This is actually a subject I've thought about a lot in the past. The idea of making a state graph for a maze game is an interesting one to me.

First, you can use the state graph to programatically prove that the maze you've created is solvable. One of the advantages of this is you could have the computer randomly generate a bunch of mazes, and guarantee that none of them are impossible.

Second, you could use properties of the graph to determine how "hard" or "easy" a maze is, or other properties. This can't replace the subjective human experience, but combined with your own experience, you could come up with some rules for what makes a maze hard/easy and fun/tedious. For example a graph could have many choices from each state, and a very short path to the goal, or very few choices at each state, and a very long path to the goal. Also, are all moves reversible, or are some moves one-way only? Are there any unwinable states that you can get into? Are there a lot of loops that lead back to previous states, or do most wrong moves lead to dead-ends?

One more thing to note about this game is that there is almost always a move from each state that kills you. These could instead have been illegal moves that you were prevented from doing, which would not change the complexity of the maze but would change the feel of the gameplay (removing the threat of pushing the button at the wrong time.) I think the choice to allow falling into the void is good, becuase it fits better with the loosely-physics-based theme of the puzzles (all rules except for the teleportation switch are based on gravity and the weight of the block.)

Making rulesets that roughly correspond to players' real-world experience is a good practice for games and makes them more accessible. 

The next level would be to randomly generate a solvable maze and have the player design an NPC and threw it in there, then watch how it fares.

The CPU would handle all the avatar's 'thinking', throw in a few enemies and maybe the option to continually create NPC's as the first tries to solve the maze and you've got game.

What happens if two of the NPC's meet? Depends again on how the player created them.

That's a world away from the topic, but I can see it working, though maybe a bit too D&D for mainstream (who don't like tweaking curiosity vs. aggression stats, usually).

With enough variables it could work, but the concept would need nailing down so as not to go off on a tangent during production. 

 

The next level would be to randomly generate a solvable maze and have the player design an NPC and threw it in there, then watch how it fares.

That's a cool idea, though i'm not sure that a maze is the best choice of environment. A more open environment with obstacles sort of spread around might be better.

Actually, it would be cool to see an RTS where you design your own units. I guess you'd have to design them before the battle started. In this type of game, I would make the cost of a unit based on the complexity of its ai program, so players would have an incentive to get the best results out of the simplest AIs.

This would also mean that even if you had a super-smart AI that could win the round for you (like he knows the exact route to the enemy base or something) he would be so expensive that you'd have to wait until rather late to build him, and you'd be vulernable in the meantime. 

game where you designed units was Master of Orion.
That was fun!
The AI designed spaceships too.

Of course, like with many games, there were somewhat clear optimal solutions. And when you have some experience playing the game, they become obvious and thus remove any decision making from the game and make it just boring. (Have I talked about the CIV3 settlers already?.)

That's a thing most game designers haven't realized - you have to keep the player making decisions that are not obvious. (At least in non-twitch games.) 

It's hard to make a game that is not random (rock, paper, scissors) and not obvious.
Chess is one - there is no one sure tactic to win. 

That's why I said about having more than one NPC trying to get through the maze - maybe when they meet each other they could fight, get in each others way (forcing a different route) just ignore each other or try and help an injured comerade - all depending on how they'd been built.

This still might be boring - drop only one guy in there and he'd eventually finish.

But how about multiplayer - maybe four different people dropping thier creations into a maze.

Chess is rigidly structured, it's the other player with thier behavioural input (not least your own as well) that gives any one game randomness.

In this game idea that'd come from how many characters a player dropped into a maze and how they built them. 

metlslime knows it all.

been playing Flatout2 with coworkers, great fun 

nice. camera's a bit wonky but I really liek this. 19% in according to the save game. 

I think that's what bothers me with Picross DS, it's all about being methodical and you never really have to change strategy. 

