I've had a though inspired by Rob Kuntz over at Lord of the Green Dragons and the indomitable James Maliszewski at Grognardia.
One of the benefits of a well run old school game is that it requires more active effort on the players' part to interact with the game environment. For instance, if the characters come across a pool of water in a dungeon room, modern and old school D&D take two different approaches:
Modern D&D: The characters make Perception checks. On a success, they notice the skeleton deep in the pool, half-buried in the mud.
Classic D&D: The characters haul out a 10 foot pool, poke around at the base of the pool, and dislodge the skeleton.
In the first case, the players apply the game system to the environment.
In the second case, the players use their experience/skill, along with an assessment of the environment, to come up with a plan of action.
On the face of it, the second case is more immersive and engaging, but I think the first approach does have its benefits. It's faster, and it establishes a protocol of sorts between the players and the DMs.
When a player says, "I make a Perception check to inspect the pool" the intended outcome is clear. If there's anything weird in the pool, the player wants to know. The old school approach has the potential to short circuit that. Perhaps the DM expects a specific action (dredging the pool's bottom, as opposed to poking at it with a 10 foot pole) to uncover something of interest.
IME, that disconnect can undermine an entire game session, with the players stuck Zorking an object or NPC until they hit the specific action the DM is looking for. The skill system helps avoid that by creating an abstract level between intent and action.
I think a middle ground approach nets you the benefits of both new and old school gaming, but it takes a little work for a DM. Here's what I propose: the players make a Perception check, but success only points them to their next action, an action that requires them to approach the situation in a more immersive, engaged manner.
Here's an example:
The players enter a room with shelves of books along its walls. The players make Perception checks to search the room. One of the players hits the DC. What happens next?
Modern D&D: The characters find a secret cache of gems.
Hybrid D&D: The characters notice that four of the books on the far right shelf are clean, while the rest of the books are covered in dust. It's still up to the players to figure out what that means (there's a secret compartment in the wall behind the books).
If the characters poke at the books in greater detail, you can continue to make skill checks but only in response to the PCs' actions. For instance, one of the characters removes the books and flips through them, looking for clues. A History check reveals they are relatively new volumes on herbalism. Another PC pokes at the bookshelf, and a Perception check reveals the hollow space. Finally, a Thievery check allows a character to inspect the wall and points to how to open the space.
The overriding idea is that the PCs' skill checks give them information in proportion to the resolution level, as in screen resolution, of that check. Broad checks give broad information, while specific checks give specific data or overcome obstacles.
Now, like a lot of techniques this one has drawbacks. A bad DM can use it to hen peck the players with endless skill checks, and it can lead to some drawn out scenes. However, used correctly I think it preserves the strengths of a skill system while allowing for more immersion and more direct, concrete interaction between the players and the game world.
The important thing, though, is that this method forces the DM to insert detail into the space between "I make a skill check" and "This is the end result".
Lovecraft was right
1 day ago