Gish: ANBUKaptain’s Lament

So I was finding world 3 difficult once again, and I started thinking that maybe I’d stand a better chance of getting through it if I played on Easy difficulty. Reluctant to start over and lose my progress, I sought online to see what difference it made. My guess was seven lives instead of five. The truth: infinity lives.

OK, that’s a bit of an exaggeration. You still get only five lives per game, after which you can continue, but your score starts back at zero. But that hardly matters to a score-ignorer like myself. The big difference is that when you run out of lives and continue, you continue from the beginning of the level, not the beginning of the world as in Medium difficulty. This makes the number of lives per continue unimportant. Suddenly, legitimate incremental progress is possible! You still have to make it through each level within a single life, but with as many do-overs as you need. Even a player who aims to finish on the highest difficulty setting would find this useful as a practice mode. And as a result, I’m already partway into world 5, which may be the last one. (Steam’s description says something about “34+ story levels”, and there are seven levels per world, so. I suppose the vagueness is justified by the presence of “warp zones” off the main track.)

Now here’s the fun part: This is the page where I learned the above. This person really doesn’t like the game at all, and has fairly detailed complaints, although they mostly come down to the same thing: that the game provides insufficient guidance in the use of its frankly unusual mechanics — or, less charitably, “I didn’t figure stuff out, and I’d rather blame the author than myself”. Either way, there was a communication failure there that I think I’ve mostly avoided.

He’s somewhat coy about exactly what he didn’t figure out, and I don’t really understand why, considering that these are anti-spoilers he’s talking about, information that makes the game worse if you don’t have it, as his own experience shows. I will say what he does not: Gish has the ability to throw small blocks. You do this by getting the block on top of Gish (by turning sticky and rolling over it like a katamari), then, when its weight is deforming you, stop being sticky and tense up to resume your shape. There’s one part in world 4 where you have to get three blocks (out of a larger number available) up onto a ledge that Gish can jump to easily, but not while carrying a block: “carrying” means sticking to it, and turning sticky tends to adhere you to the floor and prevent you from jumping. ANBUKaptain apparently got past this by painstakingly building a staircase out of blocks and then somehow hauling blocks up it without pulling it apart — only to then get killed later in the level and start over again. This is clearly the wrong way to do it, not least because it’s less fun.

So, why did I discover this capability that he did not? I can state right off that I did not discover it because I was looking for it. It’s just the sort of thing that you notice while noodling around — playing with the game, as opposed to just playing it. But my usual approach is quite goal-oriented, and I can easily imagine myself in ANBUKaptain’s shoes. I suspect the real difference is I replayed the first two worlds so may times, partly as a result of putting the game away for months at a time, partly because I was trying to find all the secret areas, partly because of the crashing. When you play through the same stuff over and over again, your brain starts looking for shortcuts, more efficient ways to do stuff. You become less methodical, more willing to take risks. There’s one bit in world 3 involving a series of hanging platforms over a lava pit; the first time I encountered them, I carefully jumped from one to the next, but eventually I discovered that you don’t even need to jump: get a running start, and you can just barrel over the lot, carried over the gaps by your momentum. I’m willing to bet that ANBUKaptain never made that leap. I probably wouldn’t have in other circumstances.

In fact, there’s one bit where I rather think I did miss the point until later. The world 4 boss chases you back and forth in a hallway whose ceiling sports three large stone blocks with crumbly blocks underneath them, supporting them. There was a similar set-up back in world 2, where I had passed it by jumping up, clinging to the crumbly blocks, and breaking them by tensing up and increasing my weight (and then getting out of the way quickly before I was crushed). The world 4 boss level makes this approach impractical: the ceilings are just a little too high to jump to, and the tiles around the crumbly ones are slippery ones, making it impossible to just climb the wall and move horizontally. I had notions of pushing loose blocks underneath so I could jump from a higher vantage, but my adversary kept pushing them away. At some point in this process, though, I discovered that I could demolish the crumbly tiles by throwing the loose blocks at them — difficult to do with any precision while you’re being chased, but easier than the alternative. And I suddenly understood something I hadn’t before: why there was a loose block sitting under those crumbly tiles back in level 2. I was supposed to have used it the same way.

