D&D 5E A case where the 'can try everything' dogma could be a problem

[Agamon]

I wouldn't necessarily classify that example as failing forward. Eventual success (bashing down the door) was inevitable, given the task was possible. Failure (making a lot of noise) only occurred if the first check failed. You're not forcing the story ahead despite a failed attempt; you're enforcing consequences directly related to the PC's actions.

While the result of the check has to do with noise or lack thereof, the check is actually to open the door. Binary would be: succeed - door opens, fail - door doesn't open.

Fair forward, at least from what I've read about it in DW and 13A, is about the the PC still doing what they wanted, despite a failed roll, but some other obstacle occurs as a consequence. It may not be the awkward example from DW where the PC succeeds his climb but griffons attack, but I think it is similar.

But that's semantics, I guess.

 

[Agamon]

Out of interest, then, do you use passive Perception and Insight?

That's kinda, sorta what's going on. Like a highly perceptive PC will spot something without rolling, the knowledgeable PC will know the answer to a question without rolling.

 

[iserith]

Out of interest, then, do you use passive Perception and Insight?

I use passive checks when a player has established that the character is doing a particular thing with an uncertain outcome constantly over time. I don't treat passive Perception as "always-on radar." You have to have established that you're keeping an eye out for hidden threats* or are searching for secret doors while moving around or whatever. The trade-off here is the resource of time and not being able to engage in other tasks at the same time (unless you're a ranger in favored terrain) in exchange for never rolling below a 10.

Passive Insight doesn't come up much. This is generally a normal ability check, the result of which is commonly that the character can or cannot determine the truthfulness of the person he or she is observing. Sometimes I'll reveal the NPC's truthfulness on success or failure, but with a complication like "She knows that you know she's lying..." or "He goes from Indifferent to Hostile because he thinks you're grilling him." Something along those lines that works in context. I never tell a player what his or her character thinks or believes though. I only do or do not reveal truthfulness. What the player chooses to have his or her character think or believe or do at that point is up to the player.

I can see passive Insight come up in an extended scene as something of an abstraction. For example, the party is at a masquerade ball and one PC hangs out by the punch bowl to observe the goings on over time and see which attendee is really Lord Vampire or whatever. Or the steward allows a PC to secretly watch the boy-king's advisers conduct business in council to try and determine who has the crown's best interests at heart. Something like that.

I show how I adjudicate a lot of this stuff, including passive checks in my thread on How to Adjudicate Actions in D&D 5e.

* I consider keeping an eye out for hidden threats to be the default activity of skilled adventurers unless they've established themselves as doing something else that is reasonably distracting.

 

[neobolts]

That's all well and good, but very often it's obvious not just that there is need for a check but also what type of check is called for. In such cases, I'd be just as happy for the player to jump straight to "I want to make a Religion check" (and, indeed, just rolling the die and telling me the result) as to go through the "What do I know about this icon we've found"/"Okay, roll a Religion check"/"17" cycle each time. It's just more efficient.

I agree. We are pretty casual and don't worry about this at our table either. An in-character "I think back on my XYZ studies and inquire if there is a library nearby" or an OOC "I make a XYZ check" are one and the same, especially if its a fairly routine check. If there's something special going on or some real gravity to the check, then I'll prompt the player for a more detailed in-character response.

It's really a matter of your group.

 

[Sword of Spirit]

This doesn't require secret backstory, however. All it requires is GM authority over backstory authorship. But the GM can actually perform that authorship in real time, at the gaming table.

Not really. Because if the DM inserts elements in the moment, you are no longer dealing with discovering a shared world--you are discovering what's going on in the DMs head.

I'm not unfamiliar with this style of play - I've been playing D&D since 1982, and have read plenty of 2nd ed and 3E modules which emphasise the playstyle you describe. But I don't entirely agree with your description of it. For instance, the GM is not discovering the world - s/he is authoring it (or, in the case of a sourcebook, s/he has already read it).

It's true that if the world is, literally, randomly generated then the GM discovers at the same time as the players. This is a style that Classic Traveller emphasises to a much greater extent than D&D. But very little D&D play, particularly in the present day, involves literally random content generation. It is mostly pre-authored by the GM or the module writer.

