CGD Podcast Episode 15: “Tharsis” Article and Redefining Elegance


In this episode, I first discuss a new article on Gamasutra that discusses Zach Gage’s new highly random (and proud of it!) game, Tharsis. For more on my positions on randomness, I would check out this article or this video.

In the second half, I talk about a new, more specific definition for “elegance” as applied to game design. Relevant is this episode of 3 Minute Game Design, where I talk about the “old definition”. Let me know what you think of my new thoughts on the topic.

As always, you can support the show by visiting Thanks for listening!

  • Michael Sinsbeck

    Good thoughts on elegance, you have there. One could start by defining elegance as the ratio (delivered value / cost).

    The rule complexity is a pretty good measure for the costs. The player has to learn the rule before he can play, so this is kind of the investment the player makes.
    As you point out in your podcast, the depth of a game is not a very good measure for the delivered value. A player does not immediately enjoy depth.

    Therefore the ratio (depth / complexity) is maybe only an approximation of (delivered value / cost).
    A better measure for delivered value could be “provided decision material”. So elegance would be a question of how much decision material a game can deliver with a given rule complexity. This new definition would contain a requirement of “depth”, because, as I believe, depth is required to generate decision material. But, of course, there is more to it.

    Decisions are very closely related to heuristic learning, in my opinion. It should be possible to get better at a game without solving it. This “getting better” consists of discovering heuristics (=rules of thumb). I am a huge fan of Go, but I agree with the point you criticize: When I started learning Go, it felt so unintuitive and there were no easy to learn heuristic for getting better at the game. It would have been much easier if I had had at least a rough idea how to differentiate good moves from bad moves. But once I was on a certain skill level in Go, that changed. So, to summarize, one could say that Go does not provide a lot of decision material for a beginner, because there are no clues how to approach the game. But at a higher skill level, Go provides a lot of decision material. So at that level, I perceive Go as a highly elegant and valuable game (and by the way, I do not do a lot of look-ahead reading, when I play, and I still enjoy it a lot).

  • Rack

    The question of randomness is a really interesting one, and I’m not really sure I have any kind of answer to it. I don’t think though you can answer it by first assuming it is inherently problematic any more than you can by assuming it is inherently fun. I think it’s also important to distinguish pre decision randomness from post decision randomness as each has very different pros and cons.

  • Tom

    I’d argue go is elegant, not because pure *complexity* emerges out of simple rules (in that case, 50×50 go would be even more elegant!) but because an *interesting* game with very *human-like* decisions emerges out of simple rules. When you’ve played go for some time (and yes, the learning curve is steep and the climb long – that’s the fundamental reason for its low popularity) you think entirely in meta-condepts, like ‘weak groups’, ‘sphere of influence’, ‘ should I invade or reduce’, ‘do I run into the middle or hide safely in the corner and concede strength to my opponent’, etc. It’s also to me the only truly two-dimensional game, where you’re often faced with the choice of pushing a group left, right, up or down.

    I think the best way to understand the complete centrality of intuition is to see two decent players play: yes there’s plenty of reading, but also lots and lots of vaguer logic as well. I’d heartily reccomend any of the Sibicky vs Jackson duels on YouTube: they’re great to watch, and as an exercise count how many times they say ‘I actually need go read here’ vs logic like ‘I’m worried about my group, but its ok because yours is weak too.’

  • Luke Oram

    Hello, I am a go player so obviously quite biased, but I have some thoughts:
    Firstly, the size of the board was not particularly *designed* as much as gradually developed over a very long time. This particular size was probably arrived at because it has a meaningful play difference to small board- which is what most of us learnt on to begin with. It is at the 19 x 19 scale that it becomes a deeper game.
    It’s probably the most widely played game in the world so I don’t know if it’s fair to describe it as ‘niche’.
    I do agree with you about the difficulties of learning the game for beginners.
    I’d also like to put forward the notion that the difficulty of the game, and dedication it takes to even properly learn it, is what has made the game enduring.

  • Michael Sinsbeck

    I would like to add a short note: I was surprised by your statement that Go would not be successful, if it were invented today. So I asked in the Go-reddit-thread what other people thought. The vast majority of Go players agreed with this statement.

    Here is the Reddit-Discussion:

  • Matthew Kaemmerer

    Great insights, Keith! I hope you’ll do a follow up soon. This episode, seemed to focus mainly about what *doesn’t* make for effective game design, but I’d love to hear more on what does constitute elegance. Or more specifically, can you share some examples of games that are elegant in ways that Chess/Go/Poker/etc aren’t?

    I noticed a tension between a couple of your observations. Namely: “Go (on a regular 19×19 board) is inelegant because it is nearly opaque to beginners” and “9×9 Go is uninteresting because it seems to be almost solved”. To me, this seems to be the central challenge in creating elegance. If we design for a short lookahead horizon, then we are making a trade-off towards how “solve-y” our game becomes. If we allow a long lookahead horizon, we may confuse and alienate beginners.

