The Perfect Game The World Forgot

The story of the best physics game ever made. And what the modern gaming world can learn from The Fight of the Sumo Hoppers.

The first time I played Fight of the Sumo Hoppers, I didn't stop until my hands gave out; either the webbing between my fingers started burning or my forearms got sore from splaying my hands so wide for so long. I can't remember, exactly. It was mid 2000 at the latest.

It was clear to me then that this dead-simple DOS title was a great game. Today, more than 10 years later, I think it might be perfect.

Sumo Hoppers has no plot and offers virtually nothing in the way of player progression; there's hardly even a sense of a persistent "player" at all. Its graphics are purely functional: lines just thick enough to see, players distinguished by primary colors. The physics engine, as far as I can tell, is incredibly simple. Characters have mass, their feet have both zero width and infinite traction, and gravity is about a third as strong as it is on Earth. Two players are locked together with an inflexible bond; only their legs can move. First one to hit the mat — or a wall — headfirst loses. That's it.

The game was released in 1997 and updates appear to have stopped in 1999. It was an indie game before indie gaming had a name; it was released as the internet was maturing and DOS games were dying — in other words, into a perfect storm of obscurity. Its lone creator, Tuomas Korppi, hasn't made a game since.

But Sumo Hoppers can count among its fans some of the most interesting and influential minds in the gaming industry. Bennett Foddy, Oxford professor and creator of ultra-simple, ultra-addictive games such as QWOP and GIRP, has even created a web-based tribute called Get on Top. "It was actually Lee Perry, the designer behind Gears of War, who put us on to it," says Foddy. "I had played a bunch of DOS freeware physics games back in the day," he says, "which had a big influence over me making my physics games."

"Doug Wilson (of JS Joust fame) was talking to Lee at GDC last year, and he mentioned that Sumo Hoppers was one of his all-time favorites," Foddy explains. "We were really excited when we saw it. It's exactly in the spirit of the games we're trying to make: hardcore, ridiculous hot-seat sports games for two people on the one computer."

Korppi, the Finnish mastermind behind the game, considers it a high watermark of a past life. "When I was younger, I wrote quite a lot of simple computer games," he says, "but Sumo Hoppers was by far the best of them. So maybe it is better that I do not mention the other ones."

He seems almost bewildered by his game's underground success. "I knew Sumo Hoppers is a good game, but it was a bit strange, so I was not sure if it would be appreciated by others," he says. "Fortunately, it turned out that also others appreciated it."

Korppi doesn't design video games anymore. In fact, he doesn't even play them. In the time since releasing Sumo Hoppers, he's turned his attentions elsewhere: He ran for Helsinki city council under the banner of the Pirate Party — "in our system, it is advantageous for a party to have as many candidates as possible," he says of the ultimately unsuccessful campaign — but spent most of his time focused on academia. "I have ... done some mathematics, obtained a Ph.D., and even managed to publish two papers on the interplay of nonstandard analysis and algebraic topology," he says.

Sumo Hoppers was born out of a physics experiment, more or less. "I was experimenting with spring systems," says Korppi. "I had an idea for a karate game where the karatekas would consist of springs, and the player would control the karateka by controlling individual springs."

"When you develop physics-based games, you do not program the gameplay directly," he says. "You program the physics simulation and hope that it will provide a good game."

Early versions were too complicated. "The game was too difficult," he says. "Trying not to fall was enough challenge."

Foddy has dealt with similar issues in his games. "One of the nice things about using physics simulations in games is that you get good 'game feel' (something that designers really strive hard for) for free," he explains. "In my view, the downside to physics games can be that they get too chaotic and out of control," he says. "It can be hard to understand why you won or lost in a game like Crush the Castle or QWOP. It's too hard to control the outcomes, so it can wind up feeling a lot like trial and error, unless you practice a lot."

Korppi's approach was to tie the game down a bit — or more accurately, tie it together. "I once watched sumo on TV and happened to think of the bodies of the two wrestlers as one system," he explains.

"The genius of Sumo Hoppers is in tying the two players together with their arms, which is an elegant way of radically simplifying the game," says Foddy. "It's still deep, chaotic fun, but you have just enough control to be able to develop a sense of mastery."

The result is, indeed, chaotic, and the learning curve is steep. But it's short: After a few dozen games, your fingers get used to the controls — five keys for leg placement, three for extension — and the game pulls you in. The only real flaw is outside the game: In multiplayer mode, two players share 16 keys across one keyboard.

"My major surprise was how good a game Sumo Hoppers turned out to be," says Korppi. "When I got a prototype working, I remember that I thought that I can never make another game that is as good."

Sumo Hoppers is still free to download here, and runs in DOS or any DOS emulator (Mac users should try Boxer). Foddy's Get on Top is a simply designed homage; it's browser-based and playable here. "I let the players flex their shoulder muscles," says Foddy, "and I think that turns it into a pretty fundamentally different game than Sumo Hoppers, in terms of how you play it."

Otto-Ville Ojala's Wrestle Jump is a more faithful remake, also playable online. Korppi, who wasn't aware of either of the tributes until our interview ("quite nice," he said), tells me he won't be making another Sumo Hoppers game: "The major improvement I'd like to see on Sumo Hoppers is the play on the net, since that would solve the problems with two people on the same keyboard. But I do not think I will ever do that. If you will see other games from me, they will probably be board/card games."

I Iike to think Sumo Hoppers' legacy, or perhaps its lesson, runs deeper than a few tributes. The year Sumo Hoppers came out was, coincidentally, a big year for physics in video games: Jurassic Park: Trespasser, a disastrously bad game that happened to be the first open-world first-person shooter with a realistic physics engine, shipped. It took quite a few years for the first good physics-centric shooter to come out — that would be Half-Life 2, the sequel to the best game of Trespasser's era — and since then there have been plenty more. Yet the trend more recently has been to limit physics, not unlike Korppi did in Sumo Hoppers. "Most AAA games have physics these days, but the physics is confined within 'hard' gameplay rules in order to avoid glitches and exploits," says Foddy. "For example, in FIFA Soccer, the physics is subordinate to the motion-captured player animations, and you'll very rarely see a player land on his head. Or in Dark Souls, your enemies collapse into dynamic corpse ragdolls, but they're not allowed to exert any force on the player's character." Or take, for example, the Crysis series, which started with an open-world, physics-rich, gameplay-deficient title and appears to be culminating with a locked-down linear shooter that, while flawed, is better.

The difference, of course, is that Korppi simplified his (admittedly far simpler) game without compromising the purity of his physics simulation — nothing is scripted, just selectively fixed in place.

That, for a game based on dynamic rules about how the world moves, is the holy grail. And Sumo Hoppers is one of the few games to ever find it.

Skip to footer