CaddieOpen in CaddieDownload for Mac

How I make puzzles for my indie magnet game (Developing 9)

Shared by Caddie Library · Updated
Game Maker's Toolkit12:08Open on YouTube
Chapters show their main points

The bottom line

A puzzle matrix that pairs every two game mechanics gives a designer a concrete prompt, so puzzle ideas come from experimenting rather than from inspiration. That structure turned stalled level design into steady progress, with 26 levels in a month.

Key takeaways

  • Level design stalled because "make an interesting puzzle" is too vague a goal. 2:45
    • Building a new mechanic starts with a clear goal, like a block the robot can push left and right. That work felt like hyperfocused game development, with no lunch eaten.
  • Pairing every two mechanics in a matrix gives each puzzle a concrete starting point. 3:49
    • Electromagnet plus drill: a magnet stuck under the drill escapes quickly near the tip and slowly on the far side. That timing trick became the whole solution, and the level was built backwards from it.
  • Each finished puzzle made the next one easier. 6:07
    • He learned to watch how many actions the player has at once, and to lure players into a wrong assumption.
    • Every new blocking mechanic joined his toolbox, and rewiring buttons down to one click made experimenting faster.
  • A puzzle bible that lists the mechanics in each level shows where the progression jumps. 8:37
    • Level four added orange buttons, green buttons, lasers and moving platforms at once. That gap became a set of new levels that teach those mechanics first.
  • The result is 26 levels and a visible road map, but only playtesting can judge them. 9:45
    • Not every combination makes a good puzzle, and progress was sometimes still slow.
    • Creators cannot gauge their own puzzles' quality or difficulty, so a demo goes to family, friends and Patrons next.

So far it's been like climbing a mountain where the peak is obscured by fog, but in January I felt like I could finally see the summit.

Mark Brown

Chapter summaries

0:00 A 30-puzzle challenge that stalled

Concrete goals energised him; vague goals drained him
Mechanic workLevel design
Starting goalBlock robot can push left/rightMake an interesting puzzle
StaminaHours on end, skipped lunchA couple of hours a day
How it feltHyperfocused, like old timesLike pulling teeth
ResultsSteady progressFew good puzzles, felt like flukes
The difference wasn't the task's difficulty but whether it began with a clear, concrete goal.

Making puzzles on demand was slow and draining, even though building new mechanics still felt great.

  • Coming up with puzzle ideas on demand was painfully hard. 0:57
    • The goal for January 2023 was 30 puzzle chambers of real content, not throwaway test levels.
    • His few good puzzles felt like flukes, so progress was like pulling teeth.
    • He could only work a couple of hours a day before getting bored and burnt out.
    • He wondered if he was done with the game after more than a year, plus a four month break.
  • Mechanic work felt easy because it started with a clear goal. 2:11
    • New buttons, doors and blocks meant art, code, testing and tools, and he lost track of lunch.
    • A level only has the goal "make an interesting puzzle", which is too ambiguous.
    • A mechanic goal is concrete: a block the robot can push left and right.

3:09 The puzzle matrix

Pairing every two mechanics in a grid turns a vague goal into a concrete experiment.

  • A matrix of mechanic pairs gives each puzzle a starting point. 3:39
    • Patrick Traynor, creator of Patrick's Parabox, wrote the post that inspired it.
    • Screenshots of every mechanic go across the top and down the side, with duplicates removed.
    • Each cell is two mechanics, such as the electromagnet and the drill, dropped into an empty Unity room with a few extras.
  • Experimenting with a pair reveals a trick, and the level is built backwards from it. 4:28
    • A magnet stuck under the drill escapes fast near the tip and slowly on the far side.
    • The player must trap it away from the tip to reach the doors in time.
    • Other pairs tried: electromagnet plus laser, block plus drill bit, laser plus wheelie block.
    • Motivation came back because the matrix gave him a prompt and a jumping off point.

6:04 Why each puzzle got easier

Puzzle design sped up because skill, mechanics and tools improved together.

  • Practice builds puzzle sense like a muscle. 6:13
    • Too many available actions confuse the player.
    • Too few let the player repeat the obvious move until it works.
    • Repeating the obvious move solves the puzzle "without actually solving it".
    • He reused an idea from his earlier puzzle video: lure the player into a wrong assumption, so they get stuck and think laterally.
  • Each new blocking mechanic closes shortcuts for good. 6:58
    • A puzzle must force one specific action, so easier routes need cutting off.
    • Examples: a laser that stops the player but not the magnet, or the reverse.
    • Other examples are a button that must be held down, and a door that flips a switch each time you pass through.
    • He swapped a flippy lever for a spinning fan blade, which combines with more mechanics in the matrix.

8:37 Ordering levels with the puzzle bible

Level four introduced too many mechanics at once
Level 13
Level 21
Level 32
Level 44
Gaps like this became targets for new introductory levels, each with a clear goal.

A slide per level shows which mechanics each puzzle uses, so progression gaps become visible.

