The Stack

My foray into obsolete hardware continues to provide puzzles and frustrations that would not be out of place in an old-school text adventure. I’m seriously considering adapting them to that format.

One bit of progress: I managed to burn a bootable Windows 98 CD. I probably burned several, actually. I wasted a number of CD-R’s, trying different software each time (including the built-in CD burner in MacOS X), but the machine I’m trying to install Windows 98 on didn’t recognize any of them as bootable. But with this last one, I thought to try booting it in my Windows 10 machine, and it worked there. This is most peculiar. The other machine is willing to boot other bootable CDs, such as my Windows XP install disc. The BIOS even displays “BOOTABLE CD DETECTED” in a text-graphics box during the startup sequence, so it’s easy to tell that it doesn’t consider my bootable CD-R to be bootable. Maybe it’s prejudiced against CD-Rs? Is that a thing that can happen?

Over on the other fork, I actually managed to procure a PS/2 keyboard, which allowed me to use the Windows 98 installer boot floppy. But this just led immediately to another blocker: the floppy runs DOS. To read from a CD-ROM drive, it needs a DOS CD-ROM driver, and, while it has several drivers on the floppy, it doesn’t have one that works with the one it has, or for any of the others that I tried swapping in, such as the one from the Windows XP box. I am once again impressed at how running DOS removes functionality that’s in this machine’s BIOS. It’s like an anti-operating system.

Now, it’s actually not hard to find DOS CD-ROM drivers online. There exist sites with incredible numbers of drivers from different manufacturers. But that just leads to the problem: How do I get them from the net to the install floppy? The obvious solution was to mount a floppy disk drive in the XP box, but by this point, between taking out the CD-ROM and installing a new CMOS battery, I seem to have rendered it unusable. For a while, it sometimes showed the POST screen when turned on, then didn’t do anything else. At this point, it isn’t even doing that. I can’t even access the BIOS. So much for having a working XP box.

I do still have one working machine that could be of use, though: the Windows 10 machine, my primary gaming device. Could I install a floppy drive in that? It looks like I can’t; the motherboard doesn’t have the connectors for it. But I could take things to a greater extreme. I know I can boot the Windows 98 install CD on it. What if I were to disconnect its hard drive, swap in the hard drive from the other machine, install Windows 98 there, and then swap it back? This might or might not work — there’s no telling what the Windows 98 installer would make of that hardware. And it has the additional risk that I might wind up permanently breaking the Windows 10 box as well.

Another possibility I’ve considered: Start with Windows 95. I have Windows 95 entirely on floppies. Once I have that installed, I can upgrade to 98 from CD. The 95 boot disk is kaput, though. I could presumably download a replacement boot disk, but then we have the “how do I get it onto a floppy” problem again.

Sometimes, you really have to regard retrogaming as a journey-not-the-destination thing. I don’t for a minute believe that the experience of finally playing Galaga: Destination Earth will justify the effort I’ve been putting into it. The only experience that can justify that effort is the experience of the effort itself.

When last we left off, I had more or less given up on running this game on my usual gaming machine, even in emulation. So this weekend, I dug some older hardware out of the closet. First up was my previous rig, in an ingeniously-designed compact case made by Shuttle. It turned out to be completely intact — the last time I upgraded, I upgraded everything. Once I hooked it up to a monitor and keyboard, it booted into Windows XP without problems — it grumbled about the CMOS, due to the battery being run down, but automatically figured out what hardware it had anyway. G:DE made no claim that it would work on XP, but I figured it was worth a try anyway, because at least it was a 32-bit OS and I had vague memories of its compatibility mode being more reliable. Well, no dice. It had exactly the same problems as under Windows 10. I contemplated downgrading the system to Windows 98, but gave up when it failed to recognize my Win98 install CD as bootable. Just as well. I can imagine a working XP machine being useful someday.

Going back another generation took a little more work. My pre-Shuttle mid-sized tower case was missing a graphics card — presumably because I had transplanted it into the Shuttle box when I first got it. But I found a suitable disused one in a box of loose cards. It’s very likely the one I had removed from this machine in the first place. Strange how upgrading graphics cards used to be such a routine part of gamer life, but at this point I haven’t bothered in years. Getting it in was a little awkward, due to the case coming from an era before people got case design really figured out. Oh, it was fairly innovative for its day — the motherboard is mounted on a section that slides out for easier access. But “easier” is relative, and the device’s innards are almost inevitably an intestinal tangle of cables, just because that’s how things were back then.

