Making an Even Better Map?

As of two weeks ago, I've been working on Golemancer for over a year! I initially intended this to be a quick and simple project, just to experiment with making a roguelike and get another game under my belt. The reason it took so much longer than I planned is partly because of overambitiousness, but mainly because of my lack of discipline when it comes to actually working on it.

...Anyway, that's not what this post is about! I'm here because, despite my efforts in this 11-month-old devlog, the map generation in Golemancer still leaves a lot to be desired. In particular, how many times has this happened to you?

"Choose a path..." my foot, the only choice here is how long you spend yawning at the complete lack of meaningful decisions this presents. These doors do technically lead to different places, but because you can't see the map ahead of you (unlike many other roguelikes), as far as the player is concerned, they're all the same. To make matters worse, battles provide almost no benefit compared to buff rooms, so if you see a buff room vs. a battle, that's also a meaningless choice.

My previous attempt to remedy this was what I refer to internally as "senses". Sometimes, you get these little popups hinting at what's in the room ahead of the room you're looking at.

In the example above, your choices are:

• A battle, with a health buff room ahead of it.
• A health buff room, with a sigil battle ahead of it.
• An attack buff room, without knowing what's after it.

The sense system does help a little sometimes, but you may have noticed that in the above screenshot, it doesn't actually change anything. The only reward that the battle promises is a health buff room... but there's already a health buff room right in front of you. 

As I've shown in previous devlogs, a full view of a Golemancer map looks something like this:

(If you're curious, the icons on the exits at the bottom indicate whether they go to the Fae's Grove or the Yeti's Pass.) 

Bigger is better, right? A small, mostly linear map would be boring... at least, that's what I thought when I designed this system a year ago. But honestly, what's the point of having a dozen different paths, if the player has to choose just one - and they don't get to see what's on it first? It's a lot of needless complexity. Moreover, it adds a lot of unpredictability to what sorts of paths players will end up taking. Not the good kind of unpredictability; the kind of unpredictability where Player A will get nine buff rooms in a row and breeze through the boss in one turn, whe Player B will get stuck with hardly any buff rooms and get wiped out by the harder battles.

Now, at this point, carelessly making any big changes to how the map generation works would completely throw off the balance of the three existing areas. So, the first thing I did was make an analytics tool for the current maps. Here are the statistics it gave me for the map I showed above:

(The huge number of possible paths surprised me, too. Remember that it's counting EVERY possible permutation of rooms you could visit, so exponential growth makes it get out of hand pretty quickly.)

From running the analytics tool on several maps, I learned a few interesting things, but the most important one was the average number of buff rooms, which was always pretty close to seven. Keep in mind that this is averaging across all possible paths, but a human player would probably seek out buff rooms deliberately, so the average number they get would be higher.

So, my goals now are:

• Reduce the likelihood of pointless choices between multiple battles, or battles vs. buff rooms
• Make the number of buff rooms a player will visit more consistent (to prevent the Player A vs. Player B scenario mentioned above)
• Keep the average number of available buff rooms about 7, maybe a bit higher

To achieve all three of these, I present the new helix map!

(I called it the "helix map" because my initial concept for it looked kind of like a helix, but it doesn't look much like that now.)

Outdated names aside, here's how it works:

• Like before, the first room is always a golem recruit. The room after that is always a battle. The fifth, eighth, and eleventh rooms are also always battles, and there's only one of them (rather than multiple in different rows).
• Between those fixed points, there will be two or three paths that connect them. Each of these paths are two rooms long, and which two rooms are chosen from a premade list of pairs, some more common than others. Most of them are buff rooms, but sometimes battles will show up there as well, usually paired with a stronger buff room (like a golem recruit) to balance things out.
  • When choosing the pairs of rooms, it is forbidden for a row to contain two of the same room. (the two battles in the third row up there are a special case; please ignore them)
• Extra diagonal lines may be drawn between the multiple paths, so the player will sometimes have an extra choice about which buff rooms to visit.
• The fourteenth room will be a boss fight, and there will be two of them, with the door afterwards going into different areas. (The player can choose which area they want to go to next by picking the appropriate direction after the eleventh room.) 

These thinner maps are much more predictable, which means it's less likely the balance will be thrown off by a series of good/bad luck. They maintain the approximate number of buff rooms from before, while preventing you from getting too many in a row (since the battles are fixed at certain points). And they make it so that when you're making a choice about where to go next, it will almost always be an actual choice.

I think this system seems pretty good, but I can't be sure until I test it, and of course I'll want to get some opinions from others as well. So, please let me know what you think by leaving a comment! And, naturally, thank you for reading.

P.S. The one other thing I'd like to do to make maps more interesting is add more types of buff rooms to visit. So, you can expect to see more of those in the next update. In fact, you can see one in the map view already...

...who could this be?