In Defense of Perception Checks
The zeitgeist in OSR/NuSR spaces these days seems to be opposed to Perception mechanics; in other words, having a skill or roll to notice things, like secret doors or traps in a dungeon. Here's what Knave 2E, Shadowdark, and Into the Odd have to say:
Knave 2E
SEARCH CHECKS: Hidden things are either automatically discovered after enough time is spent searching (usually ten minutes for dungeon rooms or a full watch for wilderness hexes) or they cannot be found without taking in-game actions. Obvious features of an area should be described right away to PCs, and details should be described as players ask questions and investigate.
This approach rejects random chance when it comes to searching: either you find it or you don't. I find it a little odd that the rules don't tell the GM what hidden things in a module might be automatically discovered and what might be gatekept behind specific in-game procedures, though the gameplay example at the end of the book suggests that PCs should automatically discover "tells" for things like secret doors. The guidance to spend 10 minutes per "room" is also strange, considering rooms can vary substantially in size.
Shadowdark
A thief always finds a trap if searching in the right area.
If you take the time to scan the sky for threats or examine a stretch of wall for a secret door, you simply succeed.
Later in the book, in the section on traps:
Characters who search a specific area or object for a trap automatically find it.
Like Knave, Shadowdark's searching are very permissive, though one thing that strikes me is (as far as I can tell) it doesn't actually specify how long it takes to search a given area.
Into the Odd
As a general rule the presence of a trap or other hazard is always noticed by characters unless they are running, visually impaired or distracted. After this the characters may be harmed through further inaction or lack of caution. Most Hazards should allow a DEX Save to avoid danger at the last second.
Similarly, Into the Odd suggests giving the players as much information as possible when it comes to hidden things, at least in the form of traps. This is backed up in the Remastered rulebook by the sample adventure, The Iron Coral, where hidden things are always marked with some kind of clue or tell that is immediately apparent to the PCs.
Compare and Contrast
Though I'm going to be gently arguing against the approach these games take, I don't think it's necessarily a bad approach, and I have nothing against these games. In fact, Into the Odd is one of my favorite OSR games. Nevertheless, I think it's interesting that these games, which are frequently used to introduce new people into the OSR playstyle, are quite averse to a more traditional method of discovering secrets.
Old-school games, like early editions of D&D, and their more direct retroclones1, have more mechanized, procedural takes on exploration. A good overview of these different approaches is Beau Rancourt's A Survey of Searching for Secret Doors. Generally, most OSR games adopt OD&D or B/X's approach to searching, in which a character can spend a (10-minute) exploration turn to search a single 10' x 10' area with a small chance of success, while Elves have a chance to passively perceive any secret doors. Note that a 10' x 10' area is quite small (just one square on most dungeon maps), meaning searching an area is impractical unless the players are already fairly certain that there is something hidden there. Only in AD&D, where searches take 1/10 as long, is searching an entire room using the searching rules actually feasible.
I think this bit of clunkiness when it comes to finding hidden things, at least in the B/X and OD&D-derived rulesets, is why a lot of GMs' handle hidden things quite differently in practice from what the actual text of these games would seem to suggest. I think there are basically two ways to give information about hidden things without using some kind of random mechanic:
High-Information
This is more or less what Knave, Shadowdark, and Into the Odd suggest, giving the PCs a lot of information up front, and then seeing how the PCs deal with that information. I think this is a reasonable approach, and works well if you're designing a dungeon around this approach to information delivery. I think of this approach as the "Do You Stick Your Head in the Statue's Mouth?" approach because it becomes difficult to actually surprise the PCs with a trap or secret door: most ostensibly hidden things become invitations for PCs to interact with a bit of dungeon dressing. It can also be difficult to use this approach with classic OSR modules, because a lot of hidden things in old-school adventures are just plain doors or trapdoors in innocuous sections of floor or wall. In my opinion, this is kind of a shame because one of the best things about the OSR is that it's a kind of "lingua franca" that translates well across many different systems.
