WEBVTT 00:00:02.710 --> 00:00:07.880 In the manual for Elder Scrolls 2: Daggerfall, Bethesda left a message encouraging players 00:00:07.880 --> 00:00:11.230 to avoid the “replay the save game” strategy. 00:00:11.230 --> 00:00:16.930 They said, “most computer gamers use the save game to maximise their playing ability. 00:00:16.930 --> 00:00:20.960 Anytime something goes wrong, they return to a saved game and replay it until they get 00:00:20.960 --> 00:00:22.030 it right. 00:00:22.030 --> 00:00:26.099 The final history of their game looks like an endless streak of lucky breaks and perfect 00:00:26.099 --> 00:00:27.099 choices. 00:00:27.099 --> 00:00:29.890 [But] role-playing is not about playing the perfect game. 00:00:29.890 --> 00:00:32.890 It is about building a character and creating a story. 00:00:32.890 --> 00:00:37.230 In fact, you will never see some of the most interesting aspects of the game unless you 00:00:37.230 --> 00:00:39.469 play through your mistakes. 00:00:39.469 --> 00:00:43.899 If your character dies, by all means return to your last saved game and replay it. 00:00:43.899 --> 00:00:49.030 However, if your character is caught pickpocketing, if a quest goes wrong, or some other mundane 00:00:49.030 --> 00:00:51.719 mishap occurs, let it play out. 00:00:51.719 --> 00:00:55.269 You may be surprised by what happens next”. 00:00:55.269 --> 00:01:00.190 This is a noble stand against “save scumming”, which is the art of returning to an old save 00:01:00.190 --> 00:01:05.340 file the second you get spotted in a stealth game, or lose a beloved team member in an 00:01:05.340 --> 00:01:09.460 RPG, or get a bad roll in a game with random numbers. 00:01:09.460 --> 00:01:13.881 And it’s sorely tempting to simply undo your mistakes or reset an unlucky event, but 00:01:13.881 --> 00:01:19.210 - as Bethesda says - if you do that, you might just miss some of the best anecdotal stories 00:01:19.210 --> 00:01:20.659 the game has to offer. 00:01:20.659 --> 00:01:25.050 You know, like coming “this close” to dying, but turning it around and winning anyway. 00:01:25.050 --> 00:01:28.950 Or killing a panicked guard, seconds before they can raise the alarm. 00:01:28.950 --> 00:01:32.759 Or an exciting extraction when your stealthy plan starts to go hideously wrong. 00:01:32.759 --> 00:01:37.299 Or having to sadly move on through the story, while bearing the loss of your favourite party 00:01:37.299 --> 00:01:38.299 member. 00:01:38.299 --> 00:01:40.119 And so it’s a lovely sentiment. 00:01:40.119 --> 00:01:45.200 But, ultimately, there’s no point putting this sort of thing in the manual. 00:01:45.200 --> 00:01:49.119 If you’ve got an intention for how players should experience the game, you’ve got to 00:01:49.119 --> 00:01:51.570 build it into the game itself. 00:01:51.570 --> 00:01:55.420 So how can we make games where players don’t want to return to a quick save at the first 00:01:55.420 --> 00:01:57.060 point of adversity? 00:01:57.060 --> 00:02:01.909 Games where you are, in fact, encouraged to play on past mistakes, failures, and setbacks 00:02:01.909 --> 00:02:06.439 - and potentially see some of the most interesting aspects of the game? 00:02:06.439 --> 00:02:12.420 Well, for some games, the strategy is to make sure that setbacks are tolerable - and not 00:02:12.420 --> 00:02:16.890 so critically damaging that you’re better off just returning to a previous save point. 00:02:16.890 --> 00:02:20.620 One way to do this is to give the game a really wide “failure spectrum”. 00:02:20.620 --> 00:02:26.220 That’s a term dreamt up by Tom Francis - creator of Gunpoint and Heat Signature - and it describes 00:02:26.220 --> 00:02:29.860 the range of states between perfect success and total failure. 00:02:29.860 --> 00:02:34.640 Think of a game like XCOM, where you can successfully finish a mission with all your team members 00:02:34.640 --> 00:02:39.560 alive, or botch your objective and leave with two injured units, or come home without a 00:02:39.560 --> 00:02:41.430 single living soldier. 