  • Listing mechanics per level exposes jumps and turns them into new levels. 8:37
    • Each level gets a Keynote slide, so stages can be reordered to introduce mechanics slowly.
    • The goal was 30 levels in January. He finished about 26.
    • Level one has the red block, electromagnet and touch buttons. Level two adds blue blocks. Level three adds the drill bit and winch.
    • Ordering by difficulty waits until after playtesting.

10:19 The road ahead

The finish line is now visible, but many months of work remain.

  • The remaining work is a clear list. 10:32
    • More levels, a story with cut scenes, and a level select screen.
    • Music, polish, bug fixing and release.
    • He expects "still many many months" of work.
    • The visuals were also overhauled, so the game looks like "a proper, bona fide video game".
  • A demo goes out for playtesting first. 11:05
    • Patrons can play it in the "Developing with Mark" thread on the GMTK Discord.
    • The next update is hoped for in March.

In the growth library

Mark Brown, creator of Game Maker's Toolkit, making his first game. Teaches a repeatable way to generate puzzle ideas by pairing every mechanic against every other, then ordering and filling gaps in the level list. The worked example is the electromagnet plus drill bit puzzle, where the magnet's escape time depends on where it is trapped.

Lessons that draw on this:

More from Game Maker's Toolkit · The whole library, and ask it anything

Transcript124 lines
0:00

Hi, my name's Mark and I am making a video game about magnets. So in the last episode of this series I said that my next goal was to work on some actual content for this game: the puzzles for this puzzle adventure. All of the stuff I've made so far has just been test levels and prototype stages, stuff that ultimately ends up getting thrown away when I change the code too much. But in the spirit of 'finish the game', it's time to knuckle down and really get to work on the content that should - hopefully - end up in the final game. So my challenge for January 2023 was to make 30 individual puzzle chambers. How did I get on? Well stick around to find out.

0:57

So on January 2nd, I sat down in front of Unity with a big mug of tea and a great deal of determination and went "all right! Puzzles. Puzzles... puzzles". Man! Turns out it's really hard to just come up with ideas for puzzles. I've made a few good ones during the development of this game but they honestly feel like flukes that came about at random. The thought of doing that on demand, 30 times in a row: I'm not sure how to do that. And so in the beginning, progress on this challenge was pretty slow. I made a few levels but it was like pulling teeth. And this really shot my motivation for working on the game. I could only do it for a couple hours a day before getting completely bored and burnt out. And so at this rate I simply wasn't going to meet the goal.

1:53

Now I did start to think that maybe I was just done with this game after working on it for so long. It's been well over a year now, plus I just took a massive four month break at the end of last year. But no. You see, sometimes when I was bashing my head against one of these puzzles I realised that I needed to make some kind of new mechanic: a new button, or switch, or a door, or block, or whatever. And so I would do the artwork and then write the code and then test and debug and make tools in Unity - and suddenly it was like game development used to be! I could just hyper focus on the problem for hours and hours at the time without even realising that I haven't had lunch today.

2:34

But then I go back to level design and it was slow and difficult and tedious. But I realised that there was a fundamental difference between these two modes of working. You see, when I was working on a game mechanic it would always begin with some kind of goal: like I need to make a block that the robot can push left and right. And then I can work towards that goal. But when I'm making a level, the only goal I have is "make an interesting puzzle", which is pretty ambiguous. What I could really do with is a bit more structure. So I looked for resources to help me come up with ideas for puzzles and came across this great post by Patrick Traynor, creator of last year's wonderful game Patrick's Parabox: a sort of mind-bending, meta-level recursive Sokoban gem. It was one of the honourable mentions in my video on the most innovative game of 2022. It's a great game. Anyway, this post has loads of great advice on coming up with

3:33

ideas for puzzles but there was one thing that really stuck out to me. so here's what I did. I took screenshots of all of the mechanics in my game: that's the magnets, the drill bit, the one-way platform, the electromagnet, the laser, and so on. And then I put them along the top of a table. I took those same mechanics again and put them along the side of the table and got rid of all the duplicates. This left me with a matrix, where every cell is at the intersection of two game mechanics. This one, for example, is at the intersection of the electromagnet and the drill. I would then take those two game mechanics and put them into an empty room in Unity, alongside some other bits and bobs like the character, maybe one of the magnets, some buttons, some floor tiles, whatever. And then I just start playing around, experimenting, just kind of messing with the system, and looking for interesting interactions, or

4:28

interesting consequences of these two mechanics. Like, for example, if the drill bit is over the electromagnetic beam then my little magnet gets stuck. And then if you retract the drill bit, if you are near the tip the magnet will quickly shoot up and escape, but if it's on the other side of the drill bit it will take much longer. And so this little realisation, this little trick: that can be the solution to the puzzle. That is what the player has to figure out.

