The steampunk caper that proved crime can pay
Burglary is an inherently high-risk profession. There is a chance you’ll make a clean getaway with all your loot, but if you’re spotted you might be forced to make your escape with only a meagre haul to show for your efforts – or, worse, you might get caught. By its very nature, then, a game about breaking into buildings and stealing as much cash as you can has a roughly equal chance of leaving you fulfilled or frustrated. Well before The Swindle’s release, Dan Marshall was well aware it wouldn’t be a game for everyone. “Oh, I knew some people would hate it,” he tells us, “I could see that immediately. And that’s perfectly fine, because it is so difficult and unforgiving. So some people love it and some hate it. That’s videogame development, isn’t it?” He laughs. “Show me a game everyone loves universally.”
The Swindle began development after Marshall had finished Privates, a sex education game for Channel 4. He had a completely blank slate, but his mind kept returning to an idea he’d had around two decades earlier. As a teenager, he’d made a prototype – “a cyberpunk game about a bloke in a long, flappy trenchcoat breaking into buildings” – and now he began to envision that with a slightly different theme. Cyberpunk quickly became steampunk, and Marshall started to experiment with what he self-deprecatingly calls “a silly revamp of a teenager’s stupid idea.”
As with most games, the embryonic version of The Swindle was very different to the finished title. Indeed, Marshall kept encountering design and technological roadblocks: having spent some time working in XNA, he attended the Develop conference, only to be told categorically that XNA was dead, and that it wouldn’t be supported by Windows 8. “I’m not entirely sure if it turned out to be true,” Marshall says, “but it basically felt like the game was on thin ice from the start.” Marshall moved the game over to the Unity engine and began to refresh it. It was, at heart, the same game, but a little more polished in its new form, with Unity’s immediacy allowing him to quickly get back up to speed.
At the time, the levels were hand-crafted, expansive environments designed to offer a number of different points of entry, and to cater to a range of approaches. “The idea was that there would be eight or so buildings that you would keep on breaking into in different ways, and finding different areas of them each time,” Marshall says. These sprawling stages were essentially guarded by an AI director who would respond to your actions on the following attempt. “Let’s say you broke in, and you did something stupid like laying a bomb at the front door,” he continues. “So you blasted your way in and stole what you could and then ran out again. The AI director would think, ‘OK, there’s a vulnerability’, and would install security cameras and put guards on the door and bolster [security] with blast-proof doors and stuff like that. And when you went back, you’d see that and maybe go through the sewers and up through the service elevator instead.”
It sounds like a great idea. In practice, Marshall was unable to make it gel into a satisfying mechanic. The results were the same each time: upon facing extra obstacles, players would invariably give up on trying that entry point and simply find another path, making the AI director all but redundant. “I was constantly shifting it around and trying to make it work, and it just never really clicked,” Marshall admits. He knew it was time to step back, and essentially cancelled the project. He made the most of his free time, developing co-op shooter Gun Monkeys in four months to clear his head.
It was here that Marshall began to play around with procedural generation, a technique he’d considered during the The Swindle’s formative stages of development but dismissed once he realised it meant he would have had to scrap six months’ work and start again. “But by this point there was nothing to lose,” he laughs. “I put together a very simple building generator and thought I’d give myself a week to make a fun prototype. And it immediately worked. I mean, it looked awful – it was just rectangles hitting rectangles with rectangles, and made absolutely no sense visually. But it was fun to navigate the environments and run around the buildings stealing stuff.”
With the decision made to use procedural generation to build the stages, The Swindle had finally started to take shape. This did, however, result in some tough choices for Marshall, including the excision of the game’s narrative. Marshall had imagined a storytelling device he likens to Half-Life 2: just as Dr Wallace Breen delivers exposition via TV screens, so The Swindle’s antagonist would sporadically appear to chide and belittle the player. But it was now an arcade-style game, and the plot was starting to get in the way of the fun. In the finished title, you quickly mash a button or key to start a new heist; back then, Marshall says, players would do the same to skip the story. “It was infuriating,” he says. “It wasn’t a game about stopping and watching plot unfurl, it was a game about telling your own story in some small way. The stories I could write about your heists were never as good as the stories you were telling yourself about what was going on and why.”
In the end, it was a cathartic process for Marshall to get rid of it all. “Just think how much writing and voiceover and bug testing and stuff that completely saves!” he laughs. “From a practical point of view as an independent developer, it was a lovely decision to make, because I could suddenly put a big red cross down a colossal column of things that needed doing. It was actually a relief that it worked better as instant gratification, a telling-your-ownstory kind of thing.”
Marshall based the building tech on a topdown Roguelike dungeon generator, which would carve out rooms and use corridors to connect them, the key difference being that they would be viewed from the side rather than above. It was so successful in its execution that game designer and computational creativity researcher Michael Cook invited him to speak at a conference about procedural generation. Marshall had to turn the offer down. “I couldn’t stand in front of a load of people and teach them to suck eggs,” he says. “I was still making it up as I went along and I didn’t really have any grandiose things to say about it. I’d just read an awful lot about an awful lot and done my best.”