So, the lesson here is to trust Edmund McMillen. If you’re doing something difficult and tedious, there’s probably a better and faster way. If you find a loose block, it has a purpose. And this suddenly makes me rethink another place where I found a seemingly-purposeless block, sitting on a sliding floor piece that I needed to shift but had difficulty getting a purchase on. I had gotten through that through awkward shuffling and stickiness, but now I suspect there’s a more elegant solution involving Newton’s Third Law.

Gish on Mac

One nice thing about the Steam Play initiative (Valve’s nascent cross-platform support) is that it makes it very easy for me to find out when games I’ve purchased become available for the Mac. This is an important thing to know for those games that don’t work right on my PC. Just the other day, I noticed that several of my indie bundle games had been quietly ported while my attention was elsewhere. My first instinct was to finally try And Yet It Moves, which I haven’t yet been able to get to run on my Windows machine at all, but I can’t get it to run on my Mac either: the download is eternally stuck at 99%, and attempts to run it anyway yield silly errors about the servers being busy. So instead I gave Gish another shot. I might as well; I’ve bought it in a bundle at least one more time since my last attempt, for something like five times total by now.

You may recall that the last time I played this game, it was crashing on me frequently enough that I figured out how to exploit the crashes to aid my progress. Without that help, the game is in a sense easier. I hold myself to lower standards, not seeking every secret or every coin, just trying to get through the levels as fast as possible. The first world breezes by when approached like this. It’s quite freeing; I get to do all the acrobatic stuff that I mentioned back in my first post — which, it turns out, I still remember how to do.

Which is fortunate, because it isn’t at all obvious, and this game has a pretty steep learning curve. In a recent online discussion, someone asked “Did anyone actually like Gish?” — to which the answer is obviously yes, because it won some awards, but it definitely doesn’t give the player the sense of immediate power and ease of movement that most platformers strive for, and that probably turns a lot of people off. Another discussion I recall pointed out how Mario 64 engages the player by making it look like Mario is really enjoying himself, running around and leaping into the air and shouting “Woohoo!”, to the point that it almost seems a shame to put the controller down and deprive him of his thrills. Gish enjoys himself too, opens his mouth wide in a wicked toothy smile when he’s fast and airborne, but it takes a degree of mastery to reach that point.

One thing I keep forgetting: one of the developers on Gish was Edmund McMillen, who went on to create Super Meat Boy. SMB is also too difficult for a lot of people (possibly including me, although I haven’t given up on it yet), but for opposite reasons: moving around in ordinary environments is almost too easy, with the result that you leap into sawblades all the time. At any rate, I give him credit for exploring extremely different points within the possibility space of the platformer genre, even if both of these games are at heart glorified Mario imitations.

Super Meat Boy: Following Trails

I played a little more SMB. I’m still stuck in Hell, but I managed to unlock another character: Ogmo, from the Jumper series. I recall trying one of the Jumper sequels a while back, probably Jumper 3 when it was featured on Play This Thing. It seemed a decent platformer, but I wasn’t in the mood for it at the time. But hey, if there’s one thing I can use SMB for, it’s as a framework for recommendations. Seems to me I should at least try out the games that McMillen liked enough to invite to the party.

So, I looked at the unlockable character roster. So far, I’ve got Ogmo, the Headcrab, Commander Video, and Jill. Ogmo I’ve mentioned; I’ve downloaded the original Jumper and played it a bit, but it gets very difficult very quickly. The Headcrab is from Half-Life. Half-Life isn’t a platformer, and the headcrab isn’t its player character, but I suppose someone wanted a Steam-exclusive unlockable, and this is the only thing in Valve’s library that’s known for jumping. At any rate, I’ve played the heck out of Half-Life (although I need to go through Half-Life 2 again at some point, now that they’ve added Achievements). Commander Video is from Bit.Trip Runner, which is a Wii game, not available on any system I own. One of its predecessors, Bit.Trip Beat, is out for PC, but it looks like a fundamentally different game; if further Bit.Trips are ported, I may get them as a package, but for now, I’ll give it a miss.

