I have been playing a lot of Rogue-like RPGs recently. Basically, those are RPG dungeon crawlers with perma-death. In many of them, there are mechanics/objects that are completely unexplained or documented: you just have to experiment and find out how they work for yourself.

E.g. there is often an "Altar" that one can pray or sacrifice corpses at. The first time a player come across an altar, he might not even know that he can pray at an altar, or know that praying can result in a bad outcome.

I was thinking of adding similar things in my RPG: magical fruits that players can pluck and eat. Different fruits do different things. Some heal you, some are poisonous etc. There is no way to tell except to experiment. I was hoping this (and other mysterious objects) would give players a sense of exploration: finding out strange things and trying them out.

Is this a good idea or pointless "busy work" that frustrates player? Should I instead document the effects of each fruits in the user manual (on a website)?

I guess you may put some clues nearby the element. E.g., for your altar, you may display some engraving or a mural painting ; for the poisonus tree, some squeletton laying around, etc...

Nethack is an example of one of these games, and my experiences with it (although they were frustrating and painful for me personally) are the ones I'm going to relate to in this regard.

E.g. there is often an "Altar" that one can pray or sacrifice corpses at.

Nethack may be the very game you are referencing here, and things like this were one of my pet peeves. I should not have to go to an outside source to see what things I can do with an object. The problem with text based games is that there are occasionally a lot of actions,that players can perform, but may not be immediately obvious to the player, and thus are punished for it. (seeing as praying and sacrificing were typically good things.) Instead of simply letting their mind's wander about why this object even exists, why not let things like examine give them more information? "This altar is adorned in white marble. A small circle of blood can be seen on the stone, along with a prayer (Let's assume the player-character can read) etched in to its surface." Perhaps that's not even enough information, but it gives the idea that not only have others prayed here, but there's a reason for that circle. I'm all for 'discovering' these actions, perhaps once you've examined it will give you a list of 'obvious' things you can do with it? Instead of just blindly typing 'kneel', 'dance', 'sing a gospel song', it'll say if you do 'interact' and have examined it previously. 'pray or sacrifice' for special actions.

Edit: This ties in with what I wrote down below, but my 'white marble' altar could be useful in hiding what alignment the altar was from the player (NH had I believe Chaotic, Neutral, and Lawful). Perhaps the white implies that it's likely Lawful, could be neutral, and is very very unlikely chaotic, perhaps a 0% chance even.)

or know that praying can result in a bad outcome.

To be fair, the player in this instance IS playing a rogue-like, and it should kind of be expected. If it's not, it certainly will be. It would be interesting to build off the altar example if you were to include hints about different effects, rather than spell them out for the player. "Nearby the floor is singed, as if lightning struck indoors .How peculiar!" or "The smell cleanliness wafts from this altar." (Perhaps not a great idea, as it might imply that the altar always does those things, but hopefully my point is clear enough.)

magical fruits that players can pluck and eat. Different fruits do different things. Some heal you, some are poisonous etc. There is no way to tell except to experiment. I was hoping this (and other mysterious objects) would give players a sense of exploration: finding out strange things and trying them out.

Here's another aspect that I think Nethack did right and wrong. (or perhaps I missed an action I could have done, that's totally plausible too.) In that game, fruits were fruits and there was no difference between them. However, where a similar aspect applies is in two places. One: Scrolls. Two: Wands. Potions were another case but I think due to the nature of them they were reasonable in how you experimented (Drink or break it over something's head basically). Scrolls were enigmas. It might be a Cursed Scroll of Genocide or a Blessed Scroll of Wish, and as far as I knew, beyond having a curse / bless detection spell or item, you couldn't determine anything about the scroll before reading it. There were no implied relationships between a 'good' scroll and a 'bad' one. Perhaps a fireball spell and an icebolt spell would look very similar, and if you had read one already you'd know they were similar. Maybe this is for the best, I don't really know, but it certainly frustrated me. That game was hard enough, and having to restart 90% more often because the best 'strategy' was to drink / read everything you come across for the first 20 minutes (and inevitably drinking a potion of boil or fiery death or something) was just an un-fun mechanic.

Wands however, they did it right. Wands were generally 'good', in that typically pointing it at an enemy and zapping them with it wasn't an action you were going to regret (minus wasting a charge and letting them get closer / maul you again). However, one thing you could do with them on top of 'using' them properly was write with them. Granted, it was depicted as more dragging it around on the floor, but one could tell a lot about the wand by the way it interacted with the stone. Leaves a sear in its wake? Wand of Fiery Death. Sparks shoot out this way and that? Can probably safely bet it'll be shooting a lightning bolt at the next thing you point it at. There were lots of these effects, and some were immediate and obvious, but others were not. Wand that resists you dragging it along the ground? Wand of Force (Push Spell).

Basically I think there needs to be a way to 'experiment' without using the item proper, especially if there are deadly things that could come with using it. In your fruit example, why wouldn't there be an option to nibble? Perhaps you'd only get a vague idea of whether there was something wrong with it. "You nibble the apple. You feel a little ill but quickly shake it off. You eat the apple. Your stomach rumbles and burns! (Take damage) You are now immune to poison!" Or something like that. Why not also potentially harm and benefit the player at the same time? These experiments do not have to be all knowledge granting like in the wand example ( I think all the write effects were unique, don't quote me on that though ) but for food, perhaps a whole class of bad things makes you ill, but positive effects are left to the unknown. About to die in a fight and the apple was fine but the cherry was not? Chances are eating the mysterious apple is a safer bet, even if the cherry wouldn't necessarily kill you, and might even help more (If that's something you decided to do).

