Hey folks,

I’m brownbear. Today I’ll be responding to a discussion on practice and skills development between Artosis and NoRegreT. I’ll contextualize my comments with quotes from the video so it’s not necessary to watch before you read this piece. However, I recommend that you do check it out at some point – it was a fascinating discussion.

Before I get started, let me emphasize that this is an editorial piece. It’s a bit rambly! I’ll be walking through my own opinions on this subject and talking through some anecdotes from my own life. This is by no means a rigorous scientific analysis of what it means to become world-class at something. If that’s what you’re looking for, I recommend you stop right here and turn instead to the wonderful book Peak.

Finally, before I dive in, please consider giving me a follow on social media! It’s a good way to keep up with the content I’m creating, which nowadays focuses heavily on real-time strategy games.

• Twitter – https://www.twitter.com/brownbear_47
• Facebook – https://www.facebook.com/brownbeargaming

The Question

The kick-off to the video is the following question posed by Artosis (timestamped link):

If you were going to help someone, coach somebody, to become the best pro-gamer, the absolute best, Serral-level professional gamer, championship winner, how would you go about doing that?

As with any discussion, the conversation moved from topic to topic and the key answers to this question are sprinkled throughout the video. I’ve done my best to assemble them here:

• If possible, starting at a young age is ideal.
• Play plenty of games – 2-10 hours a day (NoRegreT) or two three-hour blocks totalling six hours of play per day (Artosis).
• Focus 100% on the task-at-hand and avoid distractions while laddering – “stare at the queue” instead of looking at social media.
• Seek outside opinions on your play to uncover blind spots and weaknesses.
• Understand the full spectrum of the game – both the expected trajectory from beginning to late-late-game, as well as all the many corner cases around it (see this graph from Artosis).
• Never rationalize losses as being out of your control.
• Ideally, compete with yourself rather than with others, because the former is both more satisfying and more sustainable. Per NoRegreT, this may not be achievable.

This is a fantastic list of answers. I had a few thoughts in response which I wrote up below.

Process vs. Results

One thing that stood out to me was the exclusive emphasis on process and subsequent lack of emphasis on results. Many goal-setting exercises work differently, working backwards from the results you’re trying to achieve. You setup milestones, estimate how long it’ll take for you to reach them, and then course-correct as you succeed or fail. Generally speaking you wire this up to “fail fast”, such that if your process is not working for you, you know that sooner rather than later.

For example, when it comes to weight-lifting, you might strive to hit the following targets for a one-rep max on your bench press (taken from the legendary Fuckarounditis article by Martin Berkhan): 1.2x your body-weight after two years of training, 1.5x your body-weight after five years, and up to 1.8x your body-weight after ten years. If you’re not trending toward these targets, you’re likely not training correctly.

I actually completely agree with the implicit idea that this style of goal-setting is unhelpful when it comes to StarCraft II. But I think it’s interesting to think about why, because a system like this is generally what I would advise to someone looking to improve at a skill.

First, progress in StarCraft is non-obvious. The game is full of tiny details and you sometimes don’t even realize that you’ve fixed something until you watch someone else play, see them do what you used to do, and recognize how it sets them up for trouble. Mechanical improvement is also hard to measure – APM, EAPM, and time-to-first-action can somewhat approximate it, but not consistently enough to be helpful.

A results-driven training regimen in StarCraft would quickly become frustrating. I myself have tried it multiple times, and each time gave up in frustration after feeling like I had gotten nowhere after months of hard work. This despite the fact that I reached all-time high MMRs throughout my training and at my peak maintained a 5K+ MMR on NA – after being a legit silver league player on the launch day of Legacy. I just didn’t feel better in the same way that you feel stronger when you increase your lifts or improve at a more predictable task. And this discouraged me from sticking with my training regimen.

A second issue is that StarCraft is a highly mental game. With weight-lifting, what you’re thinking about while you’re lifting is not super relevant to your gains. Of course there’s factors like form and consistency in going to the gym, but from my perspective, as long as you’re doing the right lifts at an appropriate weight, you’re on your way.

With StarCraft (as Artosis mentions later on), your level of focus while you practice is directly related to how much you gain from that practice. Anything that disrupts that concentration – like falling behind on an MMR goal or failing to qualify for a tournament you vowed to play in – is counter-productive and hurts your ability to improve.

I would go so far as to say that if I ran a team house, I would actively avoid setting results-based goals for my players. Instead I would design what I consider to be the perfect daily schedule for each individual player – in terms of everything, sleep, exercise, diet, training times, replay analysis, taking breaks, etc – and measure how close each player gets to hitting that schedule every single day. Not only does this guarantee that they’re doing their absolute best in terms of getting better at StarCraft, it’s also very healthy mentally. Whether or not you hit your “goal” is a matter of what you do every single day, it’s not dependent on tournament results or balance patches or anything else.

