Kotetsu

monuments of mars (apogee, 1991) may be one of the first games to ever contain a physics puzzle

p.s.

1. i have since looked up a video of the solution and it turns out you don't need to do anything more than steal one box and make it to the left side, so it's not nearly as complex as it appears, but it's the last stage in this episode so i wonder if it originally was harder, and then the designer realized most people would never be able to manage it

2. the color palette here is completely different because i changed it with CHGCOLOR.COM

This game, as well as arctic adventure and pharoah's tomb, are some of the jankiest, most computer-ass games ever to come out for the PC. this is interesting to me for a lot of reasons

updated: I took a look at Arctic Adventure and Pharoah's Tomb, the other games based on this "engine" and found they were considerably different and much less janky. It looks like this game is pretty much the product of Todd Replogle alone, so I rewrote several parts.

Notably: I also found out that Replogle, after working on all the Duke Nukem games up to 3D, proceeded to lose his mind and disappear into Thailand to become a real estate scammer. His last recorded words were in an interview where he proudly declared that he had figured out how to rip off the american public by shorting precious metal commodities. You can look that up.

For one thing, this kind of mess of a game was much more common on home computers like the ZX Spectrum and PC88. it honestly makes me wonder if the developer, Todd Replogle, had seen a couple home computer puzzle platformers and was trying to recreate them from memory, because the energy is just so specific.

For one thing, there's the game mechanics themselves, which are very reminiscent of the early to mid 80s. It's a puzzle platformer with simple rulse: you pass through a series of single-screen rooms, assembled from a simple tileset of blocks, ramps, up-down and left-right moving platforms (that sometimes need to be activated with a switch), spiked ceilings and floors, vertical and horizontal laser walls (that can sometimes be turned off with a switch,) lasers that shoot when you move past them, left-right moving enemies, and keycards that are needed to open doors.

The goal is simply to reach the exit, which is almost always behind a door, but sometimes not. On a few levels you can simply walk to the exit immediately. There are enemies and platforms, but the only reason you'd brave them is to pick up bonus points. Sometimes the letters M A R S show up, and collecting them all gets you 10,000 points. There are also many challenges in the game that get you nothing but bonus items; If you don't care about score, you can ignore 1/3 of the game.

You have a gun to shoot enemies, and some levels make shooting enemies absolutely required. This is where the 8-bit home computer style jank begins: Some levels have no ammo pickups, and you start with only the bullets you had when you left the previous screen, but this means you need to manage ammo across levels. This has deeper ramifications than you'd expect for a game of this era.

Most puzzle games of this sort would have a password system; this game instead has save files, which seems cool and forward-thinking, except that your ammo is saved too. As I write this, I'm actually softlocked on episode 3 level 11 - I literally cannot move forward without a bullet to blow up a mandatory enemy, but I used all my ammo on the previous level. As far as I know, my save file is ruined. My choices are to restart from level 1 and use my bullets more judiciously, or hex edit the save.

I don't know about you, but this feels so computer to me. It feels like the kind of design decision that would be seen as unthinkable by any console publisher, even the bad ones. To persist a player fuckup cumulatively across levels, and have it catch up to them potentially hours later? I can't even name another game that does that. Not one.

There are also "mandatory secrets." If you've ever read about Tower of Druaga or Super Pitfall, this is like those. Most levels contain switches you can flip, and some have switches you can shoot, but many levels also contain invisible switches. Sometimes, you jump into an empty space or fall through a shaft, the game freezes, and the level reconfigures itself; some number of blocks appear or disappear. These are usually not for bonus points, but are actually required to reach the exit - and thus, some levels literally cannot be solved by static analysis, because the solution is invisible until you touch it. You simply have to play, get stuck, and then try jumping into random spots in case there's a solution.

I swear I've played games on the ZX Spectrum or PC88 that play exactly like this. Like, to a T, nearly identical, and with very similar graphics, just in different colors. This feels a decade older than it is - which is cool! Early 80s computer games were very creative! Unfortunately, they also usually had very shaky physics.

