Hey friends! First of all, thank you for checking out this little project. It was an interesting experiment, and I think it mostly achieved its design goals. I'm not totally satisfied, but I think this project did about as well as you could expect, given the bizarre and alienating premise.

I promised that I'd write a blog post about the story behind this game. This is that post. It'll also double as a bit of a post-mortem on the project - looking at the design goals and whether it achieved them. Note that the post assumes you've played the game, but someone was so kind as to upload a playthrough to YouTube.

1) Postmortem

Actually, let's start with the post-mortem thing. What is this project, why is it the way it is, and how did it go? This section will look at the game as an object that exists. Section 2 will look at the weird esoteric sci-fi bullshit that's happening, and what it was like to use AGS 2.72 again.

1.1) Background

Earlier this year, I decided that I wanted to make a lostwave game. That is to say, a game that pretends to be recovered lost media. That's kind of its own sub-genre by now, and famous examples include The Eternal Castle Remastered, Basilisk 2000, Shipwrecked 64 and Catastrophe Crow. It's not quite the same as "retro" games, because "lostwave" media comes with a sort of kayfabe. Or meta-narrative, if you prefer. Horror is very common here, these games are usually lost because they're haunted by a ghoast.

These games often do a good job capturing the aesthetics of their era, but that's usually where it ends. They're almost always made in modern engines, on modern platforms, and for modern audiences. They also tend to have features that wouldn't have been possible on old hardware. Of course, people do actually make real new-old games, which genuinely could run on old hardware. (You can even find native DOS games on Itch.io right now.) But that's the exception, and most of this is focused on early '90s tech at the moment.

So, I figured... why not make something like that, but for the mid-2000s? That was an awkward transitional period in personal computing, nestled in between the dot-com crash and the iPhone. We didn't really know what the future would look like, and so that era is full of weird and fascinating dead-end technologies. I decided to use an engine from this time, Adventure Game Studio 2.72, to make the thing. Why AGS 2.72 in particular? Convenience, mostly. (See section 2.1 for the full thought process.)

Also, I wanted to do something on the futur-null account again (rather than kdrgn), and continuing Laura's journey seemed like a natural fit. Cascade Failure is part of a little series of games. Together, they tell the story of Laura and Yaric and the Bay Empire. Icebound Machines was released 1st and is chronologically 3rd, Parsinomady was released 2nd and is chronologically 1st, and Cascade Failure fills in the gap between those two. Being the middle part of a trilogy fits well with the "fake lost media" thing, right? The writing is meant to be somewhat dense and cryptic, but still comprehensible, which is an interesting balancing act.

Once I formulated that idea, it was time to look for an opportunity to actually make the thing. Laura's stories lend themselves to horror themes, and I haven't really done much horror in the past. (I tend towards the wholesome stuff.) So, I looked through some spooky-themed game jams, and I ended up picking two: SCREAM JAM 2025 and Themed Horror Game Jam 22. This would give me roughly a week to develop a game with some thematic requirements - restrictions breed creativity, they say. One of the themes in the Themed Jam was "Heaven's Wrath," which nicely dovetailed with another recent obsession of mine.

I've been thinking about Uriel's Chasm lately. Uriel's Chasm is an infamously bad game, so much so that it was once the third-lowest title on Steam. And this was back in 2014, when a bad game being on Stream actually meant something. It's bizarre and inaccessible, with badly clashing visual styles and a completely incomprehensible plot - and a fake "lost media" angle, you'll notice. Nobody is quite sure if Uriel's Chasm is a joke, an interactive art project, or something else entirely. It's outsider art. I won't pretend that it's some diamond in the rough, but you can find stuff in Uriel's Chasm that a "good" project could never pull off.

Likewise for Southland Tales (2006), another infamously bad work that continues to fascinate me. It was a real trainwreck of a movie. Southland Tales is at once very 2006 (being about the George W. Bush dictatorship) and very 2025 (being about surveillance capitalism). And some of it just doesn't make sense outside of the director's brain.