1e & 2e had about the right degree of random tables for me.

One end of the spectrum:

The DM comes to the table with no world, no story, no statblocks.

One player says, "So my character is from a farm. Where do we..."
DM: The mud of the trail that you leave unconcernedly on your shoes as you enter the Hall of Lords contrasts with the emaculate dress of Lady Howenot, who turns and looks upon your character with disdain.
Player: Why am I here?
DM: That's for you to decide.
Player: "Fair lady, I'm searching for the Riddle of Mentios, perhaps you have heard of it?"
DM: "Are you from Mentios then? I had not thought a peasant from the land of milk cows and beans would have traveled so far only to muddy the elven-made floors of this hall."
Player: "Yay, I have traveled far in search of the Riddle, to save the kingdom of Vangard, for I am the lost heir to its Emerald Crown."
DM: "Interesting. Perhaps I will help you, for a time."

In this version, everything was made up on the fly by the player and DM. Sort of that "yes, and" thing taken to a world-building extreme. This may or may not be about a story. The defining element is that world-features are defined by player/DM declaration on the spot, in response to what is happening at the time, to accomplish specific ends of the players/DMs. They aren't exploring a world, they are creating it.

The other end of the spectrum:

The DM is playing in his version of the the Forgotten Realms. He's made a few general changes, and filled in some details that published material doesn't include. He also has a robust set of random tables to roll on.

DM: You sit in the pleasant, if rather bare, dining room of the Shining Serpent Inn in the city of Neverwinter. The delightful aroma of baking bread makes your mouth water, while the warm, fluffy guest robe that you wear, gives you a feeling of relaxing at home. As you wait for your breakfast, the simple comfort of the morning distracts your mind from the purpose of your visit--investigating rumors of smugglers based out of the Inn."
Player: Is there anything special about this robe? Since the robe feels warm, does that mean that it's winter?"
DM: Other than the embroidered depiction of a silver serpent, marking the robe as inn property, nothing stands out as unusual about it. It is mid-Autumn rather than winter."
Player: I'll wait for my breakfast. Who else is in the room?
DM: [Rolls on table] A dwarf, and 3 human travelers sit at tables in the room. The waitress arrives with warm bread and the inn's signature eel pie. "5 gold pieces. Is there anything else I can get for you?"
Player: "No, that will do for now. I did have a question..."

Here the DM is taking everything straight out of Volo's Guide to the North, pages 140-142. That's what the inn is like, and the player gets to explore it in the moment, while the DM began exploring it when he read up on it before hand. The other patrons in the room were determined by a random roll on an appropriate table (not found in Volo's Guide, if anyone is following along at home). The random roll means that player and DM alike get to discover who is in the room, allowing the DM's exploration of the world to continue. If the player continues his investigation, he might interact with smugglers that the DM created (either ahead of time or by random rolls).

The difference is that in the first one the world doesn't exist until the player and DM create it. In the second case, all the creating of the world has already been done. Because the DM isn't a supercomputer who can design every bystander and candlestick ahead of time, he relies on random rolls to fill in that information. Random rolls mimic interacting with a third entity, neither player nor DM, and which in context of the game is the world. The creation of the random tables is part of the creation of the world that is undertaken before the game begins. Even had the DM whipped up a random "table" on the spot, assuming he made it appropriate to the world, rather than to a story idea in his head, it would still be representative of the third entity, rather than of particular story goals of the DM.

The key element is that with exploration, you are discovering and interacting with a third entity--the world. You and DM visit the world. With creation, you are interacting with the DM's response to your own immediate choices. You visit the DM's creative thoughts.

Now, most of the time no one is going to play at either extreme. On the first end of the specture, generally you have some idea of what story you want to tell, and probably have at least conceptually established who some NPCs are and what events or locations may be relevant. In the second type of game, you might give the waitress a personality that isn't rolled on a table, and decide that at one of the tables sits a bard from a significant bardic school that you plan to have an ongoing presence in the campaign. But the important distinction remains.

I'm not sure how to express it any more clearly. I'm not a music afficionado. If someone introduces me to the differences between a couple of subgenres of a particular type of music, which I can understand and discern