On Massing Games

Artosis and NoRegreT both agreed that playing professional StarCraft requires a ton of game experience. I agree! I just want to add some data that I think is relevant. Below I’ve broken out the games-played-per-day at the 90th percentile by region and by league, as of the current season. The way to read this data is that if you see “NA [Master’s] = 10”, that means 90% of active Master’s accounts in NA play fewer than ten games a day, while 10% play more. The analysis is done this way because calculating actual games-played-per-day-per-player is not possible because of account sharing and multiple accounts per player:

Region + League	Games Per Day [90th percentile]
KR [Grandmaster]	10.3
KR [Master]	3.8
KR [Diamond]	2.9
KR [Platinum]	1.9
NA [Grandmaster]	8.8
NA [Master]	3.8
NA [Diamond]	2.8
NA [Platinum]	2.3
EU [Grandmaster]	9.1
EU [Master]	4.1
EU [Diamond]	2.6
EU [Platinum]	1.9

Unsurprisingly, players in higher leagues play more than players in lower leagues. Actually, from my perspective, a reliable way to hit lower-level goals (e.g. Diamond, Platinum, etc) is to set a daily-games-per-day target based on this data. For example, if you’d like to be in Platinum in the US, strive to play between 2-3 games per day consistently for a few months.

More interesting is that 90th percentile Grandmaster players in KR play around 20% more than their peers in NA and EU. This is something I’ve repeatedly observed over the years, and I think it helps explain the continued Korean dominance of high-level StarCraft, even if the absolute best player in the world is now from Europe. High-level Koreans just play more than their counterparts around the world. There’s also long been anecdotes that Koreans play more custom games than their peers, though this may no longer be the case.

On Age and Race Selection

NoRegreT said something really interesting here, and I agree with his overall point (timestamped link):

I feel like, if we took a Serral, and we gave him a different race, he would have a lot harder of a time being the global champion. I think if he were a Terran player, he could be abused a lot more. People could find a lot more weaknesses. The problem with Serral right now is that his strength is just his ability to be a wall, to know everything, to scout everything.

What this makes me think of (though I’m not sure if this is what NoRegreT is referring to) is the difference between “hard work” and “necessary work”. As Artosis points out later in the video, it’s surprisingly easy to go on auto-pilot and grind out a ton of time on the ladder. What’s not easy is identifying and resolving the gaps in your play. Even harder is learning to think about the game in an entirely different way.

Let me try to explain with a personal example. One concept that took me a long time to learn was the importance of subtle differences in build orders. “Back in the day”, I spent thousands of hours playing competitive Age of Empires, where randomly generated maps and a generally broader style of gameplay (more races, units, technologies, etc) led to a less-refined competitive meta where players spent a lot of time thinking-on-their-feet and strategizing as they went along. There were concretely defined build order traversals but they weren’t as strict as they are in StarCraft II, or at least I don’t remember them as such in Age of Titans and Age of Empires III.

Build orders in StarCraft II are tight. There is a big difference between building your third command center off of 1-1-1, 3-1-1 or 5-1-1. Your upgrade timings are different, your key tech comes out at different times, you have a wildly different ability to defend different kinds of attacks, etc. It’s more than just having a few less units, though even this is significant; because battles in StarCraft II snowball a bit more than battles in older real time strategy games (especially at my pleb level), a slight supply or tech advantage can turn into a walloping engagement.

I can’t tell you how many TvTs I lost because I went defensive 5-1-1 before 3rd CC “to be safe”, and then got steamrolled by someone who maxed out 30-60 seconds faster than I did with faster upgrade timings to boot. 5-1-1 before 3rd CC demands aggression!

There’s tons of paradigms like this – the importance of fighting on or off creep, the importance of abusing asymmetric interactions, the importance of taking the initiative at the right time, and so forth. Each race emphasizes different paradigms and playstyles. What I found really insightful about NoRegreT’s comment is this: if a player naturally trends toward and understands the key paradigms that happen to align with the race of their choice, they are going to perform better than if they were playing another race that works differently. They are more likely to see what they’re doing wrong and understand intuitively what they need to fix because the playstyle is a natural fit for them. I agree completely that Serral would probably be worse if he played Terran, not because Terran is worse than Zerg, but because what Serral is good at is just not as useful for Terran as it is for Zerg.

One idea that gives me pause is the intuitive notion that players already select the race they’re most suited to. But I’m not sure if this is the case. I feel that players aren’t exercising most or all of the game’s mechanics until somewhere around Master’s league, when build orders and meta-game and basic tactical play matter more than just building a crap-ton of units. Getting to Master’s will take most people several hundred games or more. By the time you get good enough at a race to get there, you might discover that that race is ill-suited to how you naturally like to play; but you might also find the cost of switching races too high, leaving you a bit stuck.

