Gameplay Balance for 1.29: Part 2 – Fixing the Metagame

Last week, we posted part 1 of this article, which was about tuning the gameplay rules for Version 1.29. In this part, we’ll be talking about the changes coming to Play Mode itself. While tuning the gameplay itself to be functional, the fact is that if the game isn’t operating at the optimal difficulty for a player, all that work of tuning the gameplay is wasted.


Depth of Skill in Auro

For a designer, it’s hard to quantify exactly how deep a game is. A designer and their team can only build the infrastructure that they believe will be conducive to deep gameplay. It’s then up to the players and the testers, usually over the course of months and years, to ultimately answer the question of how deep a game is.

My opinion is that even in version 1.28, Auro is pretty deep – I think it’s fair to say that it’s deeper than most single-player games. With that said, in version 1.28, the highest-skill players have figured out that defensive play is a dominant strategy, and so they’ve sort of effectively solved it as it is, cutting off any potential for more depth beyond that point. Version 1.29 removes that dominant strategy (see part 1 for more on that), and so now the exploration can continue.

For multiplayer games, everyone understands the idea that games are best when players of relatively equal skill are pitted against each other. This is the reason that just about every online game has “matchmaking”, an automatic algorithm set up to try to find an opponent of similar skill. Play an opponent who is significantly worse than you or significantly better, and the game almost just doesn’t work. For the weaker player, his decisions don’t really matter because it’s futile, and for the stronger player, he can just do safe, routine moves that he know will work on this low-skill player. The game just doesn’t function properly unless the game has balanced difficulty for all participants.

While Auro is a single-player game and therefore doesn’t have “matchmaking” in the most literal sense of the word, the same principle applies. The basic idea with the Auro metagame is that we want every player, no matter what level of skill they are at, to be able to play matches that gives them a 50% chance to win (or as close to it as possible). We never want players to finish a game early and feel like “wait, huh? I won?”, and on the other end of the spectrum, we never want players to feel brick-walled by difficulty.

One issue is that in version 1.28 we found that players tend to start on Rank 1 (obviously), but once they got a hold of the basics, they would quickly shoot up through the lower ranks. Ranks 2 and 3 were easy to beat once you got through Rank 1. Ranks 4, 5 and 6 also didn’t really increase in difficulty too much. For that reason, it seems to us that most players who play with any regularity start in Rank 1 and then basically had to grind up to Rank 6-8 or so. Further, it doesn’t take very long to get from Rank 1 to being a “master” player. I’ve seen numerous players do it in under a month of semi-serious play.

The other issue is that the difficulty wasn’t really scaling much once you hit Master rank (10). After that point, the only thing that happened was the score goal got bigger, and when coupled with the “defensive play” thing, that basically means games just get longer. Ranks 14 and 15 were really like endurance matches. This is a common kind of problem that happens to a lot of single-player games at high levels of play, and I’m determined to avoid it for Auro!


The New Setup

The first thing I realized that we needed to do was basically “compress” the skill range, especially now that gameplay was functioning properly. Here’s what we want:

1. There must be a Rank for every player. In other words, for every individual Auro player, there must be some rank at which that player wins 50% of his games. We don’t want any player to be too good, or too bad at the game where they can’t have a 50% chance to win balanced difficulty.

2. There must be players at every Rank. If there are no “rank 2 players”, then why do we have Rank 2? If no players are leveling out there, then that means that Rank 2 is basically a spot that rank 3 players have to grind through (playing non-optimal-balance games) to get to.

3. The transitions between ranks should be smooth. This actually is just an extension of point #1. If the step between rank 3 and 4 is too high, then players could have the experience of getting to rank 4, then getting beaten back down immediately to rank 3, then grinding back up to 4, and so on. In this case, there actually is no rank for this player. They have fallen into a “skill donut hole” of sorts, and they’ll have to play a lot of sub-optimal games (80% win at rank 3, 20% win at rank 4, for instance) before they can get good enough to have optimal games again at rank 4.

My first step was to keep the number of total ranks at 10, and just “squish them down”. We kept Easy Mode super easy (in fact, it may be even easier than ever now thanks to Kissie Rats), but made Normal mode a bit harder, and Hard mode a lot harder. My goal was to make hopefully only the very best of the best players able to compete on Master Mode, because I know that however good players are now, in 6 months when the game has a bigger fan base and everyone has had 6 months to play, everyone will be way better.

Ultimately, the squishdown worked well. The only problem was at the Easy to Normal and Normal to Hard transitions (levels 3-4 and levels 6-7 respectively), the gap was too big. So what I ultimately ended up doing was adding one more Normal rank and one more Hard rank. This means that the layout now goes like this:

Rank 1, 2, 3: Easy Mode

Rank 4, 5, 6, 7: Normal Mode

Rank 8, 9, 10, 11: Hard Mode

Rank 12+: Master Mode

Rank 4 is the new more-difficult Normal mode, but with a very low score goal. Same deal with Rank 8. In our testing so far, this has gone over pretty well!

Without these extra levels, we were actually a bit worried that the new “squishing down” would cause some players to find themselves a rank or two too high in 1.29. However, these new added ranks should mostly address that issue. In addition that, we’re now giving players the ability to take the placement test whenever they want, which also should mitigate the problem.

Overall, we’re really happy with these changes. Most of our testers have said something along the lines of “the game is the best it’s ever been right now”, which is good enough for me! We’re making a release candidate of 1.29 tonight and hopefully should be submitting later this week. Stay tuned for more news!

129aurobalancedepthGame Designgameplaypatch notesskillstrategy

keithburgun • 03/09/2015

Previous Post

Next Post


  1. Yuval Kalugny 03/10/2015 - 4:32 pm Reply

    I didn’t really get how you solved issue #3.
    The ranks are still discrete, so what prevents me from getting into the 80%/20% grind situation? [which I am totally having right now, BTW, between lvls 7 and 8].

    • keithburgun 03/10/2015 - 5:20 pm Reply

      Well, picture a situation where we kept the top and bottom difficulty *the same*, but multiplied the number of ranks by 100. Then each difference would be really, really tiny and it’s super unlikely that this problem would occur.

      The same principle applies for adding just two new ranks. Basically right now, the new “rank 8″ (first Hard rank) has a super low score goal which will help you get something more like a 50% win ratio.

Leave a Reply

Your email address will not be published / Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>