Is excellent. The sequels are pretty good as well. 

yeah have them too, along with like 15 other games I need to play :) 

Is one of my favorite games ever (so is the original PoP.) The puzzles, the level design, the artwork, the storytelling, writing, voice acting, are all excellent. The combat is the weakest part of it IMO, and even that is pretty good.

Never got either sequel because I heard so much about how they were "ruined" on internet forums. Maybe I should try them. 

But more gamey. But they kept the original play and expanded on where it lacked as well as occassionally spoiling stuff.

But the bosses in the second one were great and the fighting much more entertaining. 

 

puzzles and level design are great, I originally thought the puzzles would shit me given the lack of saving but its implemented really well, and that rewind dagger thing works quite effectively.

combat is a fair bit of button mashing for me, but it does look cool. 

Got my Dark Messiah, played a bit, is good. Seems a bit harder than the demo or maybe I practised in the demo a bit more. Nice proper fantasy style. Not played much so far.

Been playing a bit more Guild Wars: Nightfall, well as in just started it, as a Dervish. Knowing my steez about GW I've monched on up to level 7 fairly quickly. Quite good but the general setting is not that rad - give me the Shiverpeaks anyday!! Absolutely loads of new options and tweaks which is kinda cool. I like the one that pops up saying "You have been playing for 3 hours. Please take a break". Well, I did.

That is all. 

sands fo time. Defnitely a great game. Camera's bit wonky like I said and the fights get tedious. I think its because there's no combat progression, the first fights are so easy you never learn how to fight properly and then when you get to difficut ones, you have to learn a technique there.

envinments are fantastic.

Painkiller is next. 

I think its because there's no combat progression, the first fights are so easy you never learn how to fight properly and then when you get to difficut ones, you have to learn a technique there.

Yep, that's a good description of what's wrong with the combat. I never stopped button mashing until like 90% into it, when you have to suddenly learn how to block every attack to survive. 

Is great as well. The bosses are a bit stupid and random, and the expansion has some terrible map designs (forcing the player to know about bunny-hopping or giving no clue at all how to proceed). But its still great, some very nice artwork, combat and physics.

I considered switching from q1 mapping to Painkiller. 

so far, painkiller is the most 2d game i've played in some time (not counting the doom reinstall I did a few days back).

all the gihting is on eye level, wit hardly any 3d combat. Just 5 levels in. And yes, the two bosses so far have been totally random. The one you couldnt kill by firing had blood when yous hot him and the one you could kill by firing didnt show any damage. 

I gave up on that game after quite a few levels/bosses because it was just too damn repetitive and boring. The actual base combat is pretty fun (the weapons feel good to use and the enemy reactions are of course amusing), but the levels were just so damn unimaginative and plain that there was almost no game there.

Considering that the game has pretty decent combat/weapons and some interesting monster types that could potentially be used to good effect, its a real shame that all they bothered to do with the levels was basically this:

- Player enters box room (completely flat with no height variation)
- Doors close ahead of and behind player (that is, "hello, I am a generic gameplay device" doors that don't match the art of the level)
- Hordes of monsters spawn in waves and run directly at the player
- Player kills all monsters, doors open again
- Repeat x 89738924682638479287643972

The levels of course look nice (apart from the lack of 3d-ness in most areas). More so than any other popular game I've played, it seems quite obvious to me that they had a bunch of artists building the levels (possibly from game desginer sketches/notes... or not) rather than experienced level designers/builders. 

so far agree 100% with what you say, although I also have a beef with the progression in some levels, but I'm not bored yet :) 

But there are some real gems inside there as well; the venice level, hell, some of the base sections and the colliseum from the expansion.

It is just horde combats, but the game never pretends to do anything else - as complex as it gets is the gold / silver cards management.

That's what I liked about it; you're dropped into the game right away. 

Tend to agree with the above comments, but I did like Painkiller, the style and stuff made it worthwhile, and although I'm not a big horde fan, as that sort of stuff goes, it was fine.