WEBVTT 00:00:03.860 --> 00:00:09.480 It's almost impossible to imagine the game Ape Out without an ape. 00:00:09.480 --> 00:00:12.410 And yet, that's exactly how it started life. 00:00:12.410 --> 00:00:17.430 According to designer Gabe Cuzzillo, this was supposed to be a time-looping stealth 00:00:17.430 --> 00:00:21.370 game where you'd use push and grab mechanics to slink along walls. 00:00:21.370 --> 00:00:25.540 But if there were going to be guards in the game, then naturally you should be able to 00:00:25.540 --> 00:00:30.750 use those same mechanics on them, leading to gameplay that has you holding guards hostage 00:00:30.750 --> 00:00:33.350 and chucking them at walls. 00:00:33.350 --> 00:00:38.840 This turned out to be the most interesting part of the game: and so Gabe decided to wildly 00:00:38.840 --> 00:00:43.500 change direction and actually build the game around this core concept. 00:00:43.500 --> 00:00:48.320 He removed everything that didn't need to be there, like stealth and time-loops. 00:00:48.320 --> 00:00:53.480 And did whatever he could to emphasise this idea - most notably, by swapping the bald 00:00:53.480 --> 00:00:59.239 guy for a rampaging, 300-pound gorilla. 00:00:59.239 --> 00:01:03.840 This is an example of a game design methodology called "Follow the Fun". 00:01:03.840 --> 00:01:09.510 It’s the deceptively simple idea that designers should ignore their plans and preconceived 00:01:09.510 --> 00:01:15.350 ideas - and instead look to the game itself to find out where the development should lead. 00:01:15.350 --> 00:01:19.330 So take the microscopic tactics gem, Into the Breach. 00:01:19.330 --> 00:01:24.260 This game started life as a pretty standard Advance Wars-style game, where enemies chose 00:01:24.260 --> 00:01:28.240 attacks at random and hid their intentions until their turn. 00:01:28.240 --> 00:01:32.810 But one foe in the game would show you exactly what they were about to do on their turn, 00:01:32.810 --> 00:01:35.890 by highlighting the tile they were going to attack. 00:01:35.890 --> 00:01:41.030 The developers at Subset realised that this was the single most enjoyable part, and so 00:01:41.030 --> 00:01:46.630 decided to focus the rest of the game almost exclusively on these telegraphed attacks. 00:01:46.630 --> 00:01:51.009 And this actually helped dictate the rest of the design decisions that the studio had 00:01:51.009 --> 00:01:52.009 to make. 00:01:52.009 --> 00:01:55.940 Because, if you know what the enemies are going to do, can't you just move your own 00:01:55.940 --> 00:01:58.150 units out of their attack zone? 00:01:58.150 --> 00:02:02.509 Alright, maybe the game is actually about protecting static buildings. 00:02:02.509 --> 00:02:06.640 And so now it's about pushing the enemies around so their attacks will miss. 00:02:06.640 --> 00:02:11.370 But actually, you could use this to trick the enemies into killing each other. 00:02:11.370 --> 00:02:16.340 You can probably see why the designers who use this method often say that their game, 00:02:16.340 --> 00:02:19.340 to some extent, designed itself. 00:02:19.340 --> 00:02:23.180 Here's Sam Coster from Crashlands-developer Butterscotch Shenanigans. 00:02:23.180 --> 00:02:27.950 SAM COSTER: "We like to think about this process as the game discovering itself over time. 00:02:27.950 --> 00:02:33.701 Because as iterators, rather than designers, it's our job to simply play the game, listen 00:02:33.701 --> 00:02:39.200 to it, feel it, and kind of feel out what it seems to want to become - and just follow 00:02:39.200 --> 00:02:40.700 the trails of what's fun." 00:02:40.700 --> 00:02:46.980 Now, the idea of a game designing itself is surely rather exciting for those looking to 00:02:46.980 --> 00:02:48.550 make the next big hit. 00:02:48.550 --> 00:02:52.560 But, it's not like amazing game ideas just appear from the ether. 00:02:52.560 --> 00:02:54.430 So, where do they come from? 00:02:54.430 --> 00:03:00.420 Well, let's look at the origins of the rhythm-based roguelike Crypt of the Necrodancer. 00:03:00.420 --> 00:03:05.180 Designer Ryan Clark wanted to see if he could put Spelunky's quick-fire decision making 00:03:05.180 --> 00:03:08.450 into a more traditional turn-based dungeon crawler. 