The Most Inaccessible Part of Ranked RTS is Build Order Reactions

The content I want that doesn't fully exist yet

Thanks to my serendipitous combination of vanity and shamelessness, I often complement my yearly ladder grinds by critiquing the level of play of full-time streamers. “This guy plays 8 hours a day, everyday; and he’s lower rank than me!” I chortle, choking down one last bite of a chicken leg as grease runs down a stretched-thin T-shirt that may or may not have fit properly fifty kilos ago. About a year back I made a similar (though perhaps more nuanced) observation regarding the attempts by several Grandmaster StarCraft II players to grind the Age of Empires II ladder.

Appropriately portioning one’s practice time is a tricky balance. On the one hand, some players focus too much on fundamentals, failing to build the experience and mechanical comfort that comes from actually playing games on ladder. On the other, there’s the pitfall of simply playing too much - players that put the time in, but don’t periodically take a step back to correct bad habits or pivot their play in a more optimal direction. This second group is the source of the old saying, practice doesn’t make perfect - perfect practice makes perfect.

I think by now this is well understood by the community of competitive players; every mainstream RTS game has a group of content creators churning out guides on build orders and mechanics and so forth. And while this content doesn’t always perform super great (a look at the view counts of Hera’s invaluable Age of Empires II guides, compared to the other type of content he makes, is eye-opening), there’s still a decent-sized audience for it, and it’s a relatively evergreen space to create value and build a following.

But I still think there’s a major unexplored gap here, specifically the realm of build order reactions. I spoke about this in more general terms about a year back, but at the time I couldn’t quite articulate the specific artifacts I actually wanted. I’ll talk more about that today, as well the practical reasons why I think this type of content is not yet mainstream. And I think this is important not only because I think this work would improve the competitive landscape, but because it represents a way of thinking about ranked play that most players would benefit from.

How Do You Counter Lurkers?

My target audience here is serious but non-professional players - people who approach ranked play seriously (learning build orders, watching tournaments, etc) but who do not play the game professionally as a full-time job. I would say this cohort stretches from somewhere around the top 75%-ish of players (Diamond League in StarCraft II) all the way to mid-to-high Grandmaster.

(This is actually a small group - roughly ~5% of the overall player base of any given competitive RTS, though over-represented in online forums and such.)

I think most such players are well-aware that build orders are crucial to successful competitive play. And I think they’re also aware - as anyone who’s tried out a build order on the ladder can attest - that build orders often require adjustments based on what one’s opponent is doing. One can’t, for instance, blindly expand against a cannon rush, or fail to move their villagers against a tower rush because “this is a one lumber camp build”; both are recipes for an instant loss.

But how does one react? After all, the logic of a build order is that it leads to an optimal game state - a tech traversal, unit composition, economic power, timing attack, etc. The point of a reaction, however, is that “optimal” can’t solely be defined in the context of what’s best for you; it also needs to consider what your opponent is doing, and what best counters that. Thus, optimal build order reactions strive to balance minimally shifting from your existing plan alongside countering your opponent as much as possible, with the recognition that RTS games feature imperfect information and reactions are always to some degree probabilistic in their optimality (which, itself, is a reminder that part of reacting optimally includes scouting the right things at the right time).

Let me speak more plainly - you can’t just not react to your opponent. What they’re doing should affect what you’re doing. But in any given game state, there’s a range of good and bad reactions, relative to the build order you were following in the first place. And a huge component of skill in competitive RTS is not only following a competitive build order, but also knowing how to adjust that build order for different game states.

Unfortunately, it’s pretty hard to describe this comprehensively, because there are just so many possible game states. Go too deep and your advice is hopelessly obscure; too broad, and it’s overly general and hard to apply consistently. Thus, the sort of work I see in this space ends up focused on countering popular strategies (e.g. particular types of rushes or play styles) with a healthy amount of general advice (here’s how you want to play) and a sprinkling of specifics (it’s important to scout for X at time Y). Another common approach is to review replays, which enables the observer to make targeted comments about the game state and optimal moves from each player.

That’s all well and good, but I think there’s an opportunity here to do better. I’ll make the observation that at the highest levels of professional play, there’s a lot of similarities in the build orders that players follow. And because they execute these builds pretty consistently, they also execute their reactions pretty consistently, because the players they’re playing against are also executing at a similar level of consistency. And all of that consistency put together substantially reduces the amount of strategic ground that needs to be covered.

