Steel Stacks: Postmortem

Tee on 2025/10/12 16:11 of Tee

Thank you all who played my game!

Given the 20th Anniversary edition, I wanted to do something a little more ambitious than what I usually do, and I took a couple of days off work just to work on this. While I didn't actually finish polishing the game that way I wanted -- you might have noticed some missing animations, e.g. glass breaking when falling -- I think it paid off and this is probably my longest Pyweek game so far. I'm happy that you enjoyed it, and I hope the difficulty didn't turn you off too much. :)

I'll talk about SPOILERS BELOW, so please be warned!

.

I had been thinking about knowledge-gated games lately (e.g. Outer Wilds), and I wanted to try to make something that had some of that feel. There's something about the idea of a game as an object that can be unlocked by discovering its secrets that really appeals to me. In addition, I was motivated by mit-mit's Pyweek 39 game (Scuffy) to create something larger with a touch of a Metroidvania-like world. I'm usually minimalist in scope given how time-constrained Pyweeks are -- and I'll continue following that rule for most Pyweeks -- but sometimes seeing the crazy things that other Pyweekers can do within the span of a week can be inspiring (mainly looking at Cosmologicon and mit-mit recently, but also others throughout the years).

The problem is that one week is not enough time to come up with a proper knowledge-gated game. As I mentioned, my number one rule of Pyweek is to always keep it simple, so I decided to set two simpler (but still more-ambitious-than-usual) goals: 1, have a clean set of puzzle mechanics whose dynamics are not obvious and part of the game is discovering them, and 2, instead of relying on keys or triggers to progress to the next area, you discover a secret that you can use to find the next area. If you actually look at the save file, it only consists of the name of the room. You only progress the game to learn more about how the game works, there's never a variable that keeps track of your state other than your location. In fact, I didn't even want the game to track if the player had found the hub yet, you can skip the whole tutorial levels if you want by just pressing ESC in the beginning of the game and going to the hub straight away.

With puzzles, a minimal set of core mechanics is always ideal for a Pyweek game, and the Skyscraper City theme made me think about moving blocks around to build skyscrapers. In particular, I thought it would be cool if building skyscrapers was what helped you move around, and that gave me the idea of putting the player on top of the skyscraper after built, which is somewhere they both could not have reached by themselves, and cannot go back if they drop off (if the skyscraper has more than one floor). My first idea around this was actually to have blocks of different shapes (which didn't get into the game) and an inventory, but then I thought, what if figuring out how to carry the blocks around was actually a big part of the puzzle? That felt a lot cleaner as not only I didn't have to deal with an inventory, but also moving blocks around is half of the puzzle. With that, came the idea of putting the blocks on your head and having them actually interfere (or even help) with what you're supposed to do, and from then on I built everything on top of those mechanics.

The first underground city level might have been a little challenging as you'd have to discover a piece of non-trivial dynamic, i.e. pushing a block off the stack, but I didn't want to straight up tell the player because I think it's a nice aha moment to discover it. I hoped for two things: 1, the player would realize that they need a 4-stack block but cannot actually build one, and 2, they might try to move to the nook with two blocks and drop a block accidentally, and realize that that's a way forward. Hopefully that worked out for most players.

The rest of the puzzles were mostly playing with different ideas on top of these mechanics. I particularly like how the glass block turned out, I think that added some further depth to the puzzles. Some glass block puzzles involved some non-trivial timing.

Also, as with some knowledge-gated games, I really wanted something where if you knew what to do from the start, then you could beat the game really quickly. I decided to show the final levels in the beginning of the game as sort of a preview of what's to come, and make the player curious about it. Then, as you progress through the game, you realize that you're supposed to go back to the beginning of the game to apply what you've learned, and the only way to do that is by deleting your "save". Based on comments many of you might not have actually gotten through the whole game due to the difficulty (understandably), but for those who did, I hope that was a fun little realization!.

Overall, I'm very happy with this one, this was a lot of fun to create. Designing the puzzles was really fun as well.

--

On a different note: It's crazy to look back and think that I've been doing Pyweeks for most of the last 20 years (I only missed Pyweeks #1 and #3 so far). As with pretty much any human being, so much in my life has changed in these 20 years, and yet those two weeks a year of Pyweek have continued to remain a constant and that's something I appreciate. Special thank you to all who helped keep Pyweek up and running throughout these two decades (Daniel, Richard, and anyone else I might be missing)!

As always, I've enjoyed playing your games. Congratulations to Large Sandwich Models for winning the team category, and to everyone who submitted a game! Making a game in a week is never easy. :)

(log in to comment)

Comments

mit-mit on 2025/10/13 09:40:

Congrats Tee on the win and scoring THE GREATEST PYWEEK GAME OF ALL TIME! What a way to go with the 40th aniversary :)! 

Well done on a really good game: what you have mentioned about the different puzzles being structured to help you figure out a new game mechanic (like sliding the blocks off the top of the stack) was spot on and it's great how you can do this without having to be explicit about instructions: great game design!

Look forward to playing more of your games! Let's see if we can break the 4.50 out of 5.00 score barrier before the pyweek 60th aniversary :P ...

Tee on 2025/10/13 22:44:

Thanks, mit-mit! I appreciate the comments on game design: I really enjoy puzzle games that give room for the player to play around and discover what to do instead of telling them outright, but designing them requires a tricky balance and I get a little paranoid that it ends up too frustrating. I'm happy to hear that it worked out!