00:02:41.430 --> 00:02:47.400 Pretty much every game has a failure spectrum, but some have a much more generous one, than others. 00:02:47.400 --> 00:02:52.400 For an example, Tom points to the stealthy open worlder Metal Gear Solid V, which has 00:02:52.400 --> 00:02:57.650 a huge range of states between being a sneaky snake and a, uh, dead snake. 00:02:57.650 --> 00:03:01.300 So, if a guard sees you, he won’t immediately start firing. 00:03:01.300 --> 00:03:04.320 He’ll just come in to investigate more closely. 00:03:04.320 --> 00:03:08.660 If you do get spotted, you enter into this slow-mo reflex mode, to give you a chance 00:03:08.660 --> 00:03:11.250 to headshot the guard in question. 00:03:11.250 --> 00:03:14.910 Screw that up, and the guard will need to manually call up his buddies for support, 00:03:14.910 --> 00:03:16.770 giving you a chance to stop him. 00:03:16.770 --> 00:03:20.400 And even after all that, Snake can still escape and return to hiding. 00:03:20.400 --> 00:03:21.980 Or enter combat. 00:03:21.980 --> 00:03:26.350 Or even call in an helicopter and just take the whole “espionage” bit of the MGS slogan 00:03:26.350 --> 00:03:28.160 and dropkick it into the ocean. 00:03:28.160 --> 00:03:31.840 You’ll only die if you manage to screw all of that up. 00:03:31.840 --> 00:03:36.560 So this is a huge failure spectrum, with all sorts of states between completing the mission 00:03:36.560 --> 00:03:39.800 without ever being seen, and bleeding out on the battlefield. 00:03:39.800 --> 00:03:44.160 And that means that mishaps aren’t so punishing that you might as well reload - they just 00:03:44.160 --> 00:03:46.560 shove you a little further down the spectrum. 00:03:46.560 --> 00:03:51.560 But the thing about a failure spectrum is that, in a lot of cases, it’s reversible. 00:03:51.560 --> 00:03:56.510 And if you play well you can actually turn things around and crawl back towards the successful 00:03:56.510 --> 00:03:57.520 end of things. 00:03:57.520 --> 00:04:00.610 That’s a big part of Far Cry 2. 00:04:00.610 --> 00:04:05.100 This is another game with a generous failure spectrum, thanks to its big health bar, loads 00:04:05.100 --> 00:04:09.270 of healing syringes, and the buddy system - where you can get one extra chance to keep 00:04:09.270 --> 00:04:10.630 playing, after your death. 00:04:10.630 --> 00:04:13.720 And it’s also a game where you’re constantly facing setbacks. 00:04:13.720 --> 00:04:16.840 You might find your guns jamming in the middle of battle. 00:04:16.840 --> 00:04:20.960 Or find yourself suffering from a Malaria attack while sneaking past an outpost. 00:04:20.960 --> 00:04:24.740 Or have your car break down, just as you’re making a getaway. 00:04:24.750 --> 00:04:28.009 But these setbacks serve a really important purpose. 00:04:28.009 --> 00:04:33.650 You see, in a 2009 GDC talk, designer Clint Hocking explained that, originally, he wanted 00:04:33.650 --> 00:04:38.319 Far Cry 2 to be all about intentionality, which is achieved by having the game split 00:04:38.319 --> 00:04:39.639 between two phases. 00:04:39.639 --> 00:04:44.030 There’s a planning stage, where you survey the scene, look for items of interest, watch 00:04:44.030 --> 00:04:47.280 guard patrol patterns, and plan your escape route. 00:04:47.280 --> 00:04:51.560 And then an execution stage, where you actually carry out your plan. 00:04:51.560 --> 00:04:54.849 But the creators decided that they didn’t want you to either have your plan totally 00:04:54.849 --> 00:04:56.460 work, or totally fail. 00:04:56.460 --> 00:05:00.919 Instead, they wanted you to suffer small setbacks that would cause you to bounce out of the 00:05:00.919 --> 00:05:04.360 execution stage and back to the planning phase. 00:05:04.360 --> 00:05:06.969 Clint describes this type of gameplay as improvisational. 