Looking at unlockable characters I don’t have yet, I noticed one from a game that had garnered praise but which I hadn’t tried: Runman: Race Around the World, which can be described as Sonic the Hedgehog with everything that isn’t directly related to running fast taken out, including death. Downloading that, I see it’s done in a crudely-doodled style. No surprise there — I could tell that much from the screenshots and demo video. But somehow, seeing it in-game made me look at it better, and it looked very familiar — the drawing style reminded me a lot of An Untitled Story, a Metroidvania-style platformer I had played but not finished a few months ago, concerning an egg that falls from a nest and, after fighting a few bosses, hatches into a bird that fights more bosses. It had art that was clearly drawn with magic marker.

Googling, I discover that, indeed, one of the co-authors of Runman is Matt Thorson, author of An Untitled Story. Furthermore, he wrote the Jumper series, as well as a couple of other platformers I know: Give Up Robot and Moneysieze. I had played Moneysieze quite a lot last year, and meant to write it up here, but never got around to it. It struck me as fairly ingenious in its unconventional use of famliar platformer mechanics. For example, the double-jump. In many platformers, you can hit the jump button a second time at the top of your arc to gain additional height. In Moneysieze, you could perform the second jump at any point in your trajectory — which means you can use it to pass under obstacles that extend below your starting point. I thought this was clever, but now I see that the same author had already pulled tricks like this in Jumper. SMB treats the double-jump as Ogmo’s defining trait, and the warp zone where you acquire Ogmo for general use requires executing trick jumps of exactly this sort.

I’m a little shocked to discover how much of Thorson’s work I’ve experienced without being aware of him. I notice now that the Play This Thing writeup of Jumper 3 actually mentions that Thorson is half of the Runman team, but apparently that fact made no impression on me at the time. Well, if part of SMB‘s mission is increased awareness of indie platformers, mission accomplished. I considered myself pretty aware already, but it looks like I wasn’t aware of my lack of awareness. I’ll be watching for Thorson’s name in the future.

Also, for what it’s worth, Runman‘s level-selection screen plays a recording of Helen Humes singing Song of the Wanderer, the same background music as the level-selection screen for Immortal Defense. I suppose there are only so many public-domain jazz recordings out there, and Runman uses many of them, but unless this is a deliberate reference, it’s a strange coincidence. Or maybe there’s just something about that song that suggests “level select screen” to indie developers? I’m definitely going to use it for that purpose if I ever write a game with a level-select screen. It’s too good an in-joke not to share.

Super Meat Boy: Hell

I spent a bit more time on Super Meat Boy last night, beating one more boss (or, well, world-end-level; it was more of a race against time, without any real boss monster to beat) and then going to Hell. “Hell” is that game’s name for its traditional lava-and-fireballs world. I suppose there’s an ice level next? A big part of this game’s schtick is riffing on 2D platformer clichés. The first world is an idyllic forest, just like the first world in every Sonic the Hedgehog game, except there are enormous circular saws mounted all over the place for no apparent practical purpose, and I don’t think Dr. Robotnik ever went as far as to just burn the whole forest down — something played for laughs here. Sonic was marketed as the bad-boy counterpart to Mario, but Meat Boy, with his irrepressable grotesque-cuteness, his glee in the face of repeated gory death, and his utter disregard for censors or parents, has him beat hands-down in the bad-boy department without even making a big deal of it. And Hell is part of this: lots of games have lava worlds, but only a few are so forthright about what we’re all thinking. (Fun fact: the Japanese version of Um Jammer Lammy has Lammy die, go to Hell, and escape. The North American version timidly replaced this whole scenario with a tropical island, robbing Lammy of her heroic journey’s most directly mythic component.)

