Hey folks, I’m brownbear. Today I’ll be writing about a time I dedicated a hundred hours to doggedly improving at StarCraft II, only to find myself playing noticeably worse by the end.
Over the years of playing StarCraft II, I’ve gone through several spurts of dedicated ladder grinds. The way this goes is that I’ll start out with some kind of goal – say, one or two league tiers of improvement – and patiently work towards that goal over the course of sixty to ninety days. Typically this involves a combination of practicing build orders, watching streams of professional players, studying tournaments, and, of course, playing plenty of ladder. After about ninety days, my wrists start to feel sore on a daily basis, and I find it difficult to continue. So, I celebrate how far I’ve gotten, end my grind and move back to co-op or team games.
Each time I’ve done this, I’ve gotten significantly better at StarCraft II. Except for last time.
Back in November I kicked off yet another StarCraft ladder grind. I played hundreds of games, sticking with the program for a solid sixty days. I watched pro-level streams, I reviewed my replays to identify my mistakes, and I stuck with a consistent daily practice schedule.
Where did I end up? Down. I ended up a hundred MMR down.
Wow. I’ve been grinding RTS games for more than fifteen years, and I’ve never attempted to get better and not succeeded. That’s not to imply that I’m the best player in the world at whatever I try my hand at; rather, it’s simply to say that I’ve always been able to improve my performance. Maybe that means something as small as moving from Gold to Platinum – it’s not a question of where I end up, just a question of it being higher than where I was before.
That didn’t hold true this time around, a fact I found quite demotivating. Around the same time I was also getting excited about distance running (I wrote about this a few weeks back), so I had a convenient excuse to cut my grind short and focus more on running.
But I never stopped thinking about it. What made this grind different from all the previous ones?
Was I just better at the game? The last time I did a big grind, I peaked around 5.1k MMR. Many good players will tell you that improvement slows way down once you start reaching higher MMRs. Maybe I was getting better, but it was just hard to tell?
Nah. I watched my own replays to see what I could fix, and I noticed that the same mistakes came up again and again. I kept losing for the same dumb reasons, reasons that I was well-aware of, reasons that I would often passively *think* about as I made crucial, game-ending mistakes. The replays were clear – I wasn’t silently getting better. My decision-making was arguably getting worse!
Maybe I just needed to give it more time? But time is only a variable in an equation – it gets multiplied by the effectiveness of practice. “More time” implies that something about that time would make me better in the future, like improving my execution or learning from poor decisions and making adjustments. “More time” means that you would see improvement in the replays across some set of axes, like idle time, unit control, positioning, build orders, something. But if you see nothing, then all you’ve done is multiplied your time “variable” by zero, and ended up nowhere.
The Learning Curve
In moments like this, people (myself included!) often love to point to the learning curve, a graph that shows that skills development sometimes features both rises and falls in ability over time.
Here’s how it (sometimes) goes: when you first start learning something, your skill level rises rapidly as you learn the basic skills. As you learn more, you begin to understand the nuances, exceptions, and occasional conflicts within those skills. As you integrate these subtleties, your skill level drops, because you start to misuse skills you were using correctly beforehand. Eventually, once you finish integrating your newly learned skills, your skill level rapidly rises. The cycle repeats as you learn a new thing and subsequently the nuances of that new thing.
(If this is still confusing, think about it in the context of language learning. When you start out, you learn the most basic way of saying X. Over time, you learn more and more ways of saying X, with each new way often being more situationally specific than the last. At the start, you would have always used the basic phrase, and people would understand what you were trying to say. But now, you sometimes try to use the new phrases, and it confuses people because they’re situationally inappropriate. Your language ability has gone down, even though you know more.)
Was I just at a valley in the learning curve? It was a tempting explanation, but it was missing something. Generally speaking, once you know how this curve works, you start to see it happening in real-time – you recognize disparate or conflicting skills that require integration, and you expect yourself to temporarily get worse. But I didn’t see this in my play. It wasn’t like I was playing differently and needed time to integrate it into my normal rhythm. I was doing exactly what I had been doing before… just, somehow, worse, despite hundreds of games of practice.
The solution to my problem came not from StarCraft, but from Age of Empires. For context, I recently completed my retrospective on Age of Empires III. As I started working on my retrospective for Age of Empires II, I decided I couldn’t do it justice without getting at least decent at the competitive side of the game. So I sat down and started to grind.
