Misc Thoughts

[MicroChasm]

I wanted to create a place to write down random thoughts that don't necessarily belong in any specific git issue.

[MicroChasm]

Because being in sneaking mode doesn't actually prevent golems from seeing you, stealth is currently about avoiding golems entirely. I've mentioned the idea of a percent-based likelihood of being seen, which we might still want to add. It's hard to say if this would add a cool push-your-luck kind of mechanic, and making sneaking more tense, or if it would add a difficult-to-understand and communicate mechanic to the game that doesn't jive with the otherwise WYSIWYG style of the game.

Since the grass character is the pure stealth one, we could improve their stealth by adding a tall-grass ability. I'm imagining the player creating a line of tall, inter-tile walls made of grass. The player and golems can move through the grass but it blocks line of sight. Potentially it doesn't last forever, or it stays down when trampled.

I really think that grass players need a way like this to block line of sight because otherwise there will be situations where they can't help but step into the sight of a golem, and then the only thing preventing them from being seen is the chance that the golem is facing the other way.

I think that we can also mitigate this a bit by increasing the wally-ness of the levels as well. We could also bring back up the idea of peaking around walls and see how we feel about it this time.

[MicroChasm]

I was thinking about the number of options that players have. I feel like we don't want more than 7, maybe 8 at most, because we don't want it to get overwhelming. However, maybe later in the game it would be ok to add an option or two.

Mostly I was wondering if we want to increase the ability number to 4 at some point, just so that the player can do more sweet stuff. I was also thinking that, and I think this has come up before, we may want to handle rocks differently than other items-- maybe they take up a slot dedicate to rocks and stack infinity. I thnk that if we scale up the number of items and abilities over the course of the game, it would work.

[nsmryan]

These are good ideas. Keep them coming. I agree about stealth being all or nothing. Grass or tall grass could deft effect line of sight. The grass charavter could even have an ability where grass blocks LoS to make them more stealthy.

I agree about adding skills. They are fun. I guess the other option is to allow swapping them out at times.

If we want rocks to be different we coupd add an issue. I dont think its too much of a change.

[MicroChasm]

Maybe you build a collection of abilities through a run and you choose 1-4 for a particular level, depending on how far you are in the game. I've thought in the past about having different types of levels, like curses in Isaac, where one level is during a sandstorm and has reduced visibility or something. Maybe you know a few things about the level before you choose your abilities? If we do end up doing something like this, I would imagine it being much further down the line in development.

I would say we should discuss the rock concept to see if we want to go ahead with it unless you feel like working on it. If you do, feel free to make the decisions that go along with the change.

[MicroChasm]

I feel like the hierophant character can't be 100% about planning. With procedural levels, there are going to be times when no amount of planning will get you through a situation. You would have to just run and jump sometimes, and this class doesn't have much ability to neutralize situations. Even if you could go through a level perfectly, it wouldn't be much fun to do over and over unless it was tense. Our game currently doesn't have patrolling enemies, so it is pretty deterministic if you have enough information. We could always add patrolling, but I would be cautious with it. For some reason, I have a bad feeling about patrolling enemies in roguelikes, but maybe it just hasn't been done well in the past.

I believe that we have talked about the hierophant having some illusion abilities in the past. I think it would work well to be able to gather a lot of information and then, say, create an illusion of yourself to move around for a while and distract monsters. We could experiment with being able to move both you and the illusion at the same time, but I would imagine, rather, that you would get into a safe spot, create the illusion, move them for a while having the ability to break the illusion and go back to your body at any time. It might cause some awkwardness with LOS if you have two viewpoints... maybe your body just blends into the darkness and is invulnerable for a while, but you can only initiate a mimic when you aren't in LOS of a golem?

I like the idea of a mimic because it works well with the idea of planning, and it feels that you are interacting with the game in a way that is unique to the class. Ideally, you would be able to have layers of plans-- planning a route for the mimic and planning the route that you will be taking right after.