Low-Information
I mostly see people talking about this approach anecdotally, and in fairness my distaste for this style may have more to do with my own lack of familiarity with it. Basically, this style starts by giving no or little information to the PCs, and since there are no procedures for revealing hidden things, the PCs must describe in natural language what do to stumble upon secrets. I could see Knave operating a bit like this if the GM was very conservative on what they straight-up revealed to the players. I don't like this system because in practice I think it devolves into a game of "Magic Words": once the players figure out the "magic words," something along the lines of "I tap everything with a ten-foot pole," this approach becomes basically identical to the High-Information approach, which means this style has all of the flaws of the High-Information plus being a bit more annoying for the players.
Randomness
In my mind, the conflict is about how to "play fair" with the players. Like in combat, the GM should be able to present challenges, such as hidden pit traps, without either trivializing them (i.e. saying "you see a pit trap, do you step into it?") or harming the PCs without giving them a chance to avoid the danger (i.e. the PCs walk through a section of the map and die because they stepped on a pit trap they had no possibility of ever seeing).
The alternative to the high- or low-information approaches is to introduce randomness, meaning the same search can produce different outcomes based on the searcher's stats and the roll of the dice. You can go back to B/X or OD&D, but I do think there are some serious flaws in these classic rulesets which games like Knave and Shadowdark are trying to address. In B/X and OD&D, searching a small area takes quite a long time and has a low chance of success. AD&D expands the area searched, which I think is a good start.
Plunderer!
I want to talk about how I'm handling hidden things in my game Plunderer!, not to suggest that I have all the answers but because my goal with the game is to combine all of the things I like in OSR games into one ruleset.
My approach in Plunderer! is to try to fold discovery into the other systems, specifically the freeform skills of the Expert class and the encumbrance system. By having a bunch of different, relatively simple systems which nonetheless affect each other, I hope to make the experience of exploring a dungeon a series of strategic decisions made with incomplete information.
- First, I keep the ten-minute Turn and the chance of encountering wandering monsters. PCs in Plunderer! don't have to worry about torches or rations, but still have to worry about running into nasty creatures, and this means there is a time pressure component.
- Inspired by AD&D, in which a PC can search ten 10' x 10' squares in ten minutes, PCs in Plunderer! can search up to twelve squares if unencumbered, with their movement speed directly corresponding to their search area. This means the more things a PC is carrying, the longer it will take to search a given area.
- Discovering things is a WIL test, meaning different characters will be better or worse at searching, and discovery is a possible skill investment for the Expert, which gives Experts more potential options when it comes to spending their Skill Points each level.
- I separated discovery into active (examining an area) and passive (spotting things nearby) because of some problems specifying how the Elf's ability would work. Now, all characters can spot things passively, but the difficult is quite high unless you have the Elf's high WIL and bonus to discovery tests.
- Of course, if a PC does something that would reveal a secret, they simply reveal that secret regardless of whether they rolled well or not.
I'm happy with all these interlocking mechanics, though there are two things which still bother me:
- Making discovery a 2d6 test and expanding passive searches to all characters means there is a lot more dice rolling than before. While I tend to play online on a VTT and can quickly throw and calculate fistfuls of d6s that way, I worry how this would affect playing in person with physical dice.
- Even with a 1-in-6 chance of random encounters per turn I don't know how much time pressure the PCs will actually feel, compared to the threat of straight-up dying falling down a hole.
In any case, I'll be trying this new version of the system with playtesters soon and hope it'll yield some good insights.
Footnotes
Retroclones are games which are meant to be identical or nearly identical to another ruleset, as a kind of admitted plagiarism. In the 2000s, people wanted to play and make adventures for early D&D editions but the rulebooks were hard to find and still protected by copyright, so people made nearly-identical copies of the ruleset and released them under very permissive licenses so others could play and create for old-school D&D without legal trouble.↩