I've played a LOT of shareware PC action games - many hundreds - and I can tell you that most PC games that tried to mimic console action titles failed miserably, but in a very distinct way. Many, of course, were Amiga ports, but those sucked in a distinctly Amiga way. They weren't poorly programmed, they were just sludgey and miserable, with the floaty, imprecise physics typical of the platform. PC exclusive titles however were their own kind of bad.

Even the bigger titles, like Epic's Jill of the Jungle and Jazz Jackrabbit, had very amateurish movement physics, even if they had decent art and design. The smaller titles on the other hand were usually worse than amateurish. Billy the Kid Returns (1993), made entirely by one guy, is full of nearly empty rooms that reek of deluxepaint gradients and built-in textures. Motion is chunky and linear, with no acceleration or deceleration; jumping just makes you move up at a fixed speed, then stop and move back down at that same fixed speed.

Starting in the 90s, even bad console games started having solid fundamentals. If you don't believe me, go watch early Chrontendo episodes and look at how awful most 80s platformers were. Try Layla for the NES, and then go pick up any $5 indie platformer on Steam. It's night and day. When a modern indie dev sets out to make something that feels like Mario, feels like Castlevania, or feels like Megaman, even if the result isn't fun, it's still within inches of the bullseye from a technical standpoint.

But the resources that modern indies have access to are immense and inconceivable in comparison to what someone in 1993 had. It's very possible that Billy The Kid Returns' single programmer didn't even have an internet connection, let alone source code to study and learn from. It was probably just very hard to make something remotely as good as a Capcom or Nintendo game back then, and this was, tragically, one of the better examples. Many PC platformers were unplayable.

And yet, they all kinda sucked in similar ways. I don't recall ever really experiencing, like, bad platform physics for instance - it could be hard as hell to REACH a platform, or engage with a ladder or whatever, but mostly just because the level design was crap or the jump arc was very shallow. Generally, once you adapted to the movement, when you went to jump on a platform you'd land on it. Even if your gun or melee attack or whatever was weak or unsatisfying, you could still usually hit whatever you aimed at.

A lot of games on 8-bit computers were not like this; their jankitude was on a whole different level. Probably owing to the tiny, slow CPUs in these machines, they often used very shortcut-laden collision detection, so missing platforms was easy, getting stuck in a solid surface after a jump was common, bumping your character into the corner of two walls often had strange and unexpected effects, and bullets often just passed right through enemies.

Normal gameplay in these titles often looked like Metroid speedruns, where a runner lets a door close on them, then jumps repeatedly to move upwards a pixel or two at a time. While most well regarded console action games, and a lot of early PC ones, have a very concrete notion of whether you're "standing" or not, these games often just... didn't. As the speedrunners say, walls were a suggestion.

This is what Monuments of Mars is like. Everything reeks of jank - sorry, but there's no better word, this game is janky as hell. Simply walking along a flat platform has a strange, lurchy feeling to it. Climbing a slope consists of your character sliding forward a pixel at a time for a couple frames, then jerking upwards a few pixels all at once. If you walk down a slope, you see your character move forward a few pixels, suddenly jerk downwards, penetrate the slope for a moment, and then pop back up, as if the game is falling asleep mid-step, then jerking awake and realizing it forgot to check if you were on the floor.

If you step off a moving platform as it moves away from a wall, you will sink down five or six pixels and become wedged in the wall. You can then jump off of it - literally a Mario speedrun glitch strat. If you jump towards the corner of a platform and don't quite make it, don't worry! Your foot will get stuck a couple pixels into the corner, and you'll then walk up and through the corner as if it was a slope.

Jumping feels "sticky"; you have to hit the spacebar a few frames in advance of when you want to leave the ground, and then the very low framerate makes the jump look just as jerky as walking. When you land, you frequently sink into the surface for a moment before popping back up. Holding spacebar, naturally, makes you jump the moment you land, and triggers at extremely high speed if you're under a low ceiling. Classic.