00:05:06.969 --> 00:05:12.840 It’s this idea of constantly moving between planning and execution - but within a single, 00:05:12.840 --> 00:05:14.180 continuous playthrough. 00:05:14.180 --> 00:05:18.430 And so that’s another benefit of making people play through mistakes and bad luck. 00:05:18.430 --> 00:05:22.800 Because suffering adversity causes you to change your goals in an exciting, dynamic 00:05:22.800 --> 00:05:25.800 way - so you can claw your way back to victory. 00:05:25.800 --> 00:05:29.419 Like in a shooter - if you take a lot of damage you’re forced to change your focus away 00:05:29.420 --> 00:05:33.820 from combat - and towards getting into cover, finding health packs, or maybe even crafting 00:05:33.820 --> 00:05:34.820 a medkit. 00:05:34.840 --> 00:05:38.660 And in a stealth game, getting spotted means you’re forced to run away and get back into 00:05:38.660 --> 00:05:43.600 cover, or just give up on sneaking around and deal with your enemies in the old fashioned way. 00:05:43.600 --> 00:05:48.199 And this only works if players don’t reload their save game the second they get thrown 00:05:48.199 --> 00:05:49.289 off course. 00:05:49.289 --> 00:05:53.789 So to reduce the chance of this happening, Clint made sure these setbacks were small, 00:05:53.789 --> 00:05:55.950 unpredictable, and recoverable. 00:05:55.950 --> 00:05:59.889 Things like Malaria attacks and jamming guns might wreck your plans but they’re far too 00:05:59.889 --> 00:06:05.120 tiny to warrant a reload, they’re easy enough to turn around, and are often completely unpredictable 00:06:05.120 --> 00:06:06.360 when they’re about to occur. 00:06:06.360 --> 00:06:10.659 “It is exactly because the loss is small and unpredictable that players don’t attempt 00:06:10.659 --> 00:06:14.430 to reload the game to escape it,” says Clint. 00:06:14.430 --> 00:06:20.009 Now all of this falls apart if the player is rewarded for perfect play, or punished 00:06:20.009 --> 00:06:21.969 for making mistakes. 00:06:21.969 --> 00:06:25.740 Meaningless ranks and achievements for never getting spotted in a stealth game are fine. 00:06:25.740 --> 00:06:28.440 They’re aspirational rewards for highly skilled players. 00:06:28.440 --> 00:06:32.550 But if making mistakes will cause the rest of the game to be significantly harder, then 00:06:32.550 --> 00:06:37.300 it’s no surprise that a player will reload to a previous save game the second they screw up. 00:06:37.300 --> 00:06:42.080 Back to XCOM: losing soldiers and suffering casualties means you’ve now got to recruit 00:06:42.080 --> 00:06:47.139 feeble rookies, making you less likely to succeed at future missions - creating a nasty 00:06:47.139 --> 00:06:49.689 positive feedback loop of death and failure. 00:06:49.689 --> 00:06:55.080 No wonder, then, that some gamers will resort to save scumming to keep their favourite soldiers alive. 00:06:55.080 --> 00:06:56.860 It’s just too harsh, otherwise. 00:06:56.870 --> 00:07:02.009 Perhaps a better approach is to try and make failure as interesting as success. 00:07:02.009 --> 00:07:06.099 Look to the Shadow of Mordor games, where getting killed by an Ork captain means that 00:07:06.099 --> 00:07:08.939 they’ll remember you and bring up your history in a later encounter. 00:07:08.939 --> 00:07:14.069 So if there’s no tactical benefit for perfect play, and there are meaningful outcomes for 00:07:14.069 --> 00:07:18.490 imperfect play, then gamers will naturally want to roll on past their mistakes and let 00:07:18.490 --> 00:07:21.330 events play out naturally. 00:07:21.330 --> 00:07:25.960 Of course, one easy way to fix all of this is to simply remove the ability to reload 00:07:25.960 --> 00:07:27.520 a previous save file. 00:07:27.520 --> 00:07:32.270 In Darkest Dungeon, the game is always saving over the top of your file, making it near 00:07:32.270 --> 00:07:34.669 impossible to rewind your mistakes. 00:07:34.669 --> 00:07:37.190 Developer Redhook Studios did this for two reasons. 