Once it was up and booting, the machine reminded me that it no longer considered its copy of Window XP to be valid and would not me log in. Which is fine, I suppose, seeing how I really intended to install Windows 98 anyway. But, as with the Shuttle box, it wouldn’t boot from the Win98 install CD. Was it even bootable at all? Perhaps not; apparently some Win98 install CDs are, and some aren’t. When I had been trying to get Windows 98 running under emulation, I downloaded a Win98 install CD that I know to be bootable, because I booted it in the emulator, but burning it to a disc failed to produce a bootable CD. Apparently Microsoft disabled the ability to burn bootable CDs back in Windows 7, probably to make it harder to pirate Windows.

But there was always an alternative to booting from the CD: booting from a floppy disk.

This machine actually still had a 3.5-inch floppy drive mounted in it, albeit not connected. After I connected it, I found that the machine seemed no longer capable of getting through its startup sequence. It would get to the point of displaying “Press DEL to configure, TAB to continue with POST”, but no keypresses would get it to do anything more. I almost called it quits right there, but after taking a break, I realized that the only plausible explanation for this change in behavior was that I had wiggled or jostled something in the case while plugging in the floppy cable. Giving all socketed items a thorough additional wiggle solved the problem.

I’m a little surprised that my collection of floppies have survived as well as they have, considering how long it’s been since I’ve used them. Every bootable disk I’ve tried has booted successfully, including the Windows 98 Startup disk. But this leads to an immediate additional roadblock. Every bootable floppy I own boots to some kind of command line or prompt that requires keyboard input to do anything. And, although the BIOS knows how to get input from a USB keyboard, these programs do not. I have a USB-to-PS/2 adapter. I have several, in fact. But it turns out that these adapters only work on USB keyboards that know how to use them. I’m fairly sure I had a PS/2 keyboard around not so many years ago, but got rid of it because it was taking up space and collecting dust and didn’t fit into a neat little box the way those graphics cards did. The lesson here is clearly to never throw away anything.

And there, for now, I stand. My options going forward include figuring out how to burn a bootable Windows 98 install CD and hoping that it’ll recognize the keyboard once it’s into the install process, or gaining access to a PS/2 keyboard for long enough to do the install. My options do not include, obviously, giving up.

For reasons I won’t describe here, the team I’m currently on at work recently declared a month-long internal Galaga competition, planned to be the first of a series of contests around different classic arcade games. Well, it’s not without precedent for managers to officially sanction non-work-related recreational gaming. I’m unlikely to win, but I’ve been playing a little every day, and have managed to reach scores that aren’t too entirely embarrassing. But more importantly, after a few days of this, I remembered: Wasn’t there a Galaga remake on the Stack? One of those classic arcade remakes from around 2000, with 3D models and power-ups added?

Indeed there was. Galaga: Destination Earth, a largely-forgotten title for Windows 95/98 and the original Playstation. I have the Windows version, which is unfortunate, because it doesn’t work any more. I vaguely recall that it had some problems back when I first played it, too — graphics glitches and whatnot — but on my current system, although the installer runs without problems, the game itself exits shortly after starting, or sometimes just hangs, without displaying anything on the screen in either case. And that’s a pretty hard problem to solve.

Playing with compatibility modes did nothing but sometimes make it display an error message: “The application was unable to start correctly”. Googling this, I found that it could be the result of a failure to load a DLL — but which DLL? I installed a program from Microsoft called “Process Monitor” to find out, only to learn that galaga.exe was not itself reporting any failures. It was apparently just deciding of its own accord to not run.

I tried looking online for help, but this is not a well-loved game, and therefore not a well-supported one. Hasbro Interactive’s tech support website doesn’t seem to exist any more. Pcgamingwiki.com, an inestimable source of game fixes, had nothing. One disreputable-looking patch site claimed to have a fix, although it wasn’t specific about what problems it fixed. Once downloaded, it was easy to identify as just a malware installer.

As of this writing, the most extreme measure I’ve tried is installing Windows 98 under an emulator to run it there. (I still have my old Win98 installer CD, and its sleeve with the license key on it!) This hasn’t worked any better so far, but there may be a better emulator out there. And if there isn’t, I can try to put together a real Windows 98 machine out of hoarded parts, like I’ve been planning ever since starting this blog. Or, alternately, I can buy a copy of the Playstation version on ebay for five bucks. But at this point, that would feel like giving up.