Interaction with the environment is also weird. It seems like the game only polls for keyboard inputs every X frames, so sometimes when you try to hit a switch, it just doesn't take until you hit it a couple more times. And pickups are... semisolid.

Sometimes when you jump into an ammo or bonus item, you pass through it as you pick it up; other times you bonk. This seems random, but I suspect it's really a matter of how fast you're going. My guess is that motion collision and item collision are different routines which don't run on the same frame, so depending on how fast you're going, or maybe what frame you happen to be on when you hit something, either the "hit wall" or "pickup item" code runs first.

This means that if you're riding a moving platform and you try to jump past a bonus pickup to get to a higher ledge, you might bonk the pickup and INSTANTLY begin plummeting straight downwards. So to get a clear field, you have to tediously wait on these platforms as they grind back and forth, then jump up and collect the bonuses from directly underneath, one at a time.

Did I mention that the moving platforms lag behind you? It's fortunate that they actually move the character at all - many Spectrum/PC88 games forced you to walk along with moving platforms. But again, the platform and player are clearly not moving on the same frame. The platform moves, then the character moves. They are separate events that simply inform one another.

Going back to item collision: it works from both directions. The game sometimes registers the top of an item as a platform, so if you hold down spacebar while in midair, you can sometimes boost off of a pickup in midair. You can also sometimes stand on enemy heads, which is maybe intended behavior, but I strongly suspect it isn't. Both of these would make magnificent speedrun strats, letting you skip whole levels.

The falling bricks in the gif above are their own whole story. I don't think they appear until episode two, so I didn't know about them until today. They're sort of, kind of decently simulated, all things considered, but you can also tell that they're... I have no better word for it: janky.

This gets into the reason I love this game, unironically and sincerely: Todd Replogle did not care that the physics sucked. I bet he was incredibly proud of them, in fact, because there are a number of levels that hinge on these falling barrels, and they demand ASININE behavior from the player.

I wrote the other day about how Doom and Metroid frustrated me as a child because, while I could see HOW to complete them, I thought for SURE that no Professional Game Developer would ever ask a player to take actions that poorly telegraphed. Doom wanted me to run at a 45 degree corner of a ledge in order to clear a gap and pop back up onto the opposing ledge by abusing the stair physics - fuck off! You just have to GUESS that this is possible! Who asks a player to do something that unobvious?

How about Todd Replogle, who decided to start multiple levels in Monuments (and possibly the other games) with barrels suspended above enemies, moving platforms, or even the player, which INSTANTLY begin falling downwards as soon as the screen finishes drawing, forcing you to start moving immediately or lose the level. I posted another GIF above to illustrate this because it's so buck wild.

There are no "clean" routes out of this barrel mess. If you stay in place, you get buried alive; your only choice is to restart the level. I should mention that this is one of the earliest games I've played with a "fail fast" feature, like one sees in e.g. Super Meat Boy: you can press F6 to restart the level at any time, it takes about a quarter second, action instantly starts without any kind of "get ready" delay, and you have infinite lives.

Because you are expected to lose many, many times before you win.

There's no way to avoid the barrels. If they were lethal, this level would be unbeatable, but as is, you simply can't move fast enough to get out of the way. The only way to avoid getting immured is to start wildly jumping and juking back and forth, trying to perturb them to one side or the other, either by pushing them in midair, or getting underneath one corner or the other so they get nudged aside by the weird, unpredictable physics. In the end, you will escape the avalanche, on every attempt, by wriggling your way out from between two blocks that you're embedded in.

The collision code in this game is, in a very literal and non-fooling sense, nonexistent. As in: there is no possible way that the programmers believed that their falling barrels had rigid and indisputable collision boxes, because if they did, this level would be unplayable. There's no way to escape the avalanche except to phase partially into the barrels, and to use your knowledge of the glitchy collision to Mario speedrunner zip your way out. This is literally "samus aran stuck in a wall" as a gameplay mechanic. Who would ask a player to do this???

Answer: Todd Replogle.