Cascade Failure is rather more comprehensible than those two, even when it's being cheeky and high-concept. But I wanted to capture some of that vibe. Uriel's Chasm refuses to explain anything, and Southland Tales just fumbles it. Everything in these two works just has this aura of menace and incomprehension, and that's actually very cool.

Put all of these considerations together, and you get the ideas that went into Cascade Failure.

1.2) What worked

As described above, this game ties in to both Icebound Machines and Parsinomady. However, I also wanted it to work as a standalone title, as I can't really demand people play two other games first. I think it's actually fine if the experience is somewhat alienating for a first-time player, as that works with the "lost media" vibe. (After all, you're getting a disconnected piece of a larger story.) But it still needed to be comprehensible, and that's a difficult line to tread.

I think I got this one mostly right. The weirder aspects of the setting are mentioned at least in passing, and before they become relevant, though in some cases you have to go looking for them. The "metadata" is in there, so to speak. I think I got this one mostly right. The game was comprehensible (though cryptic) to people who didn't play the other two games. And at the same time, some fans of Icebound Machines or Parsinomady sent me very kind messages after playing Cascade Failure, expressing appreciation for the callbacks. (Ending up in the final room of Icebound Machines again was a hit.) I'm quite happy with that. People came away liking Laura, or at least feeling a degree of empathy.

More generally: There's about the right amount of content in this one. A week really isn't a lot of time, but I managed to cram just enough story in there. It's long enough that the setup-and-payoff structure (like the gradual decay of the world) can work, but not so long that it becomes tiresome. The extremely low resolution forced me to be somewhat disciplined with the length of the individual messages, which was probably a good thing. I am quite prone to going wordswordswords.

Also, I really like the way the art turned out. The sprites are relatively big, given how low the resolution is, and I think that style works well. (The native resolution is 320x240 pixels, but the GUI takes up the bottom 40 pixels.) I made some use of free assets, but I drew most of it myself, including the backgrounds. (The weird skies were created by taking photos and messing with the colour levels.) Given that I'm not much of an artist, I'm gonna say I did alright.

The 16-bit colour depth, and the way that AGS 2.72 handles upscaling, creates odd visual noise in places. I intentionally provoked this effect, making heavy use of smooth gradients to create color banding. But then I used dithering for the space behind Hero, so that one doesn't have color banding. It's subtle and I'm not sure anyone took notice, but it's a small touch that I'm very happy with.

Another small touch that I like: The speech portrait backgrounds. They mean something! Laura and Jordan both share a style, that being a grayish-blueish-purpleish pattern, a little reminiscent of waves. Barry shows a similar pattern, but vaguer and in a different colour. The falsepeople have a much simpler version of it, just blocks and splotches. The divine voice starts with the pattern, but it degrades and becomes hostile as the game goes on. Hero uses the same colours, but the pattern is orderly and resembles circuit lines. And finally, Yaric doesn't just share Laura's style, but the same pattern exactly. See, there's meaning in this.

One more thing about the visual style. A common problem in adventure games is that it often isn't possible to tell background decor and interactive elements apart. In my Godot games, I've implemented a thing that lets you press a button and highlight all the interactive hotspots, but that isn't easily possible in AGS 2.72. (You could do it via the RawDraw functions, but this would've required extensive scripting work.) So, I did a visual thing instead. Here, this needs a screenshot.

This is the apartment. The objects you can interact with have thick black outlines, while the decor is faded. I think I will actually continue to use this technique in the future, though perhaps the differences are a bit too drastic here. (Also, the room itself should have faded rather than black lines.) Worth considering, anyway.

Finally, the music by human gazpacho is great. I didn't make it, I just put it into the thing, so I want to shout it out here. Go ahead and click that link, please. I do think the sound design worked out pretty well in general, I'm gonna give myself credit on that. I decided against footstep sounds this time, AGS 2.72 ties these to animation frames in a way that would've been annoying to set up properly.

Alright! Let's move on to the criticism, before I get too conceited.