This time around – remembering my failure to get better at StarCraft six months prior – I decided to approach the process more methodically. I created a detailed, organized roadmap for improvement. I created a practice schedule that I was disciplined about sticking to. I forced myself to watch the replay of every single game I lost, regardless of how painfully I threw them. And more than anything – I vowed to not give up the way I did with StarCraft.
The results were solid. I moved from around the 60th percentile to the 98th percentile in thirty five days. Not a big deal – when you start out low, you are bound to improve pretty fast. But there was a moment in the middle that revealed what had gone wrong in my previous StarCraft grind.
For about three weeks, I was stuck in the mid-1400s (~90th percentile). One day I’d break the 1500 ceiling, the next I’d tumble down to 1425. I had no idea what, exactly, I was doing wrong; all the games I lost felt different and disconnected from one another. The only clue I had was that I tended to win when I opened drush FC, and I tended to lose when I opened men-at-arms or scouts. (For those unfamiliar with Age of Empires II, these are just different openings that exercise different parts of the tech tree).
As I watched more replays, I started to realize that drush FC was not an inherently superior opener. What made it different was that it was much easier to execute, at least for me. See, when you open men-at-arms or scouts, you invest significant time and resources. This means you need to find some non-trivial amount of damage in order to justify your investment. This, in turn, requires competent micromanagement, positioning, map awareness, scouting – the works. And even if you do end up killing a few villagers, that’s all for naught if you can’t multi-task and simultaneously grow a clean economy back at home.
By contrast, drushing is dead simple – you send out a few units early in the game and wall up behind it. Your units don’t have to do much, because the investment is low and forcing idle time is enough value. You delay most timings (aside from a drush from your opponent) long enough that you rarely face any aggression without being fully walled off. Once you hit the Castle Age, you can grow comfortably into whatever composition is optimal for the match-up, react appropriately depending on what your opponent does (forward siege, mass knights, boom, etc), and just generally play a pretty straight-forward game.
Drushing was a really comfortable way of playing the game in the context of what I was already good at, because it rewarded general RTS skills more than other styles. Luckily for me, I had set out to grind specifically to learn the game comprehensively, so I managed to dial back my natural instinct to stick with what was delivering wins. Instead, I dedicated myself to men-at-arms and scout openings, doing nothing else for two weeks straight, taking the myriad losses in stride, recognizing that they would help me work on my weaknesses.
Soon after, I broke my plateau – I gained 200 ELO points, and reached the top 2%.
(For what it’s worth, I can point to the specific skills I learned from doing this; it’s not magic. I’ll make a video explaining this and other parts of my ladder journey once I get a little higher.)
The Other Curve
I realized that the classic learning curve is not some pre-ordained roadmap for learning new skills. It is just one of many paths. And looking back at my November grind, I had actually followed a much different curve:
For example, in 2019, TvZ moved from my best match-up to my worst match-up. For the longest time I couldn’t explain this – and no, balance is not a factor at my level. Eventually, I remembered a pivotal decision I had made early in my grind. At the time, Hellion / Liberator openings had become less common because Zergs had gotten pretty good at countering them. Banshee and Battlecruiser openings had risen in popularity in response; but I found those very difficult to pull off, because I hadn’t practiced them as much. My execution was quite poor, so every time I played Banshee, I would lose, whereas most times I played Liberator, I would win.
And it’s here where I made the crucial mistake – I prioritized winning over improving.
I feared the work it would require to learn a new style. I dreaded the tedium of practicing builds over and over against the AI, the way I had when I first learned Hellion / Liberator. And I hated how punishing it was to miscontrol a Banshee and lose it for nothing.
So I didn’t do it. And in the short-term, this was always the correct decision – anytime I went Banshee, I would lose. But it came with a long-term cost that I wasn’t working on my weaknesses. It’s not like Banshees are some kind of unicorn with no applicability to anything else in the game; they emphasize multi-tasking, unit babysitting, and more than anything else teach you not to throw units away that might be useful later. These were all lessons that I missed out on because I failed to force myself down the learning curve. I thought that the learning curve was an inevitability, but I realized that it was actually a conscious choice: do I do something that’s uncomfortable that will eventually push me to be better? Or do I prioritize winning right now, even if it doesn’t help me later?