In addition to the idea of a mimic, the hierophant could have some other trickster abilities, like a flash-bang kind of thing, or a trapping ability. You could lay traps, then lure golems into them with your mimic as well. You would be planning out tricks and then trying to pull them off, sounds fun.

[MicroChasm]

I don't have a fully formed concept here, but I was thinking about a combat-focused class, which I believe we have discussed a bit. With the changes to weapons, I think it is worth thinking about again. I would imagine that you would have some kind of super special permanent weapon that you carry with you, and you would have abilities related to it. My first thoughts were that you would get one extra turn of stun from your attacks, and your special weapon is a longsword that acts as both a sword and a spear. Ironically, this will make you pick up fewer weapons during a run, since you already have this combo. Idk, it's probably fine, but maybe we should come up with a completely unique method of attacking so that your weapon feels special and you still spend time interacting with the other weapons in the game.

In either case, I have been trying to think of a way to make unique and interesting weapon interactions for the weapons character. I really liked how it felt in I of the Storm, when you cast spells by inputting a set of movements-- I feel like this idea or something similar could be used to make an ability that feels like doing some kind of sweet sword-fighting maneuver. Maybe the weapons character doesn't even have abilities like the other characters, but instead has maneuvers and passives. It would help to keep them really unique.

[MicroChasm]

I was thinking about some of the abilities in Sil, and especially the Piercing ability. Piercing lets you get an attack on a monster that is behind the one you are currently attacking if you are using a spear or greatsword. While I normally use the ability to get in additional attacks on monsters that are lined up in a hallway, I always think about how to line up monsters when I am fighting in a room as well. I think it would be cool, when we are thinking about how the weapons works, to have these sorts of situations where you can get additional attacks if you do things in a particular way.

The sword swipe already does this because, as we discussed, you can attack two monsters at once by moving between two tiles which are adjacent to both. Maybe we could add something where if you run while using the spear, you attack multiple tiles in a line.

[MicroChasm]

I watched a little of Dynast playing DCSS the other day, and he saw a new enemy that was added recently that can shoot lighting attacks at you, but only if you are a certain distance away. It reminds me of the archer in Hoplite. I like this idea, I think its worth keeping in mind for either a player or monster ability.

[nsmryan]

These are good stuff. The fixed distance is definitely interesting: it forces you to keep track of things and adds a level of constraint to the enemy.

[MicroChasm]

We could have another type of item, consumables, which are on the q, w, and e keys. These would be things like potions or grenades that have special effects. It adds some complexity to the game but it also adds another layer of interactions and synergies. We could either keep these simple by making them single-press items, or we could incorporate them into the other systems, so they can be used actively, or thrown.

[MicroChasm]

It would be neat to have a portal-using character type that uses mirrors. Maybe they can create two mirrors and pass between them. I also had the idea of them being able to mirror small parts of the map to change its layout. It would be cool to have some kind of optical illusion in the unity version related to this mirror idea, but I haven't been able to figure out what it would be.

Alternatively, any implementation of a portal system would be neat. maybe you could place portals under golems, or throw a portal-making granade, or create a bunch of portals that you can go through, but you can't choose which one you come out of. It could be the escape artist class (as opposed to planning, sneaking, fighting, or movement).

[MicroChasm]

It would be nice if we had some sort of way for stealthy-leaning characters to bridge the gap between pieces of terrain. I think it would feel cool to be able to do a silent roll move for instance. We could think of some way to nestle this into the existing interactions, like if you run while in the crouching state you do a silent roll, and you can't be seen while on the intervening tile.

The diagram below shows the player sprinting across a gap without being seen

[MicroChasm]

Some thoughts on Jupiter Hell.