1.3) What didn't

This project was built on top of Michael Sheail's OpenQuest. I think that was a mistake. Don't get me wrong - OpenQuest is made well, and it contains a lot of useful example scripts. OpenQuest is actually from 2007, and at the time, it was probably the best "tutorial project" available. (Demo Quest was never fully updated to AGS 2.72.) But it's also set up in a fairly peculiar way, and the inner workings are different from typical Adventure Game Studio titles. For example, it doesn't actually use the built-in "mouse mode" functionality, it fully re-invents the wheel and that's confusing.

Also, AGS 2.72 encourages certain anti-patterns in design. I remembered that going in, but I still fell for some of them. (I'll have more to say about that in section 2.) As a result, the UX is more cumbersome than it needed to be, and it's not cumbersome in the right way. If I was to do the project all over again, I think I would've just used the default game template and implemented everything from scratch. This would've probably cost me most of a day, but I also finished a day before SCREAM JAM 2025 ended, so it would've balanced out.

Moving beyond engine issues... the various scenes ended up being a little samey. I would've liked to devote more time to them, maybe implement some multi-step puzzles while I'm at it. Or maybe it would've been better to commit harder to a single scene. Just have one big building with all three ritual locations inside. Short jams like SCREAM JAM 2025 force you to think small, and that's something I tend to struggle with. Next time!

Horror elements could've been stronger. Not really much else to be said about that. There are a few scares in this one that I like, but not enough. I tend to pull my punches, and that's something I really need to work on. Next project, I should try to just go all in on something.

Finally, I think the whole "lost media" conceit ended up a bit... undercooked, I guess. As in, I could've probably done more with that. I think there needed to be more of a "meta-narrative" in place. Some supplemental material, a scan of a badly photocopied manual, maybe some design documents, that sort of thing. I think the execution here is okay, but no more than okay.

At first, my intention was to just provide the executable and like a text file with instructions. Actually running it would have been the first puzzle. But it's already hard enough to get people to look at these projects, and so I ended up moving away from this. I think it was a good call, but it does unfortunately weaken the lostwave vibes. This could've probably been fixed with a more elaborate "meta-narrative," I think. You really are playing an antique executable in an emulator, but I didn't really put enough effort into selling that.

Something to note for the future, I suppose! Despite these points, I'm fairly happen with how the game turned out.

1.4) Insights

All in all... this was a fun experiment! I don't think I'll go back to making stuff in AGS 2.72, though. There's still a lot to be said for it, but I missed the power and flexibility of Godot. Some effects that are fairly trivial to do in my own engine (like Laura's portrait changing over time) required a surprising amount of effort to implement. Branching dialogue trees in particular are really painful to do in AGS 2.72.

If you want to use an all-in-one engine to make adventure games, you should probably just use the current version of AGS. At the time of writing, this is AGS 3.6.2, which is a perfectly modern engine below the hood. If you're reading this in the future, there's a good chance that AGS 4.0 is out, which will be a full relaunch. Definitely check it out when that happens.

As I've mentioned above, this was a combined entry to SCREAM JAM 2025 and Themed Horror Game Jam 22. The game didn't exactly set the world on fire, though its performance was perfectly solid. #22/954 in Story and #174/954 overall in SCREAM JAM 2025, that's more than respectable. It didn't do quite as well in Themed Horror Game Jam #22, but scoring #7/22 on Gameplay and Creativity and #8/22 on Story is still decent. Keep in mind that I only used about a fifth of the available time. Adjusting for that, I can't complain.

I wonder how I should approach game jams in the future. I found SCREAM JAM 2025 to be a little overwhelming, both because it was short and because it had such an overwhelming amount of participation. But spending the full 5 weeks on Themed Horror Game Jam 22 would've been too much. I've found that Strawberry Jam (see the kdrgn account) is sort of the ideal spot, you get a month and there's 50-80 entries. Maybe that's how I should do it, look actively for those.

Initially, I really wanted to do the $109 Adventure Game Challenge. However, it turns out they introduced a pro-"AI" policy in 2023, and their banner is visibly "AI" slop. So I decided not to bother.

That's about it for the post-mortem, I think. This was intended to be a little snack of a horror game, and I think it achieved that goal, despite the flaws. Trust the process, I suppose.

