Sunday, 5 May 2013

being good at glitch tank


I find myself in possession of what is possibly the most pathetic skill: being really good at a competitive game that I made myself and that virtually nobody else plays.

Recently I've been struggling when I try to make multiplayer games. I get excited about ideas but lose motivation very quickly. This weighs heavy on me, even though I'm doing okay making singleplayer games; multiplayer is what I care more about. I try not to be too much influenced by people's responses, or by stupid numbers like download counts or page views or sales, but I've not been able to be unaffected by the difference in response I get to singleplayer games versus multiplayer ones. I've written before about the frustration of feeling that I've failed these multiplayer games (Glitch Tank in particular) by not being able to give them the audience I feel they deserve. I suspect at this point there is no chance of them ever finding that - this culture has a singleminded obsession with newness; if something is "old" (i.e. not released THIS WEEK) then it's not worth paying attention to; new opportunities arise only for things newly made. So I must accept failure.

Was trying to explain to someone recently why Glitch Tank is the best game I've made. I thought maybe writing something about what's good about it might help me to move on (it worked for my scarf).

(Is it the best game I've made? Possible cognitive biases at play here: 868-HACK may be better but it's still too close for me to view objectively - its compromises are too fresh; Vertex Dispenser took a lot out of me and I don't much like to think about it now; O's vulnerability to violence bothers me; solved puzzles aren't interesting.)

I suspect maybe in my being-bad-at-marketing I've failed to communicate what kind of game Glitch Tank actually is. Or more importantly, what it isn't: it's not a casual party game. I think maybe people are sometimes disappointed with it because they expect this is what it should be? It's an "indie" game, it's on the ipad, and that's what most of the other local multiplayer games on there are. Glitch Tank has humour, but it's not something that plays well without giving it your full attention. It's a deep competitive game, and it's most interesting at a high level of skill. It's an "e-sport", if that's your preferred language. It's good for getting good at, it's good for playing a lot of, we still play it. This fits well with how I like to play games: picking a few to master rather than trying to keep up with everything that gets released. It's maybe not so appealing if you prefer to dip into lots of different things. It's a high-skill game; this means it doesn't work terribly well between players of widely different skill levels. The best way to enjoy it is with a group of people getting good at it together. It's not so interesting if you just get good at it and then beat people up; you want to be challenged.

You have to make decisions in a short space of time. It's hard to quickly calculate the consequences of an action; the relevant information is spread across the map and it takes time to process it. It presents a wide range of possible situations, and subtle variations can lead to very different outcomes. The heart of the game is the duplicate action, creating a copy of your tank which responds to the same input in parallel. The same instructions produce different outcomes for tanks in different positions and orientations, and all of these outcomes matter. Attacking from an unexpected direction with one of your clones expresses quick thinking, cleverness, simultaneously paying attention to disparate things, creative problem solving.

Eventually you develop an awareness of how long it will take to absorb this information and work out the consequences, and you start to take this time into account. You enter a state of "metacognition", thinking about your thought processes, making decisions based on acute self-awareness. Weighing the advantage gained from acting quickly against the advantage gained from considering your move.

Meanwhile you are of course considering your opponent's thought processes, how long they will take to think about things. You've seen that they have an opportunity to attack you but they're thinking about something else - do you act quickly to evade the attack but risk hurting your position because you haven't taken time to fully evaluate your evasive move, or do you wait to be more certain of your move but risk them becoming certain of theirs?

A key factor that makes this work is that every action carries a risk of negative side-effects. The laser and mine actions allow for friendly fire, you can damage yourself; mine especially since you place it right next to you. This makes the duplicate action risky; because your clones start near you they're more likely to harm you than your opponent (this too is why spamming actions quickly is unwise - you'll kill yourself more often than not; the "overheating" effect was not added for balance but to signal that this is a bad idea). Just moving becomes risky when you might step on a mine or into your opponent's line of fire, and even turning can ruin you if you end up stuck facing a wall or danger. So a poorly evaluated action doesn't just fail to improve your position, it can actively hurt you.

The rare invisibility action is the fullest expression of this; it's clearly advantageous to be hidden from your opponent, but unless you act very carefully (which costs attention, slows you down) you can lose track of your own position at which point you're acting blind - unless you do a action with a visible effect, which reveals you to your opponent also.

In some ways these thought processes are a lot like other high-level competitive games (which is what I was aiming for; the Kompendium project was an attempt to make the mental states of high-level RTS play more accessible). But I feel it's different enough (at least from ones I'm aware of) to be notable. The way randomness is fundamentally incorporated into your actions means your plans must be flexible, you have to manage risks and possibilities, and there are strategic considerations that aren't possible in a low-randomness game (holding onto opportunities until they're useful, trying to cycle through actions that aren't useful without endangering yourself, ensuring your hand doesn't get cluttered with problems). But the way this happens in real-time distinguishes it from the board/card games that present similar considerations; you have to think about these things quickly and manage time.


  1. I've played Glitch Tank a handful of times, I think it's great but rarely find the right situation to play it. The main problem comes from the fact it's really hard to be an evangelist for it, because when I introduce knew players they are learning the basics while I'm remembering some of the depth. Unless people are very persistent they get fed up before they get good. (This paints a great picture of me doesn't it...)

    If you could get critical mass, then online multiplayer could be really good. Especially if you had some sort of automatch to pit you against somebody of similar level. It's much more fun if you play against of similar ability.

    An interesting thought that has just occurred to me is that you could easily handicap GT, if the stronger player had only 3 or maybe even 2 cards to choose from, or (more difficult) the distribution of draws was altered somehow.

    I'd love for you to show me how poor I am at Glitch Tank at some point!

  2. I have been trying to figure out what the solution to this problem is for the past few days. I think I have a good set of things that would solve for pro-mode play.

    1. NetPlay - this is actually mostly in support of the next idea, but I think it needs to be done to have a critical mass of players playing it. Free to play would support this as well, but that has certain baggage that you might not want to get into. More players though, so you are more likely to get the potential pros playing.

    2. Ladders - Right now there is no reward (and nothing on the line) over a set of matches. With the randomness in a single game, you (and I mean me, as a sorta shit player) can blame the system for any one loss. I know that you will always beat me at this game though. Having this ranking persistant, and on the line in any particular game would create a way that pros could actually be challenged. The net play would mean that the real pros could play each other more often, and would not need to wait for co-location.

    In short, I agree with the above comment, just long-windily.

  3. Good luck in your endeavors. And thanks for the rules of Toepen, although you forgot to mention how many players play Toepen, I just looked it up afterwards.

  4. I've tried to get into Glitch Tank a couple times with little success; a lot of things that would become irrelevant after sustained play and learning, like distinguishing action icons from one another and resisting the urge to act frantically, hung me up. The scattered hour or two I've put into it was spent mostly dealing with these distractions--I was able to perceive some rich, complicated interactions, but I was a long, long way off from being able to comprehend them.