The galling part is that in the process of googling for help, I found some complaints that the game is too short — just a few hours long, apparently. I probably could have polished it off in 2001 if I had just played a little longer.

Sometimes this blog fulfills the opposite of its purpose. I made a three posts a couple of weeks ago about Gearheads, a game that I own on physical media and that therefore qualifies as a true element of the Stack, but I stopped playing it after those two posts, and it’s partly because I doubted I’d have anything more of interest to say about it. It’s cute, and it launched a couple of successful game design careers, but it’s not very deep strategically, and it has no plot. Its whole attitude is that of old coin-op arcade games: you can pick up what it’s about in a second, and that’s not conducive to lengthy analysis.

The controls, too, are arcade-oriented, or perhaps Atari-2600-oriented: it’s clearly designed for each player to have their own four-direction joystick with one button, and the fact that it plays from a keyboard instead can only be attributed to it having been released at an awkward time for PC joystick support. The vertical axis switches which lane you place your toys on — the movement of toys isn’t constrained to lanes, but their initial placement is, which can be awkward when you’re trying to place blockers. The horizontal axis is used to cycle through your toys. Searching through your toy collection this way takes valuable time, which motivates the player to stick with one sort of toy for a while before switching. Which is exactly how the AI plays in One Player Tournament mode, thank goodness. I imagine it would be very difficult to play against an opponent who switches tactics more frequently.

Now, in a normal One Player Tournament level, you get a random assortment of four toys to use. This means the time spent cycling through your collection is never too bad, even if every second counts. But levels 10, 11, 22, 23, and presumably 34 and 35 (which I haven’t reached yet) give you access to all the toys. And despite how good that sounds, it’s basically a bad thing, because it means you can spend a lot more time searching for the toy you want. Maybe the solution is to voluntarily limit yourself to a span of four consecutive ones. Would that work? I don’t know. I only just got through level 23 today, and not by doing that.

Mainly I feel like I pass levels by luck, and finally getting through the second twelvesome of levels was just a matter of playing until all the dice fell in my favor. That is, there definitely is some skill involved, consisting of the rapid application of learned responses to changing circumstances, but there’s a lot that goes on that’s chaotic and unpredictable and beyond your control. Except, that is, in those puzzle-like special levels where both sides are limited to one toy. Not coincidentally, these are definitely my favorite levels.

Level 24 was a particularly good one: it gives the player Krush Kringle and the opponent Orbit. Winning this match-up isn’t so much a matter of getting your guys across the screen as of deflecting the opponent’s toys back, but you have to get the timing and spacing of the Kringles just right to accomplish this. Once I finally reached this level, it took me two tries — and, since I can now start from level 25, I never have to do it again. In other games, I’d take the ability to skip solved levels for granted, but here, I’ve had to restart from level 13 so many times.

And to be clear, that’s a self-imposed restriction. The game lets you start from level 25 whenever you like. But what kind of completist would I be if I didn’t play through all the levels?

I feel like paying attention to special promotions on Steam these days loses you some cred. Steam had something special going on back when they weren’t the incumbent, but Itch is what’s hip now. Plus, Steam’s special promotions just aren’t that interesting any more. Back in, say, 2011, they had grand metagames, things for which they’d get developers to put new Achievements and even new secret levels into their games. Today, it’s all predictable annual sales and things to do with trading cards.

Nonetheless, this past Memorial Day weekend, there was a Steam promotion that I think bears some scrutiny. Billed as a “Spring Cleaning” event, it offered a trophy (which is to say, a badge, worth 500 Steam XP if fully leveled) for completing certain tasks. The interesting thing is that the tasks weren’t designed to convince you to buy more games. On the contrary: they were all about playing the games you already have, with tasks like “play a game that you’ve played for less than an hour” and “play a game you’ve played for more than two hours, but haven’t played in a while”. Each task, when clicked on, yielded a list of suggestions — one task, which could be claimed afresh on each day of the promotion, was simply “Here’s a few randomly-chosen games that you own. Play one of them.”

