt. You would have to assume that the difference in macro mechanics should lower the soft cap in SC2 compared to SCBW, but there's a possibility that something else is happening. The extra time given from easier macro in SC2 doesn't have to go to waste.
I believe that in both SC2 and BW, the soft cap is not being hit. Whether the cap is higher or lower doesn't actually matter if no human being can reach the lower of the two. Whatever time is saved by easier macro is instead shoved into other important soft-cap activities. Quickly splitting units that want to clump up takes a ton of skill. Likewise, creating arcs, focus firing, perfect spell casting, and pulling back injured units all come to the forefront. If you can't just out-macro your opponent, you need to be able out strategize your opponent or make some very small subtle moves that have an extreme impact on the game. And the players at the top have been able to consistently use better strategy and make those subtle moves. It's a skill and it has a massive impact on professional win-rates, it just isn't quite as obvious of a skill as others.
I believe that as SC2 gets more and more developed, we'll see even better control in those extremely intense moments right at the beginning of a battle. The people who can really handle those situations will excel and dominate the scene while everyone else will be held back. Meanwhile, the hardcore observers of the game will notice the little subtleties and truly appreciate top notch plays that they could never make while the casual fan or even BW elitist misses the subtleties and thinks the game inferior.
The Conclusion
as close to a TLDR as you'll get
Starcraft 2 cannot be called an easy game without also calling Brood War an easy game. The relevant numbers are too similar to be able to make a distinction between the two. Instead, I believe the "easy game" complaints may be based on nostalgia or they may be based on the way in which skill is demonstrated in SC2 versus SCBW.
This article does not claim that either game is superior. Nor does it even claim that the methods that SC2 uses to demonstrate skill are as good as the ones used in BW. It simply states that the methods to excel exist in a relatively equivalent amount between the two games within the limits of the human ability to learn.
So the next time someone calls SC2 an easy game, you can just link them to this blog.
Answering the Criticism
On September 13 2012 05:54 Itsmedudeman wrote:
BW pros caught up in 3 months. Do you really believe SC2 players could do the same in BW?
BW pros caught up in 3 months. Do you really believe SC2 players could do the same in BW?
The implication is that BW must be more difficult because SC2 pros could never do the reverse. However, if SC2 pros were able to take their current mechanical skill and be teleported back in time to the early 2000s, yes, I think they could compete within 3 months. Due to the constantly shifting meta-game in SC2, every single non-kespa pro has only had a few months inside the current meta-game... just like Kespa pros. The time they have spent on previous meta-games has become irrelevant because better strategies have been figured out. So the time that someone put into 5-rax reaper (for example): learning how to do it, optimize it, how to counter basic defenses from a strategic and tactical perspective, and learning how to transition out of it is all wasted time. Blizzard patched that one completely out. Other strategies have simply been figured out.
From the beginning of the game until approximately 6 months ago, only the mechanical skills non-kespa pros worked on is relevant. And as everyone is aware, the Kespa pros had plenty of mechanical skill to transfer over. So it's not a shock that some Kespa pros were able to make a smooth transition to a new game. Yet, if we reversed the situation, the non-kespa pros would have to compete against years of legitimate meta-game practice that is still relevant today. The situations are not equal.
On September 13 2012 06:31 Birdie wrote:
This, so much this. There are plenty of other things I could use to argue that BW is a harder game but this is the killing blow right here. Both games have skill ceilings that have not yet been reached but the low skill floor required to be able to play BW and do the same things as in SCII is much higher. You pretty much require 100+ APM to be able to macro and micro simultaneously even a little bit in SCBW; in SCII you can do it with 60 APM.
Your post just shows that among top BW players, the differences between them are statistically the same as the differences between top SCII pros. ELO is an indicator of relative skill among players, and the ELO curve of one game should correlate very closely to the ELO curve of another, becase that is how ELO works
So no one misses it:
THE ELO CURVES OF TWO GAMES WILL ALWAYS BE ALMOST THE SAME
This, so much this. There are plenty of other things I could use to argue that BW is a harder game but this is the killing blow right here. Both games have skill ceilings that have not yet been reached but the low skill floor required to be able to play BW and do the same things as in SCII is much higher. You pretty much require 100+ APM to be able to macro and micro simultaneously even a little bit in SCBW; in SCII you can do it with 60 APM.