4:58

And then I can essentially work backwards to create a level where the only way to solve it is to do that one thing. And so if I just add a few more mechanics, like so, here we get a level where you have to trap the magnet underneath the drill bit, away from the tip, so you have enough time to get here before the magnet escapes and makes these two doors switch places. That's not a bad little puzzle! What else is on the puzzle matrix? Electromagnet plus laser beam... okay not bad. Block Plus drill bit... quite like that. Laser plus wheelie block... yeah pretty good.

5:36

Suddenly, puzzle design wasn't so difficult anymore and my motivation was coming back. I put this largely down to the puzzle matrix: it's all about giving me that prompt, giving me a suggestion, giving me a jumping off point. Just some kind of structure to help me find ideas. Now of course it didn't always work. Not every combination leads to good puzzles and sometimes progress was still difficult and slow. But overall this made puzzle design so, so much easier, and it got easier with every puzzle I made. I put that down to a few things. One is that I just got better at puzzle design. I learned the tips and techniques that you get when you do something over and over again, and it's like building a muscle.

6:23

So, for example, I started to see the importance of the number of actions the player has available to them at any one time. If it's too many, the puzzle can feel confusing and overwhelming. But if it's too few then the player can just do the thing that's in front of them, over and over again, until they solve the puzzle... without actually solving it. I also thought back to my video on puzzle design where I talked about luring the player into making a wrong assumption about the puzzle, so they get a bit stuck, and then have to think in a lateral way to find the real solution. So I tried to incorporate that into some of my puzzles as well.

6:58

Puzzle design also got easier because I boosted the number of mechanics I had available to me. As I said before, so much of puzzle design is about forcing the player to do one very specific thing, so you need to cut off other routes that would allow them to finish the puzzle in a much easier way, and render the whole thing pointless. That usually comes down to some kind of mechanic like a laser beam that can block the player but not the magnet, or a beam that can block the magnet but not the player, a button that you have to hold down for it to operate, or a door that operates a switch every time you go through it.

7:31

Every time I made one of these to solve some kind of problem in my puzzle design, I would then have it in my toolbox to solve that same problem, much faster, in the future. I also swapped some of my mechanics for ones that would be more open to combinations in the puzzle matrix. For example, this flippy lever is pretty cool but I swapped it for a big spinning fan blade which can do things like this... that could be a puzzle!

7:57

And the final thing that helped make puzzle design easier was investing even more time into my tools. Like, before, if I wanted to wire a button up to a mechanic it would take many clicks to assign the right function to both the on and off triggers for the button. So I rewrote the whole thing and now that whole thing can be done with a single click. So much of this process is about playing around and experimenting and iterating, so I want it to be as friction free as possible for me to edit these levels.

8:27

So between the puzzle matrix, and the tips and techniques, and the added game mechanics, and the improved level design tools, I was able to make a whole bunch of levels for this game. And each time, I would create a slide in my puzzle bible in Apple's Keynote. On each slide I would jot down which mechanics I used in that puzzle, so then I could easily reorder the stages to have a nice progression of slowly introducing the player to the different game mechanics. Eventually I'll also want to do this for difficulty as well, but that will come after play testing. And then when I looked at the puzzle bible I could see places where the progression just kind of jumped, so like the first puzzle I have introduces the player to the red block, the electro magnet, and these touch buttons. Then the second puzzle adds blue blocks. And then the third puzzle adds the drill bit and the winch. But then the fourth puzzle adds orange buttons,

9:22

green buttons, laser beams, and moving platforms - woah, that's way too much stuff! Slow down! But I could see this gap as fertile ground for making more puzzles, and again with a really clear goal this time of introducing the player to the mechanics that they would need for that fourth level. And so each gap in the puzzle bible would add even more levels to the total. And so did I meet my goal to make 30 levels in January? Not quite! I got about 26 before I had to stop and make this video. I am still running a YouTube channel, after all. But I am really proud of my progress and I'm really excited for the game now, especially after I overhauled some more of the visuals. I think it's starting to look like a proper, bona fide video game. And so in the space of a few weeks I went from hating my game and wanting to delete it and go live in a cave in Serbia to being really excited about this game coming out.

10:19

And for the first time - the first time this entire process, I feel like I can see the end. So far it's been like climbing a mountain where the peak is obscured by fog, but in January I felt like I could finally see the summit. I can see the road map I need to follow to get this game done. I need to make a bunch more levels; I need to do some kind of narrative, story, cut scenes, something; I gotta make a level select screen, I've got to do music, and I've got to do the general polish and bug fixing and release. That's a huge amount of work, still many many months, but it does at least feel like a road map. It does feel like something I can follow to get this game done. But the next step is play testing these levels, making sure they're actually any good. Puzzles, more than anything else in game development, I feel, are something where the creator just cannot gauge their quality or difficulty. So

11:23

I need to bundle them together into a demo and give them out to people to play. That will start with family and friends but then it will go on the GMTK Discord for Patrons. Make sure you're in the Developing with Mark thread if you want to have a play. That's all for now, thank you for watching, thank you for your patience, and I'll see you - hopefully - in March!