There was a task to play the very first game you ever acquired on Steam — in my case, this was the Orange Box, so I had quite a few choices. Another daily task asked you to play a game that’s in your library but that you haven’t played at all. For me, this was not a problem — I have many games I haven’t played yet; that is the entire premise of this blog. But I was curious to note that the list of games it recommended for this task included several that I had in fact played, and even ones that I had Achievements for. A bug triggered, perhaps, by having too many games? It tried to pull up my play history and gave up after the first hundred thousand lines? Who knows?

At any rate, the reason I’m describing the event here is the big question it provokes: Why? Why is Valve, as a corporate entity that’s not primarily concerned with encouraging people to finish their backlogs for its own sake, bothering with this nonsense? I guess they’re in favor of anything that keeps the players engaged. I also have a sneaking suspicion that it also serves analytic data-gathering. In the Bundle Age, it must be difficult to discern a person’s actual tastes, so here’s a random assortment of games that you already own; which will you click on?

But also, this is a remarkably backwards-facing promotion. It showed me a bunch of games that I haven’t thought about in years, and that made me think about how great they were back in the day. And that serves Valve well. Recall what I said about Steam being past their peak hipness. Well, if they can’t have hip, at least they’ve got a near-monopoly on several years worth of PC gaming nostalgia.

OK, let’s enumerate the toys. The manual contains a list giving their basic stats and special behaviors, so I’ll try to make observations not found there.

In approximate order of increasing interestingness:

Ziggy, the cockroach: the fastest and lightest of the toys, capable of crossing the screen with the least winding, provided it doesn’t run into anything. Which it likely will, because it tends to veer wildly left and right. If it gets bumped, it flips over on its back until it gets bumped again. Still, when it’s available, it can be the easiest way to sneak in a few extra points. The real fun comes in when the opponent is using Ziggy as well, because you tend to get large clusters of supine bugs that way, and it turns into a sort of tug-of-war (push-of-war?) with both players trying to push it past the goal line — toys don’t have to cross the line under their own power to score. Even toys that have completely wound down stay on the screen for several seconds, and score you points if they’re pushed through the goal.

Big Al, the bulldozer: The polar opposite of Ziggy. Powerful, heavy, slow, takes a whole lot of winding, moves completely straight. Once set on its path, it cannot be diverted. It can only be slowed down, preferably by another Big Al. This can produce more push-of-war situations, with both sides sinking their winding time into putting more pushing power into a single row of Big Als. Such things seldom go anywhere; eventually they just wind down and vanish. So the best way to deal with it is to cut out early and set off some fast stuff elsewhere.

Disasteroid, the gold mecha-anime-looking robot: Slow, heavy, moves straight, and has the ability to blast other toys directly in front of it, destroying them instantly. The only toy that’s immune is Big Al, which is the game’s most blatant and artificial rock-paper-scissors-ism. The blaster takes a good long while to recharge, though, so once you’ve sacrificed one toy to it, you’re safe. When two Disasteroids face each other, one of them will be destroyed — I think the one with more energy remaining wins, but I’m not sure of this.

Walking Timebomb, the walking timebomb: Moves straight ahead fairly quickly, and as such can be used as a point-scorer. More importantly, if it runs out of energy, it explodes, destroying everything within a radius except Disasteroids, which are therefore the best defense against them. This is the one toy you have a strong incentive to underwind, although figuring out exactly how much to wind it is tricky, so I haven’t used it much. Fortunately, the computer isn’t very good at using them either, and tends to throw them out in clusters that blow each other up before they reach your guys.

Zap-Bot, the wacky-looking robot with electrical plugs for hands: Another offensive unit, but in a different way. When it runs into a toy, it zaps it, draining its energy and slowing it while it zaps. What’s more, it moves diagonally, allowing it to sneak up on things like Disasteroid that focus on what’s directly in front of them. Its big weaknesses are that it’s hard to aim and that it’s undiscriminating about what it zaps — if you release two Zap-Bots from the same spot, and the one in front pauses to zap something, the one in back can run into the one in front and start zapping it. So the main use I’ve gotten out of them is just taking advantage of the diagonal movement to wiggle them past blockers for points.

