Warrens presents a fairly new problem for suggestion games: as the structure of your game world approaches that of your average video game world, the likelihood you'll encounter problems inherent in video game worlds approaches one.
For me the problem is backtracking. In a video game world, this is actually no problem, as it allows you to do some things you otherwise wouldn't have:
1. Kill old enemies for the random drops you missed.
2. Notice secrets (loot, entrances) you might have missed
In a suggestion game, these are flipped around and become problems.
1. You're running a turn based game. You cannot afford the pretension of "skill." At its most complex, your game will be a combination of resource- and risk-management, so instead of interesting obstacles, enemies are actually puzzles, and once you've solved the puzzle, throwing more of the same enemies at the readers just slows them down.
2. Who says they're going to notice? Suggestion games are all about consensus. If no one's noticed there was a Double Jump sign in room B4, you'll waste fourteen updates muddling around. I consider that a fault of the GM, though, rather than the players.
The Warrens "solution" (scare quotes) was continuity. Even when backtracking was necessary, all rooms retained their state, meaning players never got distracted from the reason they were backtracking. However, the game's constant rewarding of exploration meant that backtracking almost never got consensus (because the only time it was suggested was when we'd missed something huge in an old room that would provide rewards equal to exploring a new one).
I'd like to explore alternate solutions to backtracking, especially in a visual game that resembles Castlevania or Metroid in feel.
One thing I've considered is, instead of enforcing continuity on the game world, I could enforce it on the player. Games like this, even Metroid, experience the dilemma of choice: characters grow in power by gaining
options. Because all obstacles are puzzles, the only way to grow in power is to make the puzzles easier to solve, take away portions of that risk and resource management. This works fine for combat (although I've noticed options that work consistently tend to make people forget about the more effective but less useful options), but it does NOT work well for exploration.
Instead of simply giving the player character a double jump ability, I could give him or her a jump ability and an attack ability. Later on, after beating a boss, he can get another jump ability
at the cost of his attack ability. So it's fine to maneuver through the old rooms - backtrack through reset enemies - because he's now forced to come at them in a completely different way. For this to work, I would need to either plan the game SUPER carefully or introduce a(n) Aria/Dawn of Sorrow system where the player character can swap abilities by defeating foes or similar. Entire portions of the game world could open up just by adding another ability slot. The downside to this system is it doesn't leave too much in the way of freedom of choice for dealing with obstacles. It becomes a video game where the value of a suggestion actually decreases.