If The Swindle’s random elements made it more unpredictable and surprising, it also led to problems Marshall hadn’t foreseen, some of which he says made a few people “quite angry”. By removing any kind of guidance, the early game encouraged players to experiment with the systems; this became a tutorial after a fashion, generating rooms players couldn’t access without bombs. “So you need to blow through that wall to get to that computer,” he says. “That was my way of doing a tutorial; a more interesting tutorial. I mean, I could bring up a little text box to explain that you can destroy walls with bombs if you’d bought them. But isn’t that much less interesting than having the player work that out themselves?” And yet some players, aware of the game’s procedural element, would simply assume it was broken because it had generated a computer in a room they couldn’t reach.
Marshall spent some time wondering whether or not he should address this, eventually concluding that it would harm the game to do so. “One of the ace things about making games is that you can take risks like that. Most people are probably just going to ignore it and not even realise it was [an issue]. But I sat there and sweated and sweated over it, because people were leaving negative reviews about how bad it was, just over this one thing.” He laughs, paraphrasing a representative Steam user’s appraisal: “‘I really enjoyed this game and played it for 45 hours, but it generated a building once where I couldn’t get to a computer’.”
This was, perhaps, a corollary of the intense pressure the game puts you under from the off. With a time limit of 100 in-game days hanging over them, some players resented feeling as if they’d wasted any of that time. It was, Marshall says, intended to evoke a similar sensation to a first, failed attempt at XCOM, with countries having pulled out of the council and your squad overrun by aliens. “The Swindle is not a massive game, it’s not a huge burden to restart it,” he begins. “Like when you restarted XCOM and you blitzed through those first few levels, because you suddenly understood flanking and you understood how to use grenades and all this sort of stuff. It was supposed to be that kind of thing.”
He’s right to insist that the 100-day limit is vital – and, more importantly, that it’s actually rather generous. One player on the Steam forums has recorded a completion time of 13 in-game days. Additionally, you can purchase an upgrade that allows you to hack into Scotland Yard and buy more time. The countdown is simply there to stop players from cutting and running at the first sign of trouble; otherwise, there’s nothing to prevent you from stealing the first bag of loot you find and scarpering. This way, failure really means something, and as the deadline nears, there’s an even stronger disincentive to cowardice. Not that players didn’t try to convince Marshall to relent: he received numerous requests to add an endless mode, but without the time pressure, the game lost one of its biggest hooks. “So much of The Swindle is standing on a ledge, looking at too many bots below you, but knowing that you need to [jump down] to hack the computer to get the next upgrade, because you’re already running low on days,” Marshall says. “It’s such an overarching thing that [the thought of] taking it out was awful. I couldn’t do that to the game.”
Marshall is happy to admit that The Swindle is a demanding game, but that, he says, is the secret of its success. “When I first started making it, everyone was talking about how difficult Dark Souls was, and then Hotline Miami came along about the same time. That’s what spurred me on to [realising] this game would be better if it was really unforgiving. That’s why there’s no safety net.” The addition of one-hit kills were another turning point, and a response to the popularity of more challenging games: prior to that, there was a regenerating health system. “It just upped the tension even more,” Marshall says. “I mean, there’s tension in every aspect – the 100-day limit, the health, the security cameras, the police turning up. It just worked so much better that way.”
After The Swindle was released, there was one further complication: the controls. A small handful of complaints had Marshall doubting himself, as he wondered how he might fix a problem of which he’d not previously been aware. After all, he’d taken the game to PAX East and watched attendees play it for four days straight, without once seeing anyone struggle. Distribution service GOG.com soon put his mind at ease. “They sent me an email, saying, ‘Dan, the controls are fine. No one brought it up.’ Because they have this filtering process, and no one mentioned the controls at all.”
Marshall was still concerned enough about a potential technical issue that he began to solicit data from these players, only to discover that it was really a matter of assumptions being challenged. “Basically, people were expecting it to be a stealth game,” he says. But it’s an arcade game; it’s got quite old-school controls. Perhaps they were expecting something like Mark Of The Ninja’s controls instead of the BBC and Mega Drive platformers I had in mind when I was making it. Do I put up a big splash screen at the start that says ‘The Swindle is not a stealth game – adjust your expectations accordingly’? I haven’t got the luxury of endlessly tinkering with it. Eventually, you have to [concede] that, OK, you’re just going to have to not like it.”
Fortunately for Marshall, plenty did. It’s happening less and less, he says, but in the weeks after launch, he would receive tweets on a daily basis from people who had screengrabbed their scorecard upon completion. The reward, it seems, was well worth the risk. “People are so overjoyed at having completed The Swindle that they feel the need to tell me. That’s amazing, right? I’d much rather some people say, ‘This is awful’, and then have people at the polar opposite utterly loving it than make something safe that everyone thinks is all right.”