Your post just shows that among top BW players, the differences between them are statistically the same as the differences between top SCII pros. ELO is an indicator of relative skill among players, and the ELO curve of one game should correlate very closely to the ELO curve of another, becase that is how ELO works
So no one misses it:
THE ELO CURVES OF TWO GAMES WILL ALWAYS BE ALMOST THE SAME
Not true. Tic-tac-toe clearly will not have the same ELO curve. I'd also throw in nearly any game with inherent luck, such as Poker, not being the same. Superior players in poker will lose to inferior players far more often than a superior player in SC2 or other skillful games. Hell, I could play heads-up against the best poker player in the world and have a legitimate shot at busting him despite only being a very mildly profitable player. I have no chance against any pro Starcraft player. However, over time and thousands of hands a poker pro will be able to take his slight edge and work it into consistent profits.
Alternatively, the best runners/swimmers in the world win much more consistently than even the best Starcraft players. How often does the 8th best swimmer in the world win a match against the top 7? Essentially never. Now look at Starcraft 2. If a tournament was held between Taeja, Squirtle, HerO, Seed, Curious, Life, Creator, and Marineking, the odds of any one of them winning are reasonably good. The curve in swimming/running would be much steeper than SC2 or BW.
If you can accept that tic-tac-toe, poker, and running/swimming will not have the same ELO curve as SC2 or SCBW, then you should be able to understand that games will have different ELO curves and your basic argument is flawed. I would welcome anyone who would like to repeat this data collection for something like Chess or Running to show how different the curves are.
On September 13 2012 06:55 Birdie wrote:
But that still means that the OP's 'statistics' are irrelevant.
I don't think anyone can deny these points:
1: BW has a higher mechanical skill floor to play at what is considered a basic level.
2: BW has a higher mechanical skill ceiling.
3: Both SCII and SCBW have theoretical skill ceilings for general skill (not just mechanical skill) than what has been reached so far.
But that still means that the OP's 'statistics' are irrelevant.
I don't think anyone can deny these points:
1: BW has a higher mechanical skill floor to play at what is considered a basic level.
2: BW has a higher mechanical skill ceiling.
3: Both SCII and SCBW have theoretical skill ceilings for general skill (not just mechanical skill) than what has been reached so far.
The first two points are both theoretical. Even without evidence, I agree with the first. However, I disagree with the second. SC2 is a faster game due to things like the Terrible Terrible Damage philosophy. At a theoretically perfect level, the amount of APM needed to perfectly control an army in SC2 or BW dwarfs the amount of APM needed to macro perfectly in either game. When you hit those theoretical levels, the speed of SC2 likely gives it a higher hard skill cap, but both are nearly infinite.
Point #3 is the most important point. It is the part that gets away from theoretical and back into reality. In both games, not even the soft cap on skills is being reached. If you want to claim that BW has a higher theoretical cap and is therefore "more difficult", you can go right ahead. Unfortunately for you, you're then living in theoretical world, not real world. In the real world, players aren't hitting any caps and their ability to perform relative to their peers remains equal in both games. For all real-world practical applications, SC2 is just as hard as SCBW based on all current evidence.
mechanical skills in BW > SC2
As a general response to many people. Since when does mechanical skill of just a few portions of a game equate to the be-all-end-all of a game? The skill in a game such as SC2 or BW is about much more than just mechanical skill. And from many of the responses I'm seeing, it doesn't appear that people are getting that point. If you only take bits and pieces of a game, you can make an argument in favor of BW, but when you look at the skill of a game as a whole and how it relates to actual statistics, SC2 stacks up quite well to BW.
)![[image loading]](http://i.imgur.com/F6WAz.png)