Deadhead, the skull: Very slow-moving, and requires maximal winding to get it across the screen under its own power, but that doesn’t matter, because its purpose is primarily defensive. When any toy bumps into it, it pauses to scream, scaring the toy into reversing direction. Note that “any toy” includes toys moving in the same direction that bump it from behind, so Deadhead imposes some pretty severe limits on the person who launches it. They’ll even scare other Deadheads, and can get tangled together in a perpetual scream. They wander a little, so it’s important to keep them separated. Probably the best way to use them is to wind them just enough to scare one thing before they wind down, although this can be difficult to control. Still, turning the opponent’s toys around is a very powerful move, especially for fast-moving types that can recross the screen and score you a point before the opponent reacts.

Krush Kringle, the Christmas-themed professional wrestler and single weirdest toy idea in the game: Moves slowly, periodically thumping the ground to make toys within a certain radius reverse direction. Two Krushes side by side will repeatedly reverse each other, forming a sort of vortex that other slow-moving toys can’t escape. There’s a timing element here that’s difficult to use effectively, but also difficult to combat — with the right timing, you can get a Disasteroid in to kill it between thumps, but you risk giving your opponent a free Disasteroid that way.

Orbit, the flying saucer: Just as lightweight as Ziggy, but a little slower and needs a little more winding. Orbit moves straight forward until it hits an obstacle, at which point it tries to navigate around it. It’s the only toy that knows how to do that. Effectively unblockable, the only good defenses against it are Disasteroid and Deadhead.

Presto, the magician: Moves forward at a moderate speed, but periodically teleports to a different lane, apparently at random. Can get stuck for a while between teleports, but is ultimately not very blockable. Not the fastest way to score points, but sometimes the only real option you have.

Kangaruffian, the boxing kangaroo: Moves diagonally, like Zap-Bot. When it hits something, it punches it. If it’s something light, like a Ziggy, it can send it flying backward with enough speed to get it past the finish line and score you points. But it’ll add some momentum even to the heavy things, and can be used to break a Big Al stalemate. I haven’t been using them much, and have only recently started to appreciate how useful they are.

Clucketta, the hen: Basically a toy factory. Flies forward in bursts, passing over other toys, then settles down for a while to lay an egg, then repeats. The egg hatches into a Small Fry, a little chick that rushes forward. The really impressive thing about Clucketta is the way it can come to dominate the board. Because they don’t move forward very often, you can wind up with a whole bunch of them together for a long time, clogging up the board and getting in the way of everything else. I suppose this is what the Walking Timebombs are for.

Handy, the glove: Rushes forward straight, faster than anything else except Ziggy. When it runs into any toy other than another Handy, it attempts to lock onto it and start winding it. This can bring depleted toys back to life, or give extra power to insufficiently-wound ones — if you summon a blocker in a hurry, you can put a Handy behind it to extend its life. The big weakness is that it will wind your opponent’s toys too. If it weren’t for that, it would make an ideal point-scorer, moving both fast and straight, but you don’t dare put it own in an empty lane where the opponent can co-opt it. Its best use, then, is to follow behind the similarly straight-moving Disasteroids and Big Als, like a Heavy/Medic pair. And once you have that, why not launch some more Handys in their wake, letting the guy in front keep the lane safe for point-scoring?

I’ve played Gearheads a little more, but I can’t honestly say I’ve made any progress in the “One-Player Tournament” mode. This isn’t the sort of game that saves your progress. It’s the sort of game that keeps a high score list. It gives you a limited set of lives for each session, and expects you to start over whenever you run out — a play pattern already antiquated in 1996, when even the fading coin-op games let you buy your way past death. You don’t have to start from the very beginning of the sequence, mind you. You can start at level 1, 13, or 25. (This is part of why I think there are 36 levels, something not actually stated in the docs.) I’ve been starting at level 13, and haven’t yet made it to 25 from there. I suppose it might be worth it to start at level 1 for the sake of accumulating extra lives if you’re not just doing a practice run, but at the moment, practice runs is all I do.

Every third level is a special one, where you can use only one type of toy. (This is the other part of why I think there are 36 levels: that’s exactly enough to have one special level for each of the game’s 12 toys.) The opponent also has only one toy type, which might be the same as yours or might be a different one, depending on the level. These special levels are a little puzzle-like — there’s always some specific tactic that will let you pull ahead, but depending on the toys, it might require confrontation or avoiding confrontation, winding up your toys fully or only enough to get them across the screen. Still, there are only so many possibilities to try out, and once you’ve found something that works, you can just remember it for the next time you play that level. Generally speaking, the special levels are a relief.