Granted, the big issue with what I just typed up...there's no punishment for experimenting. Why not nibble everything? That I can't answer at this very second because I haven't thought of a solution! But having that element of randomness I think is always a little exciting, I just hated having no way to determine what things were other than blindly using / drinking / reading them. I loved wands for this reason. It was always exciting to not look things up, but feel like I had some semblance of knowledge in the game. That is of course, after looking up that I could write with them in the first place. Figures~

Roguelikes and RPGs are very different beast. Actually, I personally find more similarities between RPGs and first person action shooters than between RPGs and roguelikes. Especially from designer's point of view (since for a player roguelike can seem like an RPG at a first glance). Do not let the identical theme misguide you, these are different genres.

So my answer would be: if you are making roguelike (tactical, resource management game with permadeath which you play only if highly focused) then yes, make many secrets and mysteries; if you are making RPG (story driven, progressive power game with load/save feature which you can play even if tired and sleepy) then clarity is a priority.

And yes, there are exceptions :D

I don't know if the majority of players prefer all mechanics to be visible, or for them to discover the effects/mechanics on their own.

But I think there's no clear right or wrong. I think this is an important design decision that will give your game a different feel depending on how you chose. If you hide mechanics, you gotta make sure they're interesting and useful to discover (like ingredient properties in oblivion), and not just pointlessly hidden (like how the #$@! do I drop inventory items in Oblivion?)

I agree with Archbishop that a player shouldn't have to leave the game to find necessary information to play. But I don't think that means that you should clearly describe all mechanics. Again, depending on how you want your game to feel, you could hide stuff like effects at praying at alters or potions or scrolls.

But if you do, you can also prevent these from being clearly documented on some wiki by introducing a randomized element each time a game starts. So on one playthrough, red potions will heal you, while on another playthrough, red potions might curse you with blood curse and lower your hp. If you can mix and match elements with effects, you can keep the feeling of unknown danger present each playthrough, forcing players to go through discovering things on their own each time, making it an integral part of your game. This should render most outside documentation useless - but again, it will give your game a totally different feel.

Like I said - I think that there's no right choice here - but if you pick to hide mechanics, it makes discovery a gameplay element, so you should do your best to make sure its fun and has a purpose.

It might be worth drawing a distinction between 'mystery', and 'lethal mystery'...

I'm all for letting players discover effects/mechanics on their own, but if many of those effects are lethal (as is the case in many rogue-likes), then you are basically making experimentation a core mechanic, and simultaneously punishing the player for that experimentation.

undocumented items are essentially puzzles. handle them as you would any other puzzle in a game. not too hard, not too easy, not a show stopper if you can't get it, etc.

in this context, puzzles are a form of minigame within the larger RPG game. design accordingly.

once you start thinking of undocumented items as puzzle minigames within your RPG, how to best use them should become pretty obvious.

and they definitely sound like a cool way to enhance a RPG, just don't make them show stoppers.

Funny you should post that.

A few days ago, I was pondering at unexplained gameplay, and how it contributed to my fond memories of retro-games.

While this can cause frustration / friction for the players, and may label it as hardcore, I find value in this type of gameplay.

It might be even better if you include an unexplained feature that is purely cosmetic (being able to alter the color of something in the world when interacting, without there ever be any gameplay purpose for this). This would keep players wondering, and the mere existence of something that serves no purpose, and doesn't advertise itself as interactable, will make the players interact with your game differently, and seek the un-obvious.

Just like the mere prospect of perma-death makes you double-check everything, so does the idea of not advertising interactive elements, or not give away their use.

One note however. In Ultima IV, I found that the colored potions which didn't advertise their use was on the frustrating side: it really didn't bring much in terms of player exploration, but it did end up being quite boring (ah, it was poison then...).

I'd restrict it to elements in the environment that could serve a purpose.

As a matter of fact, you could make it so that door and key coexist in the same room, but that you'd either:

1 - Reward the exploring player for trying combination of items, and suddenly unlocking the door (that would probably feel very rewarding to get a "click" after tinkering for minutes/hours with random pieces of furniture, rocks, etc).

2 - Have an NPC related to a quest to help you determine what the "key" looks like, so that, if you have this knowledge when coming in, you can "solve" it much more easily, without need for exploration

In this case, you'd cater to two very different crowds with little effort.

Unfortunately, the Internet age has made this type of experimentation a lot less common. Remember when game help hotlines were popular? It's crazy to think that people used to pay to call a help line to get past a difficult part in a game. (I admit that I did it a few times. Damn you Quest for Glory!)

I have a feeling that rather than trying things, a majority of players will just google it and instantly know which fruit is dangerous/beneficial. To combat this, you could do what some roguelikes do and change the properties of potions each game (ie - the blue potion might heal, poison, cloak, etc on different play-throughs)

This might require a little more planning and development time on your part though.

You should put the purpose of the item underneath the name of the item. That way, the player can conveniently understand this correspondence.

It can go both ways though.

If the player does not know what the item does, you will generate a buzz or a discussion in regards to said item in a forum thus increasing the exposure of your game. This can also lead to players' creativity in your game.

And the other hand, this can also backfire and frustrate the player which leads to a negative aspect of your game.