You could also argue that races are flexible enough to accommodate different playstyles, which I generally agree with. But, I would argue that aligning with a race’s natural strengths is going to work better at the highest levels of professional play.

On the Importance of Outside Opinions

Artosis said something really insightful here (timestamped link):

There are so many ways to look at StarCraft. If you are trying to do it all on your own after a certain point – I think there is something to be said for learning to a certain degree on your own, because then you kind of develop a firm understanding and you’re not just copying what other people say – but, yeah, you should be looking for as many different viewpoints. A lot of times, two people that know what they’re talking about will look at something completely differently.

This reminded me a lot of the various review processes in software engineering – requirements review, code review, technical design review, etc. One thing I’ve discovered over the years is that no amount of training courses, textbooks and classes can ever truly replace a review by your peers and seniors. They just think differently than you do and raise up things that you may not even have thought of. Also important is the way peer review forces you to explain not just what you’re doing, but why you’re doing it that way. The why question forces you to validate assumptions, crisply understand the problem you’re trying to solve, consider the pros and cons of alternatives, and so on.

I think the same applies to StarCraft. Later on in the video, NoRegreT and Artosis discuss what SpeCial could have done to beat Serral in the prior WCS tournament. NoRegreT pointed out that some of the weaknesses of SpeCial stem from how he plays. I agree; to flesh this out, I took a quick look at SpeCial’s recent series vs. Reynor and documented the openers he used:

• Game 1 – Win – Helion Marine into Cyclone + Blue Flame Push
• Game 2 – Loss – Helions into quick BC
• Game 3 – Loss – Proxy marines into proxy reapers
• Game 4 – Loss – Helion Drop

Game 3 is the one that stands out to me. SpeCial secured a bunker and successfully challenged the natural hatchery, putting him in a good position. Instead of doubling-down on this with the more common approach of more bunkers and marines followed by a factory at home, SpeCial gambled and switched into reapers. This didn’t pay off, and turned what arguably should have been a quick win into a late-game brawl and a loss.

The interesting thing to me is not what SpeCial is doing, because I can see the internal logic of his strategy (thanks to Max for also pointing out the backstab potential if Reynor teched into roaches). What’s interesting is why he’s doing it – why not play the “stronger style” (in my opinion) in a high-stakes tournament? This is the kind of thing that I think a coach can help identify and do something about. It reminds me a bit of Maru’s earlier professional experiences; there his teammates explicitly stated that reaching out for help produced good results, and it’s implied that this is at least partly what allowed him to reach the next level of play.

On Quality of Practice

Artosis regularly brought up an interesting point about quality of practice. To paraphrase, he noted that focusing 100% on practice is significantly more efficient than constantly diverting one’s attention to other tasks, like reading the forums or posting to social media.

I agree. Here’s how I think about this in my own work: while your subconscious mind does the heavy lifting in terms of performance, it works more efficiently and draws stronger, more relevant connections between thoughts by having space in consciousness. This idea stems from a book I once read regarding the nature of consciousness; it argued that consciousness serves as a “working space” or “blackboard” for subconscious thought processing. AI researchers have been trying to leverage the same concept to see if it improves the performance of artificial intelligence.

Let me try to explain with an example. When I study Chinese characters on a study app, it’s easy to go on auto-pilot: I see a prompt (e.g. gòngxiàn – to contribute, to dedicate, to devote) and I immediately begin to write the character (貢獻). I can review hundreds of characters in this way without breaking a sweat. But there’s a catch – after awhile I discovered that my recall of these characters was quite poor, and if I had studied something more than six months prior, I was unlikely to be able to write it without a hint.

To fix this, I started to do something similar to what Artosis recommended: instead of blazing through my study sessions, I stopped and first visualized each character in my mind, drawing it out stroke-by-stroke, before then inputting it into the app. This required focused concentration. Honestly, it was painful and slow at the beginning! But I discovered that this process significantly improved my ability to recall characters (as-measured by how long it takes me to complete my daily review of previously learned characters + learn a few new words). Of particularly note was that I much less frequently confused characters with one another, even when they were very similar.