The regular levels are harder. They give you a level-specific set of four toys, but the opponent’s toys seem to be randomized. You can’t memorize a per-level successful strategy when you’re facing a different enemy every time you play the level; you have to learn to be reactive, to use the tools available to combat whatever happens to come up. Usually the opponent releases a bunch of toys of the same type in sequence — I could probably find patterns by counting them, if I need to go that deep. So there are opportunities to see what the opponent is doing and counter it — sometimes launching one toy on your side can counter a whole bunch of the opponent’s. But on the other hand, sometimes the opponent has a toy that I just plain don’t know how to counter with the ones I have, and I just have to hope that the random number generator will be kinder next time. I’ll go into more detail in my next post.

At a recent board game night, I had a chance to try a game called Quantum that used dice to represent spaceships with different capabilities. The box credited its creation to Eric Zimmerman. “That’s a familiar name”, said I, and looking it up online afterwards, I found that, sure enough, it was the same Eric Zimmerman that co-founded Gamelab, the company that developed Diner Dash, among other things. This is a person who is partially responsible for creating a genre. But also listed in his ludography was something I didn’t expect to see: Gearheads, a 1996 game about wind-up toy battles, co-created with Frank Lantz of Universal Paperclips fame. Apparently it was Zimmerman’s first published game. And it just happens to be on the Stack.

So, obviously I had to dig out the CD and give it a play. Windows gave me some guff about that, complaining simply “This app can’t run on your pc” when I tried to run either the executable or its installer, even in Windows 95 Compatibility Mode. This was a new one on me, but apparently it’s how 64-bit Windows 10 reacts to 16-bit Windows programs. Apparently there are ways to enable 16-bit support in Windows 10, but I opted to play it safe and instead run it under the copy of Windows 3.1 that I had installed in DOSBox back in 2010, which I still have around thanks to file-sync apps. This worked with no problems.

The game is essentially two-player, with both players using different parts of the same keyboard simultaneously, but the computer can fill in for one player, and it has a “One Player Tournament” mode, a sequence of 36 increasingly-difficult levels that I’m taking as the basis for completion. Gameplay consists of letting loose wind-up toys on one end of the playfield in an attempt to get them across to your opponent’s end, while your opponent does the same to you. You get to choose where to set each toy down and how much to wind it up, but you don’t control them after they’re placed. Each level gives you access to a different subset of 12 toy types, each with its own virtues and special powers. For example, there’s a wind-up cockroach, which moves very quickly but erratically, and tends to get flipped over on its back; a bulldozer, slow-moving but capable of easily pushing other toys backward; a chattering skull that scares other toys and makes them reverse direction. There’s an element of extended rock-paper-scissors to it, but also some opportunity for combos, like using boxing kangaroos to punch depleted cockroaches over the finish line.

I hadn’t thought about this before, but it’s a lot like Magic: the Gathering – Battlegrounds. Both games are all about summoning creatures that automatically march across the screen to score points and/or block your opponent’s creatures from doing likewise. And a lot of the same tactical considerations apply to both, like choosing whether to try for a mainly defensive summon to keep the opponent away or just try to outscore them with a horde of small quick things. I think the gameplay is more chaotic here, though. Nothing is entirely predictable, and there’s a lot of it going on at once.

Really, Tower of the Sorcerer spawned a mini-genre. In addition to DROD RPG and Desktop Dungeons, there’s DungeonUp, which I haven’t mentioned before — I got it from some bundle a while back and played it blind, and found it a delightful little variation on the now-familiar theme.

I imagine there are other examples of TotS-like out there that I haven’t discovered, and if anyone reading this knows of any, I’d like to hear about them. The defining characteristics of the mini-genre are, to my mind, passive monsters that stay still, possibly blocking passageways, until killed or otherwise acted on, and deterministic combat based on the formula “damage = attack – defense” or something similarly simple.

Beyond that, there are some notable similarities between TotS, DROD RPG, and DungeonUp that Desktop Dungeons is notable for rejecting. All three of what we might call the synoptic TotS-likes feature: Unlimited hit points, with healing potions simply adding to your current total like in Ultima 1; machines that let you purchase upgrades to attack, defense, or health for gold; multiple dungeon levels, with the benefits of health potions and upgrade machines increasing by level; locked doors in multiple colors, with corresponding keys. There’s no notion of experience or character levels, since their purpose is absorbed by health potions and upgrade machines. DROD RPG adds mechanics derived from DROD, including diagonal movement and facing rules. DungeonUp adds randomized dungeon layouts and adventure-gamish “Aha!” puzzles. But there’s so much shared foundation here that the games have fundamentally the same feel and tactics.