00:07:37.190 --> 00:07:42.729 For one, they wanted players to have to live with bad decisions, and truly awful dice rolls. 00:07:42.729 --> 00:07:46.349 This is a game where rotten things happen, and you’ve got to deal with them. 00:07:46.349 --> 00:07:50.789 And also because they wanted players to really struggle over whether they should take certain 00:07:50.789 --> 00:07:55.479 risks - knowing that they can’t just restore a previous save game if things don’t go 00:07:55.479 --> 00:07:56.479 their way. 00:07:56.480 --> 00:08:00.180 In a 2016 GDC talk, designer Tyler Sigman says 00:08:00.180 --> 00:08:02.460 TYLER SIGMAN: “Permanent consequences were what we were after. 00:08:02.460 --> 00:08:07.680 We want you to, at all moments, be like, ‘should I go a little further and get a little more treasure. 00:08:07.680 --> 00:08:11.260 Do I think I can make to the end of this quest even though these two characters are afflicted 00:08:11.260 --> 00:08:13.340 and this one is almost dead?’ 00:08:13.340 --> 00:08:19.620 And to do that you know we needed this terrible save system that is just really really mean”. 00:08:19.620 --> 00:08:24.500 Other games do this too, like survive ‘em up The Long Dark, which automatically saves 00:08:24.500 --> 00:08:28.469 over your game the moment something bad happens - like getting attacked by wolves or hurting 00:08:28.469 --> 00:08:32.830 yourself - so you’re forced to keep playing from that most dramatic point. 00:08:32.830 --> 00:08:36.930 And ultimately, this is pretty common in console games, where saving your progress isn’t 00:08:36.930 --> 00:08:40.650 so easy to do and you must rely on actual save points or checkpoints. 00:08:40.650 --> 00:08:44.770 This doesn’t mean you can’t let players easily save their game when they need to take 00:08:44.770 --> 00:08:45.810 a break, though. 00:08:45.810 --> 00:08:50.440 In Dark Souls you can only permanently save your progress at bonfires - but you can suspend 00:08:50.440 --> 00:08:52.290 your game at any point. 00:08:52.290 --> 00:08:55.750 This quits the game, and lets you continue from that point next time you play. 00:08:55.750 --> 00:09:01.180 But then that save is deleted, meaning you can’t rewind to that point if you do something silly. 00:09:01.180 --> 00:09:05.090 So, there are lots of compelling reasons to keep players in the game - and not reaching 00:09:05.090 --> 00:09:07.910 for the quick load button the second something goes wrong. 00:09:07.910 --> 00:09:11.200 Screwing up causes you to dynamically shift your goal, and do something different for 00:09:11.200 --> 00:09:12.200 a while. 00:09:12.200 --> 00:09:14.800 Awesome stories can occur when things go horribly wrong. 00:09:14.800 --> 00:09:19.200 And risky play is far more meaningful if you can’t just rewind and try again. 00:09:19.200 --> 00:09:23.430 But it can’t be up to the player to enforce this pure way of playing. 00:09:23.430 --> 00:09:28.160 I’ve quoted Civ 4 man Soren Johnson before, who says “given the opportunity, players 00:09:28.160 --> 00:09:30.770 will optimise the fun out of a game”. 00:09:30.770 --> 00:09:34.650 So if designers really want to keep players in the experience, they’ve either got to 00:09:34.650 --> 00:09:39.500 lock off easy save scumming, make setbacks so tolerable that they’ll want to keep going, 00:09:39.500 --> 00:09:44.270 remove rewards for perfect play, or make failure as fun as success. 00:09:44.270 --> 00:09:46.820 Only then will players “let things play out. 00:09:46.820 --> 00:09:50.560 And be surprised by what happens next”. 00:09:50.560 --> 00:09:51.840 Hey! Thanks for watching! 00:09:51.840 --> 00:09:55.970 Game Maker’s Toolkit is made possible thanks to everyone who pitches in over on Patreon. 00:09:55.970 --> 00:10:00.780 You may have noticed that I’m back on YouTube for streaming - Twitch didn’t really work out. 00:10:00.780 --> 00:10:05.800 I stream at 8PM BST on Wednesdays, so hit that Notification Bell thingy if you want 00:10:05.810 --> 00:10:07.390 to get a heads up when I go live.