And if this ladder can be torn off its brackets by one of those [i]things[i], will I get stranded down there?

Dead ends the player can't escape from is terrible map design.

Can I lure it towards that gas station in the previous outdoor map...

There has to be some sort of locality to puzzles. If a solution to a puzzle isn't near the puzzle itself, then you need to think long and hard about whether that's ok. Keys for example should be near locks because no one should have to back track through half a map just because they forgot to look in a particular nook. Tangentially, lock and key is a tired puzzle that needs to be put to rest. Anyways, as an extension of that I wouldn't let a player back track through the levels. Even if I always stick by this nice clean locality, the player may not trust me. He might not figure out I want him to push the rock on to the pressure switch right by the grate. He might end up searching all over hell's half acre for a none existant switch, and I'd like to save him a little bit of trouble with an absolutely unbreakable guarantee that the solution is not outside the map. 

Dead ends the player can't escape from is terrible map design.

Yah, but what if the player can create their own dead ends? 

:-P

Good post kell though! 

While we've only got a few responses here, it does seem that the direction people are headed in, is that while old games allowed multiple solution puzzles, these newer games will demand them. or maybe we could even go so far as to say that all puzzles will be dumb ass puzzles. At least when it concerns progress.

Mapping will involve a lot of knowledge - how much does an oil drum actually weigh? Which firearms work underwater? How thick must your concrete be before it will stop a 10mm explosive-tipped caseless?

But does the player know about this stuff? the point of implementing stuff like this is to create a more intuitive environment. However, for most players, the only way they would have any connection to stuff like this is through watching movies. So would it be better to not be realistic, but rather movieistic?

On a related note, how much harder or easier will it be to create a science fiction game, say on an alien world, where players have little to no intuitive understanding of how things should work? 

Yah, but what if the player can create their own dead ends?

Then like I said, that's bad map design.

So would it be better to not be realistic, but rather movieistic?

So not necessarily realistic, but believable. 

Then like I said, that's bad map design.

wrong. it's only bad map design if the mapper didn't show the player, somehow, the ramifications of his actions (or inactions) before hand.

if the mapper doesn't show the player how things work in his world, he can't be expected to be claire-voyant. it would be the mapper's fault.

if the player has seen the consequences of an action, and still does it anyway, (for whatever reason), even though the mapper went to the trouble of showing the player, then it's the player's fault, because he failed to recognize a danger.

ex: back to halflife's tentacle thing in the pit.

if the tentacle thing couldn't be heard and simply flew up out of the pit and killed the player on the spot, then that would be bad map design. how would the player know what to expect?

but as kell said, there are three instances where the player is shown the dangers inherent with tangling with that thing.

another example: Chthon. how many of you started shooting at him? that's bad map design. not a lot since you don't get killed immediatly, but it's still bad. how is the player supposed to know what to do? the player is reduced to wandering around avoiding fireballs until he can get the lightning mechanism working.

and there's also Shub. the player doesn't know that the teleporter go to wherever the teleball is. the first time i played, i teleported directly into the lava. for a long time, i kept thinking that's where the teleporter always lead, since i had never experienced a moving teledest before. finally, i was cheating and i jumped into the teleporter again. it deposited me somewhere else. eventually, i figured it out... but i never finished the level the first time i played. that's really bad map design. ;) 

But does the player know about this stuff?
Probably not - right up until they take cover from an HMG-wielding turret behind a chunk of concrete that's too thin... 

But none of your examples have anything to do with what I'm talking about. 

I'll be digesting that one for a while =)

BTW, I figured out the grenade/tele secret in The Palace of Hate long before I could actually perform the move...err sorry, why did I mention that? :/ 

Then like I said, that's bad map design.

Ok then, but at what point does giving the player more freedom to interact with their environment make it impossible to create a good map?

Again, Red Faction. By your definition, a good map must ensure that the player cannot dig themselves into a hole. In that case, you might as well remove the geomod stuff altogether.

I don't think a good map should be required to protect a player from their own stupidity... 

I think Maj got it there. Say you've got to climb out that window, but its out of reach. In the corner of the room in a washing machine/table/chair/yak. If you throw the washing machine/table/chair/yak down the stairs instead of placing it so you can climb out, you're an idiot, and you shouldn't probably even be allowed near a computer. Hell, if you just do it out of a momentary lapse of intelligence, then there's always quickload to bail you out. 