2) The Penultimate Truth

Alright, here we go, the bonus essay. This section is going to talk about the more esoteric stuff. Why did I pick AGS 2.72? What was that experience like? And is there any Thematic Resonance™ with the game? This section is going to be a lot more stream-of-consciousness, fair warning.

2.1) Confessions of a Crap Artist

Once I made the decision to create a "lostwave" project (see section 1.1), the next step was to pick a dev environment. I decided early on that I wanted it to be a real outdated game, not a fake outdated game. That is to say, I didn't want to use a modern engine to make an old-looking game. (That's just the stuff I normally do.) I also wasn't going to write something from scratch in a week, so I had to choose an engine.

A week isn't enough time to learn a new environment and make a project. Of course, I could have spent a few weeks before the jam - set aside some of September to learn the rudiments of ACK or LoveDOS. I do enjoy learning new skills. But... I have a very limited amount of time and energy to put into these projects. I don't want to expend that on picking up a skillset that I'm going to use exactly once.

This is what ultimately led me to Adventure Game Studio. Or led me back to it, I should say. I used to make terrible "student projects" in AGS! They're not really worth playing today, but I did learn how to use the engine. When I poked at it again, I was pleasantly surprised to find that I actually remembered how it worked, at least in broad strokes. So I set up a Windows XP VM and grabbed a copy of AGS 2.72, and that was that. (I used shared folders so I could set up source control without having to connect the VM to the internet.)

The interaction editor is fairly intuitive, once you understand what it wants from you. You can even use AGS 2.72 as a low-code development platform, though people generally just did everything via the run-script action. (AGS has a C-like scripting language.) There's a surprisingly in-depth help file, which doubles as a language reference. There's a tutorial and everything! That sort of documentation is rare today.

Unfortunately, the built-in editor lacks even basic comfort features. It has syntax highlighting and autocomplete, and... well, it has those two things. That's it. There are bizarre restrictions in place, like not being able to edit more than one script at a time. There's no way in the editor to reorder or renumber your rooms, animations, or dialogue files, and they all just go into a big pile together. I sure hope you laid them out correctly the first time. The project setup encourages certain antipatterns, like god objects, everything-as-globals, and magic numbers).

Adventure Game Studio also does some things that are outright silly. Like refusing to start on December 25th, or having a joke option in the toolbar called "make my game," which calls you an idiot when you click it. It's not at all a professional environment, but it doesn't try to be.

AGS 2.72 is still a powerful engine, mind. You can use it to make excellent games, such as A Tale of Two Kingdoms (2007). Non-adventure games are possible too, with Trilby: The Art of Theft (2007) being a famous example.

In short: It's easy to make a small adventure game in AGS 2.72, and pretty hard (though possible) to do anything else. But! I actually was going to make a small adventure game. So that all worked out!

Here's something that's striking about AGS 2.72. It feels ancient today - which makes sense. But it already felt ancient when it was new. AGS 2.72 was already a bizarre retro throwback in 2006.

2.2) Counter-Clock World

To explain what I mean by this, I want to briefly look at the history of the engine. Adventure Creator, the direct precursor to AGS, was inspired by the 1991 adventure game Space Quest IV. Solo dev Chris Jones began work on the engine in 1995, and made it available to the public in 1997. The first real AC game, Larry Vales: Traffic Division, came out in 2000. (Earlier titles were little more than tech demos.)

Adventure Creator was strictly DOS-based. The first Windows version was Adventure Game Studio 2.4, which finished development in July 2002, a year after Windows XP had brought an end to the DOS era. The final release in this branch came in 2006, with Adventure Game Studio 2.72. (That's why I picked that specific version. If Cascade Failure really was lost media from the mid-2000s, it would've been made in AGS 2.72.) Two years later, in 2008, AGS 3.0 came out. Chris Jones rewrote the editor from scratch for the second time, and it was now using the .NET Framework, to make it ready for Windows Vista. That's still the current branch of AGS.

