I think it is important to have aspects of combat and navigation that are unpredictable. However I think you can make a completely predictable map that still plays solidly and is enjoyable to the player. The authors who can work surprises into their maps without having them feel too gimmicky or out of place are the best authors.

A good example of doing it poorly would be in my 1000-brush Norway SP map from last year. If the player missed an invisible "leap of faith" they were teleported to this "hell room." I'm sure most of the players were just expecting to fall to the bottom of the area and take 5 damage, as there was no visible teleport there.

In retrospect I shouldn't have tried to work in such a gimmicky piece of gameplay but such are the risks you take in trying to make compelling gameplay :) 

are you talking about predictable to the level designer or predictable to the player?

Predictable to the level designer means levels on a rail -- no route choices, very regulated and scripted gameplay, and lots of measures taken to keep players from breaking the gameplay.

Predictable from a player standpoint means learning how different classes of things behave, and then being able to build strategies around them. In other words, "systems-based" design. Of course level-on-a-rail can be predictable to the player as well. 

1. Perfect level symmetry where everything is so nice and clean that I can predict what is around every corner.

2. Redundant placement of monsters. Like push a button, see some monsters. Repeat. Open a door, see some monsters. Repeat.

3. End of map climax with hordes and hordes of monsters, focusing on quantity not quality.

Things that are nice to see when they happen.

1. Monsters with walkpaths, effectively randomizing their position or occasionally surprising a player because they "weren't there" and then see the player from behind when they come around a corner.

2. Moderately complicated situations where the type of monster and the environment creates the challenge. Very small area plus Shambler/Hell Knight, area with no cover and Vore, darkness plus some scatter grunts, zombie plus no gib weapon.

3. Rocket Jump/Grenade Jump! Is Once Upon Atrocity solvable without a grenade jump? I've never found the way, if so. (I mean without the secret RL, I wish I never found that on accident).

Also in Trinca's Forgotten Tomb, I found myself surrounded by knights. I only had rockets and decided my only option was to rocket jump over them, as I had no other exit opportunity.

4. Unanticipated/unconventional geometry. Tunnels, confusing hallways!, maze-like structures where it hard to tell where you are (ac.bsp, bnt.bsp, etc.)

5. Water! (ant.bsp)

6. Slime! Why do we see so little of this and so much lava???

7. Unintuitive shortcuts or map navigation. Anyone else *really* like that area on Insomnia where you encounter the difficult, but climbable area or the part of -- is it czg01 -- where you have to make a jump across lava rocks and then later a complicated jump from the top of a box to a ceiling tile (hard!)

8. The monster that scares you. Can be a simple as a dog awkwardly placed. 

interesting comments.

RE: #2. John, i mean predictable to the player i guess. 

What metl said, systems-based.

Predictability is what gives the player 'rules' to play within. Monster behavior in Quake is an example of that. Once you can predict what monsters will do you can exploit their behaviors to be better at the game. In Quake you can, in theory, handle a room full of shamblers and fiends and vores and shit, because you know what they're all going to do.

Once you, the designer, have those rules set up, you can play off them, and do something new and, arguably, "unpredictable" instead of just re-testing the player on the same rules. You want the player to start predicting things, because once you know what he's predicting you can interrupt him, and make his predictions wrong.

The one moment in Doom3 that scared me was actually not a planned scare. I had stopped and backtracked for health and missed an Imp somewhere, and he was just hanging out behind a door on my return. Now, doom had a lot of imps behind doors, and the unfortunate effect was that "doors have pre-placed imps behind them" became one of the rules the player learned. Who else started playing the game by edging up to doors from the side until they brushed the door trigger, specifically to avoid that stupid repeated ambush?

The reason it worked on me that one time is that I knew I had cleared the door already. It was a "safe" door, and thus not covered by the rule. 

Monster behavior in Quake is an example of that. Once you can predict what monsters will do you can exploit their behaviors to be better at the game.

Yes, i agree. I find this monster predictability in quake 1 so fascinating because it seems so perfectly balanced, this seems to me one of the primary reasons quake 1 gameplay is so good.

Also, watching speedrunning in quake 1, the way they can predict everything to such a fine degree in an artifical 3D environment is amazing, i think. 

Blitz' comments kind of hit home for me. In almost all of my maps I go teleport crazy in an attempt to create unpredictable situations. In a way it works, but it gets old. I think the reason that many mappers use lava - at least the reason why I do - is largely because of the amount of control a mapper can wield with it. Slime is too survivable when you want to keep the player focused... does that make sense?
In SMs I think it's less shameful to have that kind of rails gameplay, especially with teleports, etc. I think a great example of unpredictable gameplay was Ant.bsp. The monsters were placed oh so perfectly (especially the shamblers), and it was never over the top. 

... is when the player goes "duh! i should have seen that coming." Which ties into what Lunaran says.

And uwf, speedrunning is a mix of prediction, practice and hope :-) You don't see the demos where the ogre stubbornly refused to get out of the way... 

I�m working on something atm thats very loose in play style, the player has to wander around alot, but I�m trying to keep play moving forward without them getting lost / bored / frustrated, just because it�s something new to do. Seems like all mappers want to create very strict A to B maps, and whilst this has to be the case it needn�t rule always.

And the only reason to experiment with this is because its rarely done well, but very good when it is.

I only just bought hl2 and find it kind of irritating that there�s no real wandering around sections - you always feel funneled along a path, something that the first hl removed from play, and one of the reasons it was so good. 

I think the main issue with open maps is that there needs to be a close connection between weapon progression and monster progression. In a linear map, this is easy to maintain, becuase you know when players will get a weapon and when they will meet a monster.

In a standalone map, the mapper often gives the players all the weapons in the course of the map. This means the progression is steeper.

In a map that is part of an episode or an entire game, you will progress much more slowly through the weapons. Becuase of this, it's a lot easier to have open, nonlinear layouts becuase you might spend most of the map without getting any new weapons.

Some of the Doom2 maps were very open and this worked quite well. 

The map I am making at the moment is partially linear but contains route choices. These basically amount to which route you choose first, or which way around a loop you travel, though I have given the player the choice taking one weapon instead of another also. I'm hoping that this will allow many players to get a different experience from the same map. Perhaps I could even implement difficulty by having an easy/normal/hard route through the level, though I think I'll just stick with the traditional method for now. 

in a custom level the player is assumed to have already played the original game, and so knows what weapons exist in the toolset. In the original game you have levels (eg. in q1) where the player has only shotgun / nailgun, no rockets or lightining at all - imagine these being left out of even the first two-three levels of a custom episode; it wouldn't happen because it'd likely piss people who played it off. You can do it ni the original game because everything is new to the plaer, in a custom level its assumed to be at least familiar, if not expected. 

 

I think it's the point that many quake1 mappers forget about. Environment interaction as a gameplay element was well used in many modern (and not so modern) games. It makes another experience to the player than just go-kill-monster-press-a-button-kill-another-monster gameplay. It may be called unpredictable for player in quake imho.