00:03:08.450 --> 00:03:12.970 So he made a quick prototype of a roguelike where you only have a second to make your 00:03:12.970 --> 00:03:14.420 next turn. 00:03:14.420 --> 00:03:19.410 When he played it, Ryan realised that it had an almost rhythmic-like quality - and it became 00:03:19.410 --> 00:03:25.280 obvious that the game should be set to music. 00:03:25.280 --> 00:03:29.989 Or perhaps we should take a look at the iconic mid-air movement of Rocket League. 00:03:29.989 --> 00:03:35.099 When Psyonix was working on the game's predecessor, Supersonic Acrobatic Rocket-Powered Battle-Cars 00:03:35.099 --> 00:03:37.900 - they've learned a lot about marketing since then. 00:03:37.900 --> 00:03:42.870 Uh, they had built a game about battling cars, but wanted to add a speed boost mechanic. 00:03:42.870 --> 00:03:46.769 So the devs simply applied a physics force to the back of the car. 00:03:46.769 --> 00:03:52.080 In testing, they discovered that players could use that force in mid-air to rocket about 00:03:52.080 --> 00:03:53.080 the arena. 00:03:53.080 --> 00:03:58.690 That wasn't the plan, but the developers realised that this actually added enormous depth and 00:03:58.690 --> 00:04:02.360 a whole extra dimension to the game - so they kept it. 00:04:02.360 --> 00:04:07.550 The studio says “we developed this mechanic almost by accident”. 00:04:07.550 --> 00:04:12.739 In fact, there's a whole history of games where bugs, glitches, and accidents in the 00:04:12.739 --> 00:04:14.870 development process were turned into features. 00:04:14.870 --> 00:04:20.110 For example, Hideki Kamiya found a bug in Onimusha: Warlords that let you juggle enemies 00:04:20.110 --> 00:04:22.180 in the air by repeatedly slashing them. 00:04:22.180 --> 00:04:26.139 It was fixed in Onimusha, but Kamiya developed it further, and turned it into the premiere 00:04:26.139 --> 00:04:28.710 game mechanic of Devil May Cry. 00:04:28.710 --> 00:04:34.860 The point being, this process involves taking some initial idea - however loose, fuzzy, 00:04:34.860 --> 00:04:39.289 or unoriginal it might be - and actually building a working prototype. 00:04:39.289 --> 00:04:44.910 And it's here - during the process of coding and playing - that new ideas can spring up. 00:04:44.910 --> 00:04:49.629 And so it's up to the designer to be open and attentive to what the game is saying. 00:04:49.629 --> 00:04:54.800 To realise what's interesting, and be willing to explore those aspects… even if they don't 00:04:54.800 --> 00:04:57.960 totally align with what you originally had in mind. 00:04:57.960 --> 00:05:02.949 That's how Gunpoint went from being about a robot in space who drops fridges on people, 00:05:02.949 --> 00:05:07.360 to being a puzzle game about a spy who re-wires buildings. 00:05:07.360 --> 00:05:13.439 That re-wire mechanic was just one possible idea for a hacking mini-game in a side-scrolling 00:05:13.439 --> 00:05:15.189 Deus Ex-inspired game. 00:05:15.189 --> 00:05:19.830 But as soon as designer Tom Francis started prototyping it, the game that would eventually 00:05:19.830 --> 00:05:22.349 become Gunpoint started to emerge. 00:05:22.349 --> 00:05:23.349 Here's Tom: 00:05:23.349 --> 00:05:27.360 TOM FRANCIS: “It just immediately became obvious that this should become a puzzle game. 00:05:27.360 --> 00:05:28.719 That was just a puzzle mechanic. 00:05:28.719 --> 00:05:30.649 And so Gunpoint just kind of told me what it wanted to be. 00:05:30.649 --> 00:05:33.210 It just wanted to be a puzzle game, just obviously. 00:05:33.210 --> 00:05:38.879 And I just rolled with that, I just expanded the hacking mechanic to a crazy extent. 00:05:38.880 --> 00:05:40.960 I built the whole game around it." 00:05:40.960 --> 00:05:46.360 So this process generally causes the most significant changes towards the beginning 00:05:46.369 --> 00:05:51.089 of a game’s development - which is the point that Sam Coster describes the game as a white 00:05:51.089 --> 00:05:53.589 hot ball of malleable magma. 00:05:53.589 --> 00:05:58.310 But it can still be used as development goes on, and the game starts to form and settle 00:05:58.310 --> 00:05:59.529 into rock. 00:05:59.529 --> 00:06:04.889 Like, for content generation: Jonathan Blow has said that the puzzles in Braid were simply 00:06:04.889 --> 00:06:09.689 a showcase of the unexpected consequences of his time-travelling game engine. 