By SriK / December 3, 2011
Canabalt was made in 5 days, for an amateur
game development competition themed around "the bare minimum". It's probably the
best a game made under those circumstances is going to get, but that doesn't make it a great game (or even a good one), and it certainly
doesn't make it an "instaclassic" (Edge) or "perfect" (Action Button). What it does do, however, is help explain
just why Canabalt sucks as much as it does, so let's get into that right now.
Given its lack of an end point or any significant stage development, Canabalt is obviously trying to follow an early-'80s arcade model, so first let's view it as a retro effort and see how it compares against games released in that era. Take, for example, two popular '80s arcade games: Galaga (1981) and Q*Bert (1982). Can you remember how, even though these old games didn't have much in the way of stage progression (since they weren't designed to end), they at least had difficulty progression, where the further you get the more challenging the game becomes? This is a crucial feature which Canabalt lacks; it doesn't ever increase in challenge at all, and you could be presented with the exact same set of obstacles at 200 points as at 20,000.
And that brings us to Canabalt's other glaring flaw: its randomized obstacle design. This seems to be a popular gimmick amongst
"indie" titles; Derek Yu's Spelunky and Jason Rohrer's Inside a Star Filled Sky make use of randomly generated levels
as well. Unfortunately, just because a feature will net you
a Technical Achievement Award at the IGF doesn't mean it'll make your game better; in fact, in Canabalt's case, all the randomization
does is make the game even worse. See, the cool thing about good arcade games, and good videogames in general, is that they always give you the
tools to overcome the obstacles that are ahead of you; if you screw up, then it's your fault and no one else's. However, thanks to
Canabalt's extremely limited control scheme and random level design, you can often find yourself in a situation where you're accelerating
for buildings at a time with no way to stop, and then the game throws a wall at you. The way you can control your speed is by bumping into
the copiers and other (again randomized) obstacles which are placed in your way to slow you down, but even when you're given an opportunity to do so you
have no idea whether you should go for it or avoid it just in case the game's about to segue into a mile-long bottomless pit
five seconds on. This problem could at least have been alleviated slightly if a) you could see 4x or 5x as far in front of you, so that the decision to slow down or not was more than a pure guess, or b) the level was designed by hand and you could maybe memorize
it, but whoops, looks like you're getting an entire new level full of random junk each time you play
(because hey, 5 days isn't enough time to design a lot of different and varied levels, we are on a tight schedule here!)
Combined, these factors reduce the game's depth to even lower than that of an early-'80s arcade title. There's a diminished correlation between skill and performance; either you get lucky and the randomizer doesn't fuck you over, or you don't. In short: Canabalt is Russian roulette, remade into a Flash game.
But let's stop talking about this game solely in the context of the 1980s. Even if Canabalt implemented an actual difficulty curve and fixed the flaws stemming from its randomization, at some point we have to take into consideration whether a game based on an early-'80s arcade model would even really be worth playing in the year 2009 (when this game was released), especially one with such a simplistic and limited control scheme. I mean, of course that type of game is worth making if you enjoy game development in general and only have 5 days to put something together for a contest [Assuming, that is, that for some unfathomable reason you felt like wasting your time to even enter such a pointless contest. -Ed], but it is not necessarily worth playing, because, unbeknownst to "indie" game developers and other scammers, videogames have come a long way since Galaga and Q*Bert. Recently, for instance, I've even been playing a cool 1989 arcade platformer called Chelnov, which is like Canabalt with the momentum mechanic removed and everything else done a thousand times better; levels still autoscroll, but they're designed by hand meticulously instead of being generated on the fly by a coarse algorithm. Chelnov's more sophisticated control scheme also allows you to not only just jump but also control your speed and shoot in 8 directions, and as a result enemies are able to do more complex things than just slow you down or blow you up when you hit them. A minimalist 1980s-styled platformer with no stage progression simply can't compete, especially if it's a complete failure like Canabalt.
The best thing I can really say about Canabalt is that its aesthetics aren't too bad. The game's monochromatic palette contains only 6 shades, the foreground art is fairly simplistic, and the stage background is mostly composed of building silhouettes and maybe a giant robot silhouette or two moving around, but it all comes together very nicely, and the animation is quite fluid, with the shattering glass and flying doves looking particularly cool. I'd have normally expected a bit better from AdamAtomic [Who is such a knowledgeable pixel artist and programmer that he has even managed to include an "HD" version of his pitiable five-pixel artwork by blowing it up to fill your 40" 1080p screen lol. -Ed], who not only has a pretty cool PixelJoint profile [Wow! PixelJoint! Lookout Japan, this guy are serious! -Ed] but also helps administrate Pixelation (one of the foremost Western pixel art forums [Wow! He administers a forum! That must bode well for his artistic skills, since all the great artists today are into forums, and especially into administering them! -Ed]), but hey, 5 days. The music by Danny B is actually pretty cool, though, but again, only one track, etc.
But of course, these small positive aspects can't redeem how Canabalt falls flat in virtually every other way. In conclusion: if you want a fast platformer centered around momentum and speed, play a Sonic title; those games even let you control the main character's movement, and as a bonus have actual level and enemy designs which vary from stage to stage. If you are considering Canabalt's iPhone port and are looking for a decent iPhone game to play on the metro or in the dentist's office, get Rage HD or one of Cave's STG ports; the latter will be a neutered, watered-down experience compared to the real thing, but they are the best you're going to currently get on a cellphone with a touchscreen. If you want a simple game which only needs one button to play, try a video player; I hear they even use Blu-Ray discs these days, just like the PS3, and the graphics are photorealistic. All of these things would be, at the "bare minimum", a far better use of your time than playing Canabalt.
[And if you want to see what's happening at the cutting edge of the old and venerable genre that Canabalt completely mangles, play Goku Makaimura or Hard Corps: Uprising, more on which in their respective upcoming Insomnia reviews. -Ed]