There are five episodes in Monuments, and the falling barrels are used in the first episode. Todd must have known they were weird and bullshitty before he made this level, but he not only used them anyway, he designed a level that only works because the collision is bad.

It is amazing to me that someone could just be okay with their game being so... unorthogonal. Like, imagine trying to write a manual page explaining how to do this. You couldn't! It's gameplay based on barely-deterministic behavior, that won't even be the same from one attempt to another. I doubt you could even really learn this repeatably - hell, it might even have an RNG element! And yet, they went ahead and published this game, a game that just hopes the player can figure this out. If they can't... they just don't get to finish it.

I'm not saying all games should be like this, and there are plenty of reasons not to make most games like this, but it's so, so, so cool that games like this got made once, games that struggle not just with phenomenally ahead-of-their-time physics puzzles, but basic player motion.

We don't often see indie games that are willing to attempt mechanics beyond the developers skills to implement. Generally, people stick to what they can do cleanly - we talk about how "messy" Celeste's billion-line player.cs file is, but the fact is that no matter how many corners the programmer cut, the result feels like Nintendo made it. They didn't just get their car across the finish line; the engine purred like a kitten and it steered like an F1 racer, because we know now that anyone can do that. We know too much.

Imagine deliberately making a game with motion worse than a Standard Reference Mario. Who would do it? Who could do it, knowing what we now know? How could you get into game dev in the modern world without, at some very early point, learning how to make a platformer that moves about as well as anything on the SNES? Christ, stuff like Unity (I'll say five hail marys) comes with this capability baked in. You'd have to work to make it worse.

Fun Fact! I tried exactly that. I have a partial reimplementation of Monuments of Mars Pharoah's Tomb that I started writing in Unity (I'll say ten hail marys) and it was amazing how much extra work I had to do to make it bad. Tomb wasn't nearly as janky as Monuments, but it's still pretty primitive. I tried using the 2D physics system, but it was too good! I had to manually override the motion code and make it only calculate in five-unit increments every X frames, otherwise the game was too easy. Reimplementing Monuments would require a titanic effort to implement all the bugs exactly in their original form, otherwise you'd have nothing at all.

The reason I'm obsessed with Monuments, and with all the stuff on the ZX Spectrum and PC88 and whatnot, is the same age-old hipster bullshit: authenticity. It's outsider art. It's someone who looked at Video Games, went "I'm going to make that," did their best, and then failed in a way that produced something that, while bad, was also genuinely new. There's nothing like Monuments of Mars, nothing feels like it, nothing plays like it. Any of us would struggle to make something this distinct because we have played too many games, we already know all the "answers" to questions like "how to make it fun" and "how to move the player around." This limits our imagination.

It feels, as with much outsider art, that the creator may have actually lacked the frame of reference to recognize where they had deviated from the Standard Genres - or, indeed, where they had simply done a bad job.

In my most generous mindset, I can only describe Monuments as buggy as hell. It is not just unique and special. It is badly made. The developer gave the player unlimited lives, a completely unheard of phenomenon in this era - Pharoah's Tomb and Arctic Adventure for instance have limited lives. He probably did this because he knew his game was completely unfair and incredibly hard to play, and he just didn't know how to fix it. You could easily take these exact same levels and gameplay rules, replace the player physics with those from a game like Mario or Celeste, and end up with a game that could be beaten on a three-life limit with two continues. The puzzles are very simple, it's just INFURIATINGLY hard to interact with them.

Nobody could plan this kind of jankitude. I'm sure that Todd sat down to write a platform engine, and when he stood up he was not satisfied with the results. He just didn't know how to make them better. Maybe he didn't know he could make them better. He didn't know where to begin and he had no proof it was possible.

And yet, Todd finished his game, he made dozens of levels, he handed the result to his employer, and it got mailed to thousands of people on floppy disks. This could only have happened because he never said to himself, "oh, you can't sell a game this shitty" - and I strongly suspect, having read PC game developer interviews from the era (including with Replogle himself) that he didn't say this because he simply did not know this "fact."

I cannot imagine a greater sense of bliss.