    Are there any combinatorial games (e.g. Chess, Go, Tic-Tac-Toe, Connect 4, etc) currently, that fit your framework for elegance? Could there be, or is this type of game inherently inelegant?

  • To be totally honest, I don’t think there are any elegant games yet. Haha. I would say the closest are things like Eurogames and some other designer board games.

    Games like those are inherently inelegant because the designer did not himself carefully select where the information horizon would be, and instead left it to “however much you’re able to calculate in the time you have”. More on that here:

  • Matthew Kaemmerer

    Ah, thanks for the link. I think I had been misunderstanding what you meant by information horizon. I had assumed you just meant limiting the search depth of the entire game — the fog-of-war example helped clarify.

  • Jimmy Mac

    I think a better term for “rules complexity” is efficiency. As in how much interesting emergent complexity/fun gameplay comes from the rules. Chess and Go are efficient games because they have few rules but a lot emergent complexity. Effectiveness is a good word for capturing how good or pleasurable a game is. 200 square Tic Tac Toe is efficient but not effective. Elegance is both efficient and effective.

  • KammanderKhan

    A system can have complexity elegance but can be poorly distributed to the player, giving them a way too easy time taking advantage of the depth, or a wall of unreadable possibilities and cause a drop in strategic understanding. True elegance, as you talk about in this episode, is when the design of the mechanical system works to give players manageable amounts of information, so it can be easily accessed. The ratio of inherent/emergent complexity is not ultimately important but rather, how the system provides game states with a balanced ratio of valuable strategic information to the player’s ability to process that information.

    You talk about Go players having a learned intuition about strategic choices, which likely only comes from study and practice. It’s considered a classic because it endures time due to it’s depth and long, steep learning curve, so it’s interesting for you to insist good design must have an accessible learning curve as well to be considered elegant. I think the idea that good design includes accessibility, is an obvious, but a hard learned lesson for a designer well acquainted with gaming culture. We need to stop thinking about making games for “people who like x type of game” or games with x because “people like x” and just make games for people. As you say in your argument against the Tharsis article, we have to go beyond just want people “like”. We actually have to give them something that might genuinely contribute to their lives and understanding of reality.

  • gboybama

    I am quite confused by the front part of the podcast. Casino gambling or exploitative nudity have a known social cost. We weigh that against the amount of “fun” and determine if the cost is worth it. Each of us individually and collectively as a group make that intellectual and/or moral determination.

    In contrast, I cannot figure out the terrible cost you assert of basing games on random mechanics. What about this mechanism offends you? If someone enjoys rolling dice and the outcome determines the fate of doomed virtual astronauts, where’s the harm? Thus, Gage’s statement that random chance is fun. The player gets a shot of adrenaline from the virtual risk and nobody gets hurt. Where’s the problem?

    You seem to think you have a handle on problems presented by randomness in games and you never make a case for why it’s disappointing or depressing or “negative.” You just say it is. The article on Gamasutra doesn’t address these issues because they’re peculiar to you, or at least a very small subset of people, in my opinion. Does your religion preclude you from enjoying games of chance? If so, full disclosure of that might have been advisable.

    Now, if one wants to argue that randomness is a crutch for bad game design somehow, we can have that conversation, but to link randomness of games to “regressive” thinking and social depravity is just totally left field. There’s just no linkage I can think of.

  • The cost is people’s wasted time. I believe that interactive systems can and do deliver real intellectual value to people. Getting better at chess, for example, is the pursuit of a real discipline. I think there is some intrinsic value to pursuing disciplines and gaining skill at something, and I think this is also something most people understand.

    Skinner-box-games don’t really give people that kind of value. Instead, it just hooks players into a compulsive behavior loop that’s much more like “addiction” than anything else.

    So yeah the harm is just “getting people to waste their time doing things that they don’t really get anything out of”.

  • gboybama

    I appreciate the response, but this is a quantitative / qualitative analysis where such rigor is not called for. As far as “don’t really get anything out of,” you’re projecting. The player of the “Skinner box” game presumably gets fun out of the experience or they would not play. Fun is its own reward and is the overriding factor in whether any game is successful. So, we’re really again back to what the heck is so offensive about the random, chance driven mechanic for you, given that the player is enjoying themselves. I have played about 50 games of Tharsis and find it utterly frustrating, capricious and (thus far) impossible. Yet, the challenge I enjoy when grappling with it makes me look forward to giving it another go. That’s pretty cool.

    YMMV, of course.

  • I don’t think the fact that players do something mean that they necessarily get something out of it. In the same way that in general, society advocates for people to play chess over doing heroin, we can make the same kind of judgment about playing chess vs. Candy Crush.

  • gboybama

    So, Candy Crush is roughly the game equivalent of heroin. Got it.

    C’mon man. Just because a game is empty headed and doesn’t appeal to you and me doesn’t mean it’s destructive to society or that those people who enjoy it are wrong.

    Many people love it and that makes it a success in my book. My not caring for it is merely a matter if taste. Granny isn’t hurting anyone.