Red Faction had so little deformable terrain that it was a joke.

If for any reason the level becomes entirely unwinable, then the level should fade to black and tell you that you've fucked up. One way you could do this with a bunch of special cases; if there aren't too many cases to worry about. It's more work for the mapper but far less for the programmer. Checking to see if the player is in a pit he can't escape can't be the hardest programming task in the world. On the other hand, you could attempt a generalized system for checking if the level can still be won. That would mean some pretty hard core AI, but if you pulled it off it's a system you could probably shop around to other game companies.

Imagine two games, where in both games I dodge a pink charging buffalo which then proceeds to rip a ladder off the wall. In game one that was a scripted event, and I was meant to find an alternate route up the warehouse. In game two I was meant to protect that ladder, only now I'm screwed but the game doesn't feel like telling me that. Since I don't want to give up, I'll assume that it's game one and waste all kinds of time before I find that out that it's actually game two. Only I still won't be absolutely sure, and I'll always have that nagging doubt that I gave up too easy. 

If there is one specific thing you're supposed to protect, then it's pretty easy to have that trigger an info_endlevel entity when you failed to protect it. This has been done in games for years now and isn't the issue. 

Checking to see if the player is in a pit he can't escape can't be the hardest programming task in the world.

You'd be suprised... 

Or possibly surprised. Your choice. Just don't ask me to code it. 

From what you've said in IRC, it appears that your idea of coding is to ridicule the designers and artists until they leave you alone.

Which is a surprisingly good tactic. I'm shocked that they haven't caught on and done the same to you. 

Writing code that knows whether winning is impossible in a game is easy, in exactly those games where our beloved "emergent gameplay" isn't present. Like freecell. 

Ok then, but at what point does giving the player more freedom to interact with their environment make it impossible to create a good map?

Fascinating question. I think you can it break down like this: in every game player's have what I'll call "destructive" and "constructive" abilities. simple example in quake: say there is a shallow hole. the player has the destructive ability to fall into the hole and get stuck, but also has the constructive ability to jump and get out of it.

So we have this probability space of when the player is not stuck, which is defined by their abilities. It then becomes the mappers responsibility to make sure that their maps are always within this space. One of the major reasons for this thread, imo, is to discuss what the boundaries of that space are in a game with complex physics and corresponding abilities, so that we can make better maps.

back to the question, the point when it becomes impossible to create a good map, is when the space becomes super small, if not zero.....

(....MY GOD, what kind of psedo-intellectual bullshit is this?! Probability spaces? Somebody please take away my psilocybin)

To sum it up, if you give the player the freedom to blow huge holes in the ground, then give him the ability to fly, or climb AVP style. 

<quote>I've watched several people play Half-Life and it still amazes me how many simply charge in going "A big tentacle thing! Waargh, shoot it!" and die.</quote>

I was going to post earlier about Half-Life defying your expectations *of FPS*, not really of games in general. When you suddenly realize you can do something you took for granted you couldn't. OTOH, the only thing really new was the freedom to walk away and hear your clues fade into silence.

<quote>bad map design</quote>

It's your expectation you'll always be able to get out.. although something should happen eventually, even if it's just finding some bombs to kill yourself with.

As to Chthon, certainly don't take the rune a second time without figuring something out, hey at least it didn't matter if you saved. Shub, it's the end.

As to checking if the game is winnable, I have this unformed idea of far future basic FPS where there's an AI managing the game far more powerful than anything evident inside the game. Looks like Doom, reconfigures and rebuilds entire levels different every time and in some way responsive to player tendencies, kind of like entering the world of Tron just to race light cycles.. 

aren't about finding them unless it's RPG-ish. It's about being able to complete objectives in multiple orders and the battle each and every way. 

It's your expectation you'll always be able to get out

Until I'm told otherwise, it seems like a reasonable assumption. 

it appears that your idea of coding is to ridicule the designers and artists until they leave you alone.

How is this different from any other coder? 

all depends on the coder
wazat doesnt strike me as that at all, and most of the ones i talk to just code stuff that wont be entirely map dependant when they can help it 

most of the ones i talk to just code stuff that wont be entirely map dependant

And that's because mappers caught on and stopped working with them.