When it comes to StarCraft, I think the most relevant distraction to professional players is not day-dreaming or checking social media or reading Reddit. It’s streaming! This is a pet theory I’ve held for awhile now: streaming is a low-quality form of practice. The combination of commentary and reading and responding to chat disrupts the learning process and makes it harder to get better. In an attempt to gather some data on this, I decided to compile statistics on the most recent WCS Summer tournament. Specifically, I compared the placement of players in the Ro24 with how many hours they’ve streamed in the past three months (as of July 13; full spreadsheet here):

Player	Placement	Hours Streamed (last 90 days)
Serral	2	0
Elazer	5-8	0
Nerchio	9-16	0
Clem	9-16	0
Scarlett	9-16	0
souL	17-24	1.4
Reynor	1	2.6
Hellraiser	17-24	2.6
Neeb	5-8	8.9
SpeCial	5-8	17.7
ShoWTimE	3-4	19.4
Denver	9-16	20.1
SortOf	9-16	21.9
ShadowN	17-24	23.7
GunGFuBanDa	17-24	23.9
PtitDrogo	9-16	40.8
Cham	17-24	49.9
uThermal	17-24	78.8
TLO	9-16	92
Mana	9-16	96
Harstem	17-24	253
HeRoMaRiNe	3-4	277
TIME	5-8	N/A
DNS	17-24	N/A

HeRoMaRiNe is the big outlier – both the most hours streamed of anyone and a high-ranking placement (top 4). But overall, there’s a consistent pattern: top-ranking players stream less or not at all compared to lower-ranking players. And I think this comes down to plain old time considerations. It’s hard to practice full-time *and* also a stream a lot; the easier approach is to just stream your practice games and combine the two into one activity. This worsens the quality of your practice and ultimately degrades your performance in tournaments.

Note though that there’s a lot of caveats to this. First, this is more “data as entertainment” than it is “data as evidence”; to really measure the impact of streaming on performance, you’d need to do a more rigorous analysis. Second, cause-and-effect here is not obvious. A lower-seeded player might simply accept that they are not going to win a lot of money from tournaments, and in response stream more because they view that as a better use of their time. Third, full-time streamers have pointed out that they wouldn’t play as much if they weren’t having fun interacting with chat and putting on a show, and as a result would likely not be as good at the game if they only played by themselves.

On Being Satisfied

My experiences align with NoRegreT and Artosis’s examples here of how reaching a goal hampers continued progression (timestamped link). For me, my StarCraft progress has often been hindered by my own expectations of how good I “should” be. At some point in the distant past – perhaps even subconsciously – I decided that’s Master’s in Korea off of a pure macro style (no cheesing). I don’t really know why this is satisfying to me, but I suspect it’s the nomenclature “Master’s”.

Anyway, every time I’ve tried hard to grind at StarCraft, I’ve always stopped soon after I consistently maintained a Master’s MMR. Similar to Artosis’s wish to be a billionaire, my wish to be a KR GM Terran is just that – a pipe dream that I will almost certainly never fulfill, because I’m not actively working toward it.

There really is something to the idea that to become the best, you can never be satisfied. Your goal must be the right blend of “just attainable enough that you won’t get discouraged” and “unattainable enough such that it won’t lead you to become satisfied and complacent”. Because if you reach what you were striving for, your motivation to train will dip and eventually impact your performance.

Artosis brought up a fascinating point in this regard (timestamped link):

I think that you can become really, really good, and I think it’s maybe the best thing that you can do, is try to just to make yourself improve everyday. If you’re just like, “all I care about is me getting better, and if I’m playing better today, my ladder rank doesn’t matter, who I’m beating doesn’t matter, who’s beating me doesn’t matter”, because you can’t control how good other players are.

Man, I’m on the fence about this one. On the one hand, it’s a very sustainable form of skills development. However, for me personally, I think I would get lazy with this approach. For example, with my Chinese practice, I’m striving to take and pass certain certification tests. Each certification corresponds roughly to a high-level assessment of language ability. I have a general idea of where I want to be at what time, and I’ve set my study schedule accordingly. I often fall behind! But I think having a goal keeps me more honest than simply being “better” at Chinese than I was yesterday. I could get better every single day, and still be functionally useless after years of study.

I think there’s a slightly lesser point that would be easier for me to accept: it’s better not to set goals that are determined by other people, since you can’t control what other people do. In this sense, competing with others and setting goals like “round of 8 in GSL” may be counter-productive because you have no control over how hard other players practice and study. Similar to the whole “process vs. results” thing, “process” is internal while “results” is external. I think this is what makes focusing on process a better tool for self-improvement.

Final Thoughts

This was a fascinating discussion! I highly recommend you give it a watch if you haven’t yet seen it. I often feel like every new aspiring professional independently rediscovers a lot of the lessons discussed in this video. Imagine if they started with this knowledge from day one!

Thanks for reading and I hope you enjoyed this piece! I’m super interested in this topic and would love to hear what you think. And, as always, I would love it if you followed me on Twitter and Facebook and checked out my YouTube and Twitch channels to receive regular content updates. All the best and see you next time.