I would posit, then, that if one were to limit oneself to serious players - above average ranked players that have graduated beyond the realm of doing random things - it would be possible to summarize the meta from the perspective of popular professional builds and their reactions - here’s a build, and here’s the five to ten common ways in which it’s adjusted in response to some other builds. It’s still a lot of information covering a good deal of what players will see game to game, but it’s more manageable than infinite game state permutations. And because these are common reactions in the context of optimized build orders, they should typically imply fundamentals about the generally correct way of playing the game, and this should make it easier for players to improvise in situations that aren’t covered by the most popular use cases.

(A good example of this would be the tendency of build order divergences to nonetheless eventually converge on convergent points, an important idea that shows up across mainstream RTS games (even those with randomized maps). Once you understand this idea, you tend to react better in all situations, even wild one-off cases.)

I’ll label this outcome a meta report - a highlight of the most popular builds, the most common reaction traversals, and how this has changed since the last report (thus enabling analysis of meta trends over time). Certainly, this wouldn’t cover every possible type of game one would see on the ladder - but I think it would cover the most important areas, and still offer insights on how to handle everything else. And as a result I think this would be a valuable toolkit for any serious competitive player; the sort of thing that instantly uplevels your play, and makes it an order of magnitude easier to come back from a long break. It’s genuinely something that I would pay decent money to subscribe to, and I don’t think I’m alone in that.

But the really cool thing about such a report, assuming it’s structured correctly, is the insights one would be able to glean about a game’s design and balance. Players are beginning to react to X differently - why? What’s changed? Is it our understanding of the game, a design or balance change, a player preference change, or something else? Have people simply gotten better at a different, more optimal approach? And why is this happening now, and what can we learn about the underlying game design as a result?

What’s Stopping Us?

I think the challenge with this kind of content is that it’s time-consuming to produce - it implies the creator has depth of knowledge across all races in a game (at least sufficient to understand, intuitively, how they work and play against each other), plus depth of knowledge of the professional meta, plus the willingness to follow along continuously as the meta evolves - all the while, perpetually researching and inferring what’s actually considered optimal and what’s not, because it’s not like professional players go out and explain their every decision after every game.

I’d hazard a guess that putting something like this together would be akin to a part-time job - playing the game, observing professional play, and compiling and understanding the results. That’s a lot of time, for pretty esoteric and obscure content.

But, I also think that while the target audience of people who would read this is small, I also think that a high-quality version would be received well. I don’t think this is the type of thing that can be done casually - a poor-quality report is worse than no report at all. But done well, and I think any serious Master’s or Grandmaster player would gladly subscribe to it, even if it costs them $5 or $10 a month, purely from the time savings of not having to keep up with everything.

And I guess one thing I want to stress here is that while this is an idea for content, it’s also a push to think differently about real-time strategy games. I think it’s natural to bias oneself toward the mindset and playstyle of one race, even at high levels of play where players need to understand all aspects of the game. I’m definitely prone to that. But I think it’s easier to understand and appreciate a game if one is always looking at it from both sides, much as players tend to do when they analyze games of chess.

How often do players watch replays and ask themselves what their opponent could have done better? My guess would be very few, at least not consistently; but that’s a mindset I want to try to evangelize, because I think it makes competitive RTS more enjoyable to play and watch.

So anyway, I think this is a solid idea; I’ve punted it around my head for the past several months, and I want to put it into action, at the very least to get the idea of “main’ing” a race out of my bloodstream for awhile. I write this article partly just to get feedback on this stuff early and catch blind spots before committing a bunch of time. But I also think transparency is a good thing more generally. The content ecosystem is really competitive; lots of people are doing lots of really cool things. It can be easy to shy away and work on things in private until they’re perfect. I’d rather not do that on this go around, and instead include folks in how I’m thinking about stuff. After all, being transparent about my last Age of Empires grind is what motivated me to crack into the top 200 of Age of Empires IV, an achievement and time period I look back on fondly.

I imagine Stormgate will be where I’ll try my hand at this - aside from being an exciting new upcoming competitive RTS, it’s also a game I’d like to simply understand better, and find my small niche where I can create value for the community. And spending time creating this report ought to provide a bountiful source of ideas for my weekly free articles, too (which will continue regardless, even if at my occasionally uneven cadence™). But in any case, as I begin to warm up my hands with some RTS in the coming weeks, I’ll be thinking about how this kind of thing could work, and how I could best put it into action.

Until next time!
brownbear

If you’d like, you can follow me on Twitter, Facebook, and Instagram, and check out my YouTube and Twitch channels.