Fittingly, Hell seems to be the place where the difficulty ratchets up to just beyond my abilities. I may well change my mind about this — a good platformer makes things seem like they’re beyond your abilities but then trains you up to the point where they’re not. I remember Crash Bandicoot as being particularly good about putting collectibles in seemingly-impossible places that I skipped over on the first pass but came back for later with greater confidence. The “Veni Vidi Vici” sequence in VVVVVV looked daunting at first, but yielded to persistent practice. Still, those are both matters of hunting for optional collectibles. Here in SMB it’s the main path through the game, and I’m not even on the last world yet. I’m winding up doing the opposite of what I did in Crash and VVVVVV: going back to find collectibles and bonus areas because it’s the easier alternative.

Well, I knew what I was getting into when I started playing. This game has a reputation for extreme difficulty. I’ve seen this school of game design described as “masocore”, although there are differing definitions of that — the author of Super Jill Off contends that a true masocore game has to subvert genre expectations. I suppose SMB does that to some extent, though. Just the sheer abnormal distance that you can leap is something of a subversion, in that it allows the designers to create levels where the best and safest route through a series of obstacles is to just clear them all in one go.

Super Meat Boy

Maybe my perspective on things is skewed — I don’t pay much attention to the mainstream gaming press, and the blogs I read tend to focus on indie stuff. But then, indie stuff is big enough these days to get official recognition on consoles. Regardless, it really seems to me that this year, the year that gave us the long-awaited Starcraft sequel and the most significant World of Warcraft expansion yet, the title that’s generated biggest buzz has been Super Meat Boy. (Or possibly Minecraft, but that’ll have to wait for another post.) It’s being called the apotheosis of the 2D platformer, the ultimate expression of the form. And it encourages this sort of thinking by being kind of a living summary of what’s been done before, full of references to other games.

The most obvious references are the unlockable characters, mostly from other indie platformers — Braid, VVVVVV, and Mighty Jill Off, to name just a few — most of which I’m familiar with, some of which I’m not. Like the Smash Bros. and Kingdom Hearts series, it suggests that all these games are part of the same family, a sort of indie platformer club. Also, the characters carry with them an approximation of the mechanics from their source games, which effectively makes them demos for any of the games that you haven’t played. I have to wonder how much SMB has affected sales of these other games, and how much this was a factor in the decision of their creators to allow their inclusion. (It doesn’t have to be a factor at all — game developers are quite capable of making agreements like this just on the basis that they think it would be cool.) But viewed from the other side, it’s effectively making a statement that SMB is a generalization of the platformer, broad enough to include all these other games within it.

Some of the unlockable characters are accessed by collecting bandages (collectibles in hard-to-reach places), others are located in special “warp zones” that make you play through a few levels in the style of their games of origin. There are also “retro” warp zones that use the normal Meat Boy mechanics, but in the graphical style of, say, a NES or a Gameboy (or even a glitched-out version of same) — another kind of reference to things that have come before, this time appealing directly to the nostalgia factor. Note that any reference to a modern platformer can also be an indirect nostalgia appeal, because the nostalgia factor is pretty big in 2D platformers to begin with. The three examples I gave above of games that provide SMB with guest stars are heavily based on specific older games — Braid on Super Mario Brothers, VVVVVV on Jet Set Willy 1Actually, Terry Cavanaugh says he never played Jet Set Willy and that VVVVVV was really inspired by the games that imitated it, making this even more indirect., Mighty Jill Off on Mighty Bomb Jack. I didn’t pick those three games with this in mind. It’s just that the 2D platformer genre has become so intra-referential in modern times that it’s hard to avoid. SMB embraces this tendency a little more thoroughly and inclusively than most, to the point that it becomes recursive: its referencing of other games is itself a reference to those games referencing other games.

Then there are subtler shout-outs. I’ve been through one level that’s a blatant imitation of Canabalt, but it’s only blatant if you’re familiar with Canabalt. This makes me wonder what else there is that I’ve been missing. I’ve found an article explaining how the world intro cutscenes are all shot-by-shot imitations of intros from various classic games, but it’s the sort of thing where there could easily be references that no one has even noticed yet.

1 Actually, Terry Cavanaugh says he never played Jet Set Willy and that VVVVVV was really inspired by the games that imitated it, making this even more indirect.