I chose the latter, to my detriment. Sadly, the problem with this went beyond failing to learn new skills. It also infected all other aspects of my play. One of the most consistent reasons I lost games was the decision to all-in once I had a lead instead of transitioning into later-game tech and upgrades and “getting more ahead”. This was easy to justify in any individual game. If I spend a bunch of time and attention on this macromanagement that won’t pay off for another five minutes, my opponent could surprise me and steal the game away. But if I attack, I can capitalize on my advantage.
In the long-term, it’s better to play clean and always transition, always stick to your build, always make the optimal choice, regardless of how much you’re winning or losing. But in the short-term it will cause you to lose more, because you don’t magically gain actions per minute when you start doing additional tasks. And when you apply this kind of short-term thinking to your entire playstyle, you start to play like garbage – your skill level starts to go down.
There’s a train of thought that I see quite often in the StarCraft community, which goes something like this: in order to get better, you should find a style that works for you, and work on refining it. It’s reasonable advice! It’s easy to get overwhelmed by the sheer complexity of StarCraft; focusing on one thing can help keep the chaos at bay. And anyway, the skill ceiling in the game is so high that whatever you go with will feature an endless amount of depth and learning.
But this approach doesn’t work for me. For me, “refining my style” too quickly becomes “do what’s comfortable and easy”. And in thinking over my more successful ladder grinds, I realized that I had forced myself not to do that. For example, I hit Master’s just a few months after I started playing the game, relying on aggressive builds and two-base all-ins. But following that, I doggedly played pure defensive macro games, recognizing that failing to address this skills gap would cause me to plateau. This was painful – imagine being in Diamond for months after you found it so easy to get into Master’s – but it made me into a better player.
I remember when I plateaued in mid-Master’s in 2018. Did I keep doing what I was doing? No – I went out and grinded the meta-game build orders, even the ones that I felt very uncomfortable using. I gave up on 2-1-1, even though I had done it hundreds of times, and forced myself to play Hellion / Liberator, the meta build at the time. My MMR dropped, temporarily – but slowly but surely, it came back, and then started to rise to new heights.
In each of these cases, as soon as I stopped pushing myself out of my comfort zone, I hit a plateau. And calling it a plateau is generous, because what came next was always decline. This was in part due to the cancerous nature of a short-term emphasis on winning, but it’s also a practical issue. StarCraft II is not a static game – it features periodic balance patches, a still-evolving meta, and a regularly changing map pool. The longer I stuck to what I knew, the more my playstyle became outdated and countered by modern ways of playing the game. And so, over time, I didn’t just stagnate. I got worse. And this hurt my confidence and desire to play, which in turn hurt my performance, which again hurt my confidence – it was a vicious cycle.
One Weird Trick To Master Anything
I feel like this is some kind of rule of improvement: in order to get better, you must feel uncomfortable. And if you get to a point where you feel comfortable, you need to force yourself out of that situation. Here I’ll quote from one of my favorite books on skills development, Peak (emphasis mine):
This is a fundamental truth about any sort of practice: if you never push yourself beyond your comfort zone, you will never improve. The amateur pianist who took half a dozen years of lessons when he was a teenager but who for the past thirty years has been playing the same set of songs in exactly the same way over and over again may have accumulated ten thousand hours of “practice” during that time, but he is no better at playing the piano than he was thirty years ago. Indeed, he’s probably worse.
There’s a potentially dire implication to that, though. Doesn’t this mean that your entire improvement journey will be uncomfortable? That the moments of pain and frustration will vastly outnumber the moments of satisfaction and achievement, by design? Is it worth subjecting yourself to that?
Yes! Here I’ll pull a lesson I got from distance running. When I embraced the pain instead of trying to avoid it and when I accepted frustration as part and parcel of the improvement journey, discomfort lost its potency and became a lot more bearable. “Life is a journey, not a destination” takes on new meaning when you stop thinking of contentment as a worthwhile destination.
I lost a bunch of ladder games. So what? I threw a game I would have won if I had just all-in’d. Who cares? I don’t want to watch the replay because it’s painful to think about. What? Why does it matter? It’s just a replay. The goal is to get better, and understanding these losses are the ticket to doing that.
It’s strange to say it, but for me, it’s far more comfortable being uncomfortable, than it is to be content on a seemingly endless plateau.
Now to get back on that ladder.
Thanks for reading! I write about StarCraft and Age of Empires. I’d love it if you followed me on Twitter and Facebook and subscribed to my game design-focused YouTube channel. I also stream Age of Empires II ladder five times a week – tune into my Twitch channel to catch all the action!