The game definitely has more polish and some new features since I last played. I like some small changes, like the fact that health chests now give you one medkit and one health item that is used immediately as you step on it. Just having these two levels is nice. The game starts with an opening animated cinematic, which is pretty brilliant. It shows the character shooting monsters, using walls as cover, which teaches new players how to play the game before they even start. I really like that pressing "f" makes you fire your weapon. I had been playing a little Angband, and the firing it pretty bad. to fire at a monster that is not on a cardinal direction or diagonal, you have to press "f" to initiate firing, then 0, 1, 2, etc to choose the ammo type, then to target a monster instead of firing in a direction, then to cycle through monsters, then "t" to actually fire. In Jupiter Hell, you just press "f" and it fires immediately. If you want to target a tile, you press "t" to enter target mode, then move the cursor, then press "f". The targeting mode also gives you info about the tile. I like there are some breakable objects around. You can break them by walking onto them or by kicking or shooting them, and they do what is clearly a mesh-swap with an object that has had a voronoi operation done to it. The objects fall in a satisfying way. The levels get more open as you go onwards, which is interesting because you start feeling safer when you are around walls. there is also a fair amount of branching paths in the game, which I don't understand yet. I did find a workshop level with a bunch of monsters in a small space, but lots of upgrades once I cleared them. The game asks you to play the tutorial if it is your first game. It is the kind of tutorial that makes you take specific actions, which are highlighted-- like move down, move right, etc. It will tell you what you were supposed to do if you press the wrong key. The game does not do as good a job as I want to do with making the player and other things super clear. Instead, it makes it feel more like an action game with the way things blend together. This is a fine choice, but not what I want for us. I think my favorite thing so far has been this-- in my run, I decided to take the gunslinger traits, which give you better accuracy and more crits with pistols, and eventually, you can dual-wield. In addition, you can get mods for your equipment, that you can add to your armor, helmet, or guns. When you use the mod, it gives you two options to choose from, from a list related to the type of mod. There has been a fair amount of level variety, from spaceship to caverns to some sort of religious area I think. Nothing overly ornate or complex, just different walls and misc objects. The game doesn't really darken areas outside of your LOS. Instead, it changes the light color from a warm yellow to a desaturated white, and you just can't see things between the yellow. The game appears to be segmented into chapters, I cleared the spaceship levels and went down to the planet for the next set of levels. This resets the level size, so it starts small again, which is interesting. Maybe an interest curve choice. There are environmental objects to interact with that have a certain number of charges, that you can use to either manufacture medkits, make armor or weapons, or there is a terminal for opening locked doors and other things. The Tech is good at using these. Some enemies, walking turrets specifically, create smoke when they are destroyed, which blocks LOS. You can also pick up smoke grenades, and the Scout class can turn invisible, which I found super powerful (idk if this is still the case). The effect of taking the pistol abilities and upgrading my pistols made me feel both that I was building my own custom character, and also I felt a connection to my guns. I could have taken huge automatic guns, or shotguns, or close combat weapons, or chosen a different class, but I played a marine with pistols this game, and it felt like a unique experience. In the end, I think we should use the intro and tutorial especially as inspiration, and we should consider the idea of customizing equipment. My biggest complaint is that I don't understand the mix of clearing a level and diving down, but that is something I guess just comes with experience. Overall, I had a good time with the game, and I would like to try some different characters.

[nsmryan]

It seems that there could be more levels of reward for powerful/dangerous actions. For example a running spear attack could disable a golem, while a powerful hammer attack, or some other particularly powerful attack, could destroy them. When disabled, it might just be like getting stunned, but they enable again after a long time. It would be awesome though if other golems would enable them if they are in FoV while the golem is idle.

[MicroChasm]

I'm still trying to figure out how we handle weapons. So far I like the direction that the input is taking and I am hopeful that we can come up with some interesting mechanics around the weapons. What I don't want is to just have some weapons on the floor, the player picks them up and then just uses those weapons the whole game.

I think that there are a few different ways to avoid this. One would be to have just a whole bunch of weapon types that do all sorts of things, getting pretty granular and creative with the abilities. This way, players might want to experiment with different weapon combos, and might try to combine weapon combos with abilities.

While I think that this could work, I also think that players will still find the combo that works best for them, or they think works best for them, and then just wait until they find these weapons and stick with them for the rest of the game.

One option is to have all weapons be consumables. The problem with that is that I think that having all your weapons break would feel bad. I think you will get into situations where you need to escape and all your weapons break all at once and it is very frustrating.