00:06:09.689 --> 00:06:14.379 Blow says "I had a curator role, cleaning up the answers and presenting them in such 00:06:14.379 --> 00:06:17.249 a way that they could be enjoyed by the game’s players.” 00:06:17.249 --> 00:06:19.249 More on that in this video. 00:06:19.249 --> 00:06:22.349 Or, it can be used for listening to player feedback. 00:06:22.349 --> 00:06:28.029 When Chris Hecker made SpyParty, players found all sorts of exploits and unintended ways 00:06:28.029 --> 00:06:29.499 to play the game. 00:06:29.499 --> 00:06:33.969 Instead of fixing these “bugs”, Chris leaned into them and made them an official 00:06:33.969 --> 00:06:37.900 part of the game - pushing the experience towards being more about 00:06:37.900 --> 00:06:40.680 mind-games and psychological tricks. 00:06:40.680 --> 00:06:44.999 And it can simply be used to help guide the general development of a game. 00:06:44.999 --> 00:06:47.059 Here's Subnautica designer Charlie Cleveland: 00:06:47.059 --> 00:06:49.729 CHARLIE CLEVELAND: "You kind of think you know where you're going. 00:06:49.729 --> 00:06:51.729 You have some place on the horizon. 00:06:51.729 --> 00:06:53.810 And there's many paths and you don't know how to get there. 00:06:53.810 --> 00:06:57.900 But if you listen to the game it will tell you where it wants to go." 00:06:57.900 --> 00:07:01.959 That's how his studio made a horror game, without that ever being the intention at the 00:07:01.960 --> 00:07:05.200 start of the project. 00:07:05.200 --> 00:07:10.560 Now obviously, this sort of design process can make it very difficult to predict how 00:07:10.569 --> 00:07:12.860 long a game will take to make. 00:07:12.860 --> 00:07:17.529 This is one reason why the methodology is more popular in the world of indies - rather 00:07:17.529 --> 00:07:21.179 than the fiercely regimented world of blockbuster game production. 00:07:21.179 --> 00:07:26.930 Like, when Tom Francis made his second game, Heat Signature, he hoped that the fuzzy idea 00:07:26.930 --> 00:07:32.409 of “you go inside spaceships” would magically lead to good stuff - just like had happened 00:07:32.409 --> 00:07:33.800 with Gunpoint. 00:07:33.800 --> 00:07:37.580 But… it just didn't. At least, not for a very long time. 00:07:37.580 --> 00:07:42.599 In truth, Tom realised that he had to make a butt-load of stuff in order to find out 00:07:42.599 --> 00:07:46.590 what made the game interesting, which lead to a protracted development where he made 00:07:46.590 --> 00:07:52.360 a ship generation system, artificial intelligence, a combat system, a whole galaxy map with an 00:07:52.360 --> 00:07:53.589 economy, and so on. 00:07:53.589 --> 00:07:59.770 It took Tom years to to figure out that the on-ship combat was the most interesting bit. 00:07:59.770 --> 00:08:03.649 And so that's why it’s worth remembering that the real phrase is actually a bit longer 00:08:03.649 --> 00:08:05.509 than just "Follow the Fun". 00:08:05.509 --> 00:08:09.699 I traced the coinage of the phrase back to this guy - Marc LeBlanc. 00:08:09.699 --> 00:08:14.879 He’s a designer who worked on Thief and System Shock, and an educator who helped come 00:08:14.879 --> 00:08:17.489 up with ideas like the MDA framework. 00:08:17.489 --> 00:08:22.009 When he coined this phrase, he actually started it with a well-known idiom from the world 00:08:22.009 --> 00:08:25.550 of design and entrepreneurship: "fail fast". 00:08:25.550 --> 00:08:30.139 That’s the process of throwing something together as quickly as possible, to see what 00:08:30.139 --> 00:08:32.029 works and what doesn't. 00:08:32.029 --> 00:08:35.890 It doesn't matter if you fail, because you didn't waste much time - but that so-called 00:08:35.890 --> 00:08:40.979 "failure" will tell you so much about what direction the next attempt should take. 00:08:40.979 --> 00:08:45.940 So, perhaps there are some concrete techniques for speeding up the iteration process? 