Look at those years again. Like, think about that for a moment. When Adventure Game Studio 2.72 was brand new, Space Quest IV was fifteen years old. And this was an era of rapid technological change! The 2000s gave us both an MS-DOS based operating system and the iPhone. There was an honest-to-goodness paradigm shift in the middle of that decade.

The point-and-click adventure genre peaked in the late 1980s and early 1990s, then entered a long and slow decline. Adventures were fine, they just started to move away from the old point-and-click model. The Secret of Monkey Island (1990) and Day of the Tentacle (1993) saw modest commercial success, but they couldn't even begin to measure up to Myst (1993). That was the earliest hint that the point-and-click subgenre was in trouble, I think. Big adventure games continued to come out, like Simon the Sorcerer II (1995), but the lackluster sales figures of Grim Fandango (1997) and Gabriel Knight 3 (1999) suggested that the end was drawing near. Adventure games were pronounced dead in 2000. I think this was a little premature, as The Longest Journey (2000) was a "post-death" title that saw great commercial success.

That being said, I'd put the actual end in 2004, with the cancellation of Sam & Max: Freelance Police. So not much later. People kept making point-and-clicks, but not the triple-A sector; and the adventures that were still being made weren't point-and-clicks.

And this all happened long before Adventure Game Studio 2.72 even came out.

Here's what I'm trying to say. This is simply not what games looked like in 2006.

This is what they looked like.

In short, AGS 2.72 was a living fossil.

2.3) Martian Time-Slip

So: Cascade Failure was made in a game engine from 2006, making it a weird retro throwback. But that game engine was not at all modern by 2006 standards, in part because it was itself a port of a DOS thing from 1997. And that DOS thing was again a bizarre retro throwback when it was new, since it was originally designed around a game from 1991. The point-and-click adventure genre was already seen as moribund in the days of AC 1.0, and stone dead when AGS 2.72 came out. Using AGS 2.72 really is like peering into a portal to the distant past.

... and this is where we wrap around to the Thematic Resonance™ of it all. This goes beyond Cascade Failure itself.

You'll notice that Laura is dilligent about logging her observations. She's structured, she's organised, and she pays close attention to detail. But she's approaching a problem which is a poor fit for that sort of personality. Her efforts do succeed, at least partially, but she's doing the equivalent of hammering in a nail with a glass bottle. Her prediction models work, but her methods and her results are bizarre. They just don't add up in any conventional sense. Laura's three core assumptions - that the future is predictable, that the future is mutable, and that small causes can produce large effects - are sensible and of practical value. But it's unclear if they're true, and why they're being interfered with.

Laura surely must be aware that she's caught in the guts of a dying machine. That reality is breaking down around her. We did see in Parsinomady that she's prone to mentally blocking out things that she isn't emotionally ready to deal with. Part of the fun of horror lies in not answering questions like that, so I'm afraid that this vague speculation is all you're getting. Sorry. ♡

Either way: The date jumps around, and all the characters seem to be at least slightly confused about the order of events. This is in part because there's a causality loop in place. Laura has prophetic nightmares of a cosmic crime in the future; because of this, she develops her abilities to better understand those nightmares; because of this, she ends up on the path that sees her violating the cosmic order, leading to the event that she had nightmares about.

In fact, everything in the Bay Empire seems to be in a loop. Souls are recycled. Everything important is based on recovered "Old Earth" technology - and it's never clear what this actually means. The setting is repeatedly implied to be post-apocalyptic, but not within the borders of the Bay Empire. The year is totally unclear but it seems to be in the past. There's something deeply wrong with Old Earth technology, but it's not like there are any alternatives.

Cascade Failure contains the pay-off for certain mysteries set up all the way back in Icebound Machines and Parsinomady...

2.4) Flow My Tears, The Policeman Said

... which is that the Bay Empire runs on nostalgia and hypocrisy. It's in decline, but slowly, and it never quite collapses. There don't seem to be any real challengers to it. And if the world really is post-apocalyptic, as it seems at times, then we have to note the fact that the Bay Empire is still running at all. Even if it increasingly has to use violent means to sustain itself.