Gish is a goth/cartoony 2D platformer starring a sentient ball of tar, a hero that stretches and splats and sticks to things. In some respects it’s very traditional, based around Mario-old conventions like the kidnapped-girlfriend plot 1No, Gish’s girlfriend isn’t a ball of tar. She’s a cartoon goth girl. Don’t ask me for details. , the linear sequence of levels grouped into “worlds”, enemies you kill by jumping on their heads, etc. I think the designers chose to adhere to old and even outmoded conventions as much as they did, not out of lack of imagination, but to give the player something familiar to cling to as they pulled the rug from under you on the mechanics and controls. This is a platformer where jumping is difficult to execute.

Oh, sure, there’s a “jump” button, but it doesn’t do much of a jump by itself. To make Gish do a jump worth jumping, you have to hit the button while he’s compressed — the more compressed, the better. And he’s at his most compressed when he’s in the middle of colliding with a hard surface, such as a floor. So standing jumps are a no-go, but sequences of long bounds are doable, once you’ve attained sufficient facility with the controls to carom about with confidence. Momentum is your friend, and hesitation is your biggest enemy.

It takes a while to become that confident, because, aside from the jump, Gish’s core abilities are not standard platformer fare. The core controls allow you to make Gish stickier than normal (useful for climbing walls), slicker and less viscous (useful for sliding down narrow chutes), or heavier and more resistant to deformation (useful for breaking things or sinking in water). These can be combined arbitrarily: sometimes, for example, the easiest way to shift a pile of loose blocks is to trickle into their midst through slickness and then tense up to resume ball-shape and force the blocks apart. It’s all very physics.

I may not be playing it much, though, because it’s still crashing for me. Not as badly as it was before, but I’m typically getting in about 10-15 minutes of gameplay before it exits to the desktop without so much as an error message. This is, however, long enough to make permanent progress, so I really could just keep playing, as long as I exit the game every once in a while to force it to save. (There is no manual save option.)

1 No, Gish’s girlfriend isn’t a ball of tar. She’s a cartoon goth girl. Don’t ask me for details.


In a recent blog post, Edmund McMillen talks about his confusion-and-insanity-themed puzzle platformer Time Fcuk. It’s an interesting read, but the one bit that stood out for me was the bit about the alt levels. Apparently certain levels had multiple versions, chosen at random:

i came up with the idea late one night where i envisioned people playing the game and then trying to look up hints on how to beat a level only to find no one had played the level they are on, in hopes that they would feel “crazy”. this of course didn’t have the effect i wanted…

And yeah, it certainly didn’t have that effect for me. In order to notice the variations, you’d have to either replay the game from the start and notice that the levels were different, or compare your experience in considerable detail with someone else’s. And the player doesn’t really have much motivation to do either: if you like this sort of game, you’ll probably play right through it to the end, and if you don’t, you’ll probably just quit in the middle and not go back to it. The interesting thing is that, while Time Fcuk didn’t inspire this sort of comparison, another recent game did in a pretty big way: Dungeon, a retro platformer by Cactus and Mr. Podunkian.

Dungeon is described by its creators as an “experiment”, but feels more like a prank, or possibly even a troll. The concept was that the game had a number of deliberate problems, bugs, and other causes for complaint, which caused people to post on its TIGSource message board — but different installations would provoke different problems. The game apparently uses a deterministic combination of factors such as the OS version and the current username to produce a consistent experience for each player, even as the content varies among different players. So, some players experience frequent pauses, others get monsters that move far faster than they should, others find a certain jump early in the game impossible, etc. (When I played it, I was lucky enough to get variant #7, in which the only issue is that the level titles are artsy and pretentious.)

The forum comments on Dungeon start off as confused as you might expect, with comments along the lines of “What are you talking about, that jump is dead easy”, but it didn’t take long for people to figure out what was going on. The first clues started coming in when people found that they could fix their “bugs” by running the game in some Windows compatibility mode or other, which alters the OS version seen by the game app. Speculation that the game “modifies its own difficulty depending on the machine or something” started less than an hour after the game was released; by the end of the day, people were starting to compile lists of the variations.

So, I guess the lesson here is that if you want something about your game to be noticed, make it obnoxious.