00:08:45.940 --> 00:08:51.699 Well, one is something all GMTK viewers will be familiar with: Game Jams. 00:08:51.699 --> 00:08:56.360 Those frantic game creation marathons where you have to try and make a game in, perhaps, 00:08:56.360 --> 00:08:57.870 a single weekend. 00:08:57.870 --> 00:09:03.019 Arvi Teikari, who dreamt up the award-winning puzzler Baba is You at a Game Jam, speaks 00:09:03.019 --> 00:09:04.469 to the power of these events: 00:09:04.469 --> 00:09:07.320 ARVI TEIKARI: "The whole idea is you can take a prototype that you have 00:09:07.320 --> 00:09:09.829 in your head and try to make something around it. 00:09:09.829 --> 00:09:11.910 if it doesn't work, that's fine. 00:09:11.910 --> 00:09:16.600 you can toss it away after the game jam. you are not committed to the idea for longer than 00:09:16.600 --> 00:09:18.610 the game jam takes". 00:09:18.610 --> 00:09:24.480 Another technique is to use tools that suit rapid prototyping, like Game Maker and Godot. 00:09:24.480 --> 00:09:30.769 Or perhaps paper prototypes, LEGO, or the PS4 game-creation suite Dreams. 00:09:30.769 --> 00:09:34.839 If you've already made most of the game and just want to generate content, you can develop 00:09:34.839 --> 00:09:39.889 some custom level creation tools to speed up the creative process - and get more people 00:09:39.889 --> 00:09:41.250 on board to help. 00:09:41.250 --> 00:09:46.930 So, for Mario Galaxy 2, Nintendo made simple level creation tools to encourage everyone 00:09:46.930 --> 00:09:49.839 on the team to think up unique mechanics. 00:09:49.839 --> 00:09:55.910 And to quickly focus on design and mechanics, you can use placeholder art, music, and plot. 00:09:55.910 --> 00:10:01.589 When Klei made the very first game jam prototype of Don't Starve, the hero of the game was 00:10:01.589 --> 00:10:03.200 actually represented by… 00:10:03.200 --> 00:10:04.960 Link, from Zelda. 00:10:04.960 --> 00:10:07.860 And finally, it can actually help to have something 00:10:07.860 --> 00:10:11.000 about the game that absolutely cannot change. 00:10:11.000 --> 00:10:16.620 Sunni Pavolic from thatgamecompany says the studio used a very iterative methodology when 00:10:16.620 --> 00:10:23.170 making Journey, but always with the idea that this game would explore the theme of love. 00:10:23.170 --> 00:10:27.560 This gave everyone on the team a singular direction to follow, and helped narrow the 00:10:27.560 --> 00:10:31.920 range of possible ideas that they might discover and develop. 00:10:33.780 --> 00:10:39.319 So if there’s one thing I’d want you to take away from this video: it’s to stop 00:10:39.319 --> 00:10:42.200 waiting for the perfect game idea to come along. 00:10:42.200 --> 00:10:45.820 It’s easy to look at something like Ape Out, Crypt of the Necrodancer, 00:10:45.820 --> 00:10:50.279 or Crashlands and assume that these games were designed in a flash of insight 00:10:50.279 --> 00:10:53.490 - which seamlessly transitioned into the final game. 00:10:53.490 --> 00:10:57.230 And so if you can’t come up with an idea as good as these ones - why bother trying? 00:10:57.230 --> 00:11:02.510 But as I’ve shown in this video, nothing could be further from the truth. 00:11:02.510 --> 00:11:07.569 In reality, the thing that ties all of these developers together is that they got started 00:11:07.569 --> 00:11:09.019 and they built something. 00:11:09.019 --> 00:11:13.970 And it was only then - as the designers tried new ideas, played their prototypes, and even 00:11:13.970 --> 00:11:17.829 created bugs - that the games we know today started to form. 00:11:17.829 --> 00:11:22.519 They’re great designers not because they came up with amazing ideas - but because they 00:11:22.519 --> 00:11:28.649 knew how to listen to the game, knew which avenues to follow, knew how to fail fast and 00:11:28.649 --> 00:11:34.750 fail often, and knew how to coalesce these disparate ideas into something coherent. 00:11:34.750 --> 00:11:39.339 So if you watch Game Maker’s Toolkit and think you might like to make a game - don’t 00:11:39.339 --> 00:11:41.120 wait for the perfect idea. 00:11:41.120 --> 00:11:42.579 Build something. 00:11:42.579 --> 00:11:48.550 And then you can listen to the game, follow the fun - and you might just discover that 00:11:48.550 --> 00:11:53.320 the game, in some small way, designs itself.