You also can't just flag it as an evil empire and call it a day, because there's some complexity to it. The haze of nostalgia actually works in your favour a few times. Normality covers the Bay like a blanket of fog, even as the state engages in brutal crackdowns. The military police in particular is quite deluded about its own role, and holds itself to bizarrely high standards. The law has an unexpected amount of sway over the cop-adjacent functions of the state. In Cascade Failure itself, we are told that a lot of businesses and services remain open, even as gaping wounds tear themselves open in reality. Everyone is very nonchalant about the breakdown of time and space.

Hero is the closest thing the series has to reliable narrator. Laura is either losing her grip on reality, or she's perceiving things too sharply, but it's an unhealthly dynamic either way. But Hero has an outside perspective on at least some of these loops. Hero claims that Laura has been working on the cascade failure for 163077 hours. I don't think anyone did the math on this, so I'll spell it out: That's not a random number. If you take the date when the game starts (March 11th 2007 at 05:00 AM) and you add 163077 hours to it, you get October 17th 2025 - the date when Cascade Failure was released. To be clear, the plot twist here isn't that these two(?) people are diegetically in a videogame. The larger point is that Hero, alone, is aware of how badly time has slipped out of joint. (Also, if you treat 163077 as a base-8 number, it's prime. That's just a little bonus.)

Anyway. Hero is, ultimately, still a Bay citizen - and still makes use of Old Earth technology. So even their perspective is limited, and they remain tied to the loops. Hero sees a lot more than they're supposed to, but they can't quite put the pieces together. Hero and Laura have found ways to press on reality, to move it in certain ways. Both(?) have some grasp on the how, but neither(?) understands the why, not exactly.

And then there's Yaric, the protagonist of Icebound Machines. The relationship between Yaric and Laura is never made clear. It's weird that they resemble each other so closely, when other lizards look different. They might be siblings. Or perhaps they just bonded over their shared outsider status. Either way, they're linked. We saw this back in Icebound Machines, where Laura's dialogue uses the exact same icons as Yaric's thoughts. (... were these truly Yaric's thoughts?)

Yaric performs one of these cop-adjacent functions, being someone who looks for missing people. Yaric lacks the awareness of Laura, but has more freedom to act. And let's also look back at Parsinomady for two lines of dialogue.

YARIC: Lots to do here at the ministry. Somehow people always go missing right at the start of summer.
YARIC: They have us working triple shifts at the ministry. This isn't how this normally goes. I'm... sure it'll be better soon, though. Talk to you then.

Did you catch that? Time either moved backwards a few months between Parsinomady and Cascade Failure, or perhaps it's the next year. (The former is more consistent with Yaric's statements during the phone call, while Laura thinks it's the latter.) Either way, Yaric has noticed that more people than usual are "going missing," and there doesn't seem to be any reason for it. This is the very start of the apocalypse that unfolds in Cascade Failure.

See, I planned some of those things.

3.) Closing thoughts

This essay is starting to sprawl a bit, so I think this is a good point to start wrapping up. The Icebound Machines series was, in part, an experiment in setting up payoffs that wouldn't happen until much later. It makes me really happy that I got to check those off with Parsinomady and Cascade Failure, and that I didn't have too much retconning. (Just a little.) I hope I avoided getting too high on my own supply there.

One final thing, since I mentioned plans. I once wanted to make a fully-fledged RPG sequel to Icebound Machines. We would've seen much more of the world, from the frozen north to the city of Deluge to the dry grass plains to the Bay itself. That hasn't happened so far, and... well, never say never! But at this point it seems unlikely. I already owe people sequels to much more popular things, over on my alt account. I don't want to get too metrics-brained, but it's hard for me to justify expending my very limited energy on something that's too niche. Some aspects of the worldbuilding got absorbed into Lizard Game, anyway. For example, the "half star" (⯪) was previously the logo of the Bay Empire.

But still! I like Laura and Yaric and some of the minor characters, and I... well, I don't like the Bay Empire. But I think it's an interesting place. There might eventually end up being a sequel of some kind, though perhaps not necessarily a videogame. We'll see.

Thanks for following along so far, and please get in touch below if you want to discuss things further.