One option is to have two or three weapon slots, and allow players to fill them however they want. Some players will just get the permanent weapons that they want, and stick with those, and we can allow them to do that. At least they have to experiment early in the game. Other players might want to fill only one spot with a permanent weapon, and then keep the other open for flexibility. This only works if consumable weapons are frequent.

Another option is to have two slots, one only for permanent and one for consumable weapons. This might work. It doesn't really make any sense from the perspective of the character, but we could allow it as a gamey mechanic.

We could make consumable weapons valuable or desirable in some way. For instance, it could be that consumable weapons are actually more effective in some way, so players have to balance between the stability of permanent weapons and the power of consumable weapons. While this seems like a nice balance, it doesn't really make a whole lot of logical sense. The only way that I can see it working is if, say, greatswords are always consumable and shortswords are always permanent. In Breath of the Wild, weapons break all the dang time, and they do more damage on the hit that breaks them, but this doesn't really make sense logically either.

In the end, I don't really have a conclusion here, just posting some ideas and thoughts.

[MicroChasm]

I wanted to share some thoughts about chess here, and think about what we can learn from it as a game. It's something that we have talked about in the past, but I think that now is a good time to revisit it as we continue finalizing the mechanics of the game.

I've been trying to understand and express the features of chess that make it interesting. There are a lot of them, with no real easy answer.

One obvious one is that chess is played against an opponent, which makes most games inherently more interesting. This isn't something that we can really emulate, since the golems in our game aren't supposed to be intelligent and because they have different objectives than the player. However, and I think we may have talked about this before, I wonder if it is worth considering, in the case where a golem could take one of two moves with equal weight (or maybe just similar weight) if they decide randomly between them instead of choosing the first on that they think of. It might add just a bit of randomness that could help to disrupt the player's plans by appearing to make decisions, even if any disruption is by accident. However, since this is a roguelike I don't think we really have to try to emulate chess in this way.

Another thing that makes chess interesting is that it has an escalation of power levels throughout the game. This didn't occur to me until recently, but you start the game by moving pawns and lesser pieces, then only later in the game do you move on to the queen and rooks. This is enforced loosely, just because of how the rooks move and how they are placed on the board, and because bringing out the queen early is risky. I think we will likely have a reasonable escalation of action in our levels, but it is something to keep in mind.

The pawns are a big part of what is interesting with chess. They are like a fortress that you can move. They have the biggest swing of power in the game as well, where they are considered the weakest pieces, but can also win the game. A pawn can take a piece as well as any other after all, and they can be promoted. Many of the beautiful games on Agadmator's channel involve a bold pawn push. Abilities that block movement or attacks are worth considering.

Another thing about chess that we have definitely talked about is the fact that your pieces gain or lose power depending on their position. The hard boundaries of the game, where most pieces can't move over others, and you can trap yourself with your own pieces, create really interesting positional challenges to solve. We have discussed it before but I feel that some sort of golem that attacks like a chess piece by moving itself, may be worth adding as I think it would increase the likelihood of interesting positional challenges. I do think that, even with no changes, we have some of this represented in the game with walls and golems, and the fact that attacking generally causes a move.

The concept of piece development brings us closer to my actual point in all this. The rules of chess make it so that the meaning of "doing well" in a particular game is fluid. Generally, you have the goals of developing your pieces, controlling the center, and protecting your king. If you lose material, then the other player is at an advantage. However, there are cases where you can lose material in order to gain development or to weaken the other player's king. There are cases where it is ok to lose development to save pieces as well, like the game where Bobby Fischer moved his knight back to the first rank to escape a pawn push, so early in the game that he was left with no pieces beyond the first rank (when asked about the pawn push after the game, his reply was "weak").

I think that, if it is possible to summarize the reason that chess is interesting in one short concept, it would be this shifting of priorities. The harsher and more sudden the shift, the more beautiful. The most beautiful moves are the ones where you do something that seems completely impossible, and suddenly everything changes. It lights up the brain to re-evaluate the board like this. In the game The American Beauty, the Black player moves his queen so that it can be taken by the opponent's queen or either of two pawns. At face value, it seems like a crazy move since you are putting your most powerful piece in a position to be easily captured, but in fact, any move by White after this is losing, and after this move, White resigned the game.