Desktop Dungeons, meanwhile, takes just the barest basics of TotS and runs off in its own direction with them. Some of what it does is adding back familiar RPG-isms like character classes and experience levels, but it doesn’t do this in a cowardly clinging to the familiar. It does it because of what they can add to the puzzle. For example, as in a lot of CRPGs, leveling up instantly restores your health and mana to maximum. This can be exploited! One of the game’s most basic tricks is hitting a tough monster a few times, then slaughtering something weaker to level up and get your health back, then resuming your previous fight. You can’t do that in the other TotS-likes, not just because they have no notion of leveling up, but because they don’t let you break away from combat. Once you start hitting something, you just keep on hitting it until one of you dies. There’s no reason for them to let you break away; the rules of those games provide no benefit for killing something halfway. DD provides so many reasons to do it, from healing to renewing your buffs to “I don’t actually want to kill it yet, I just attacked it because my weapon has a knockback effect that pushes it into a wall, and destroying walls pleases my god, and that gives me just enough piety for this boon I’ve been after”.

I’ve seen the use of cliché and even stereotype defended on the basis of efficiency. The idea is that it’s an expositional shortcut, a way of exploiting shared culture to lessen the heavy lifting required of both the author trying to convey ideas and the audience trying to understand them. Genre in games fills a similar function where it’s even more needed. When we choose to play games with wizards and dragons in them, it’s not typically because we’re in love with the idea of wizards and dragons. It’s because wizards and dragons don’t require a lot of explanation. Even when the familiar elements deviate from expectation, the very fact that there is an expectation helps us to grasp that deviation.

Desktop Dungeons exploits this a lot, thank goodness. It’s got so many unfamiliar mechanics that we really need familiar pigeonholes for them. I think the most intriguing example is its treatment of race. You’ve got the standard assortment of dwarves and elves and so forth, albeit with their places in society all mixed up and humorized: elves live in the slums and and are regarded as disreputable, orcs have opulent mansions and talk posh, dwarves have frat houses. It’s in the dungeons, though, that it gets interesting.

In the dungeons, your inventory space is limited, and you can’t drop items or sell them to shops. If you need to free up some room to pick up something new, you have to destroy something — or rather, “convert” it. Converting items adds points to a pool, and every time that pool fills up, you get a boost of some sort. Exactly what that boost is depends on your race. This is the only difference between the player races. If you want racial bonuses, you have to earn them by trashing stuff. It’s weird and it’s subtle and, just like everything else in the game, you have to learn how to use it effectively if you want to beat the Hard dungeons.

Races are unlocked one by one as they join your kingdom in gratitude for rescuing them from dungeons. At the start of the game, all you have is humans. The human conversion reward is a permanent increase to your attack bonus. Then you find elves, which get an increase to their maximum mana as their reward, and dwarves, which get increased health. Attack power, mana, health: these are the three primary stats your character has. They’re the three things that you can find stat boosters for scattered in the dungeon. And they’re assigned to a fairly archetypal set of races — basically, normal, gracile, and robust people. So once you’ve gotten over the weirdness of the conversion system, this arrangement feels fairly elegant, natural, and even necessary.

And then you find the halflings.

The conversion reward for halflings is a healing potion. You’ve become used to the idea of sacrificing objects for enhancements to your intangible characteristics, but now you’re turning objects into other objects. Next come the gnomes, which get mana potions, and at this point maybe it starts to seem systematic again. Health and mana potions, like stat booster objects, are scattered loose in every dungeon, in consistent quantities. It’s just giving you one race for each thing you can consistently find.

But then you get the monster races. Orcs get additional base damage as their reward. How is this different from humans? That’s a little technical. Your base damage is, by default, five times your experience level; your attack bonus is a percentage increase on top of that. So if you have items or spells greatly increase your attack bonus, you can get more out of it by increasing your base attack. Finally, after that, you get goblins, which get experience points from conversion, enabling them to level up quickly without fighting anything. And with that, any sense of pattern is broken. Conversion rewards can be anything the designer thinks up. The whole sequence, from humans to goblins, is like a little story about a weird system that becomes weirder every time you get used to it.