With this thought in mind, I think that we should have it so that you gain back all of your health and energy between levels. First of all, this reinforces the idea that everything that you have to worry about is on the board; you are not worried about what is on the next board. Second, it creates a checkmate-like situation where you could give up all of your resources just to get to the exit, but if you get to the exit square then you win. It's also sort of like a pawn push in this way, where squares closer to the exit have greater.

We have other situations where you might re-evaluate the board, like finding a trap or wall or some other resource, but I think it may be worth brainstorming some additional mechanic that changes things and makes you re-evaluate. It could be something organic that the player does, or it could be more specific. A more artificial but still effective example in game design that comes to mind is eating a larger dot in Pacman allowing the player to eat ghosts. The player takes an action, and all of a sudden they completely re-evaluate their position and goals.

[MicroChasm]

I wanted to write down some thoughts about the design of Super Auto Pets. It's worth thinking about because part of the beauty of the game is the amount of complexity that arises from clear, simple rules.

All of the abilities of the pets in the game are described in a short sentence with a clear format, e.g. "Level up -> give all friends +1 attack, +1 health" or "Faint -> Summon a level 1 tier 3 pet as a 1/1".

The game makes use of all the different events that can occur in the game. Some triggers are: the pet fainting, a friend being summoned (in battle or in the shop), before the battle, the pet being sold, feeding the pet a food item (buff), the pet being bought, the pet making another pet faint, and the pet being hurt.

These interactions can all influence one another in unexpected ways. For example, normally the Gorilla gains coconut armor (which negates and attack) when it is hit in battle. However, if you have a Hedgehog, which hits every pet on the screen when it faints, and you feed the Hedgehog a sleeping pill in the shop, its faint-trigger will proc, causing it to hit all of your pets. This is will trigger the coconut armor and give it to the Gorilla outside of battle.

It is possible to have really silly interactions occur as effects chain together. For example, a Dolphin (who can attack once at the start of the battle), might make a Hedgehog faint before any attacks have occurred. The Hedgehog will then hit everything on the field with a low-damage attack. This can trigger a Blowfish to attack, since they attack a random pet on the other team when hurt. If both teams have Blowfish, the Blowfish can trade attacks back and forth as they hit eachother and their hurt-trigger is proc'ed again. the blowfish could then hit another hedgehog, causing the whole thing to start again until all the pets have fainted, before the battle even officially began. It's not likely, but I've seen this or similar things happen several times.

One of the most interesting pets in the game is the new Rat, which was changed just a few days ago. I can't say that I think that it is good, but it is certainly interesting. When the rat faints, it spawns a low-stat rat on the other team, in the front. This give the other team an extra pet, and the rat does attack your team. However, since there are abilities that proc before attack, or after getting hit, spawning this enemy rat is a way to get an extra proc at the cost of some HP. Super interesting.

There is also an interesting interaction between incremental numbers and absolute numbers. Early on, you can purchase garlic armor, which reduces the amount of damage that a pet takes on every hit by 3 to a minimum of 1. It can be very powerful in the early game, as it can negate much of the damage of several attacks. However, in the late game, negating 3 damage become trivial, both because pets do so much damage, and because it is more common to get one-shot due to a high-damage attack. At some point, melon armor appears in the shop, which reduces the damage of an attack by 20, but disappears for that round after one attack. This one-time reduction ends up being much more powerful than the incremental reduction later in the game. However, any attack will proc and remove the melon armor, so a Blowfish, Hedgehog, or Dolphin might remove the armor when garlic armor would have remained. There is a similar interaction with adding additional damage, with the meat item being similar to garlic and the steak item being similar to melon.

Another part of the beauty of the design is that, although the game is very simple, there are ways for the designers to fine-tune the design by modifying different parameters about each pet. They have done a great job of this, as far as I have seen.

For example, the designers might decide that a pet's ability is too strong. One option is to change the effect-- for instance, Bats apply the weakness debuff to one pet per level of the Bat at the start of battle. This not only replaces that pet's current buff (melon armor, meat, one-up, etc) but also makes them more vulnerable to attacks. This was too powerful and everyone was taking bats all the time. To address this they reduced the effectiveness of weakness, so that the pet does not take as much additional damage. Bats are still quite good, but not as overpowered.

There may be situations, however, where the ability of a pet can't be modified. An example would be the Swan, which gives you +1 gold in the shop. There is no way to lessen this amount, as it is already the smallest increment. To nerf the Swan, then, the developers instead changed its stats from 2/3 to 1/3, making it less effective in battle, and just a bit less desirable to buy.

Pets can also be nerfed or buffed by changing their tier. As the player progresses through the game they encounter higher and higher tier pets, which are stronger and have more powerful effects. By changing a pet's tier, it can appear sooner or later in the game, making it stronger or weaker in respect to the other pets that appear alongside it. Also, there can be fewer chances or more chances for it to show up in the shop if it appears later or early.

This brings up the final way to buff or nerf a pet, which is to increase its frequency. Each pet has a certain chance to appear in the shop after its tier has been unlocked. By making some pets less likely to appear they are nerfed because it will be harder to upgrade them (by buying the same pet again and combining them) and therefore harder to unlock the higher-level abilities of that pet.

Speaking of buffing and nerfing stats and abilities, there is this really interesting balance between different ways to win a game. You can sometimes win games solely on the abilities of the pets alone, building a very conceptual team that weakens the enemy pets or does damage in unconventional ways. Or, you can use abilities and food to scale up your pets, so that even pets with abilities that do not scale into the endgame can still be useful in the endgame, simply on the basis of their stats. It also creates an interesting balance of judgment for each pet on your team-- how much are their worth for their ability, and how much are they worth for their stats, and is a new pet in the shop better on either account such that your pet should be replaced.

I really appreciate how such a simple system can give rise to so much replayability, so much complexity, and allow for so much fine-tuning, when simplicity is so often at odds with these goals.

[MicroChasm]

I wanted to write down some notes on the talk The Untapped Potential of Roguelikes. There were a lot of good points but these were the ones that stood out to me specifically.

One of the issues with roguelikes is that players have to redo the early content over and over. In order to make the early content engaging, it really has to be high quality work. One additional option is to offer earlier non-meta progression-- for instance how in Binding of Isaac you will unlock items and characters as you play even if you don't beat the game. The unlocks don't make you stronger, they just unlock more gameplay and give you the feeling of having accomplished something.

One design idea to keep in mind is limiting the access to tools. If the player always has easy access to the right tool for the job, it will be obvious how they want to go about solving the problem. However, if they do not always have access, then they have to make do with a new, different, suboptimal solution. Coming up with their own solution using suboptimal tools is more rewarding and interesting.

Cadwell emphasizes that what players want is variety, not just randomness. So if we can provide variety in any other way beyond level-gen, it will help. Examples would be additional abilities and classes.

One idea that I thought was super interesting was the idea of competing objectives. I was just thinking about this because I was playing more P1 Select recently. In P1 Select, a lot of the interest comes from balancing the two objectives of using your rings to get blue triangles using the Midas ability, and using your rings to survive until the end of the 9th level. You are basically trying to get as many triangles as you can without sacrificing so many rings that you get into an impossible situation. This kind of competing objective is super interesting. I think Into the Breach is another good example of this idea, where you want to keep your mechs alive, kills the Vec, and also preserve the buildings, and you are often forced to choose between these options. I can't think of how to implement this idea in our game at the moment.

The final point that I wanted to bring up is the idea of providing options for sacrifice. This is obviously something that The Binding of Isaac does well with Deals with the Devil. Putting yourself at risk for the sake of becoming more powerful, maybe glass-cannon-y, seems like an interesting idea. I don't know how to implement this at the moment either.

[MicroChasm]

I wanted to outline some thoughts that I've been having about the design of Sil and also of the Marble Machine, just to get them down and out of my head.

In the original Marble Machine video, Martin is playing this complicated wooden instrument that plays music using a series of repeated instructions laid out on a "programming wheel". The programming wheel is a cylinder that holds note instructions in the same way that a music box would. Martin cranks the machine to get the wheel moving, and the pips on the wheel move levers which translate to opening marble gates, which drop marbles on the xylophone and bass guitar. However, the machine does not handle the guitar chords; in other words, it doesn't both hold down the strings as well as pluck them. If the bass is playing, Martin holds down the strings himself.

Because the programming wheel is only a certain diameter, it has to play a fairly short loop. In order to create a song that is more than 30 seconds long, Martin has two options. One is that each instrument has a mute lever, so he can add and remove instruments throughout the song. The other option is that the marble gates can be opened by hand. In the Marble Machine song, somewhere in the middle, Martin powers down the machine using a killswitch lever and plays some notes by hand. I'm not sure if this is 100% necessary, but it adds to the music in a way because the physical composition of the machine affects the music that it is capable of playing.

In a recent stream, someone suggested to Martin that he use a paper programming setup, where the notes are punched out of a thin strip, most likely a large strip of plastic. He was excited about the idea that using this method would remove the need for muting levers, I think because the parchment could be fed into the machine and wouldn't need to repeat. the reaction from the audience, however, was fairly negative, because watching Martin mute channels is one of the captivating things about the Marble Machine Song video.

This is sort of the point that I want to get to, that the constraints around playing music with marbles in this way are what make the music so fascinating. You can hear the marbles moving through the machine, and the machine parts rubbing and hitting one another, and it's all sort of noisy and clunky, but it just adds to the overall experience. The machine is beautiful because you can hear the mechanisms that make the music, because it is the result of a lot of passion and hard work on a human level, and because it is a conversation and collaboration between the player and the machine. If the machine were simply a large music box, it would not be the same.

In the end, everything about the Marble Machine is so interesting because it goes through enormous hurdles in order to overcome its fundamental flaw, which is that playing music using marbles is not a good idea. Martin himself has said this, and no one would know better than him.

I think that there is an interesting parallel here to Sil, which I also think has a fundamental flaw. The issue with Sil is that its dungeon generation is not very interesting. I noticed this issue pretty early on, and I even talked to Quirk about it; especially about the idea of making new vaults. He said that making new vaults is not an easy thing to do because the level generation is finicky. I think that the game creates levels and then does a series of checks, and discards the whole level if it doesn't pass, so a new vault might just never show up. Quirk looked into creating some more interesting rooms but ended up abandoning the idea, at least for this release.

Because the dungeon generation basically only creates rectangular rooms with simple corridors, one of the major design goals of the game is to get players out of the corridors and fight in the rooms. There are a lot of ways the game does this, through the balance of monster strength to player strength, the morale mechanics, the bonus that monsters get when surrounding the player, and the abilities that the player gets to help handle crowds. It ends up being a very interesting dynamic, the decision of how and when to use corridors. For instance, I initially thought that Exchange Places was 100% necessary when playing a combat character because of how easy it is to get trapped, but Quirk told me that that is not the case, and after interacting with the game much more I can see that he was correct. I was simply misusing corridors initially because of my training from other roguelikes.

Because of the morale mechanics and the monster AI, interacting with monsters in Sil is less of a puzzle as it would be in other roguelikes, and more of a conversation between the player and the monsters. You might move foward and they move back, then you move back and they move forward. You either commit to the attack, run, or get the monster to commit. Monsters try to flank you or surround you in corridors. They run when they are scared, and then turn back to fight later. It's much more dynamic and interesting than most monsters in roguelike games, even though one of the results of this behaviour is that the AI can be annoying at times.

[MicroChasm]

I just wanted to get down the quick thought that we should think about some abilities that proc when you take damage. It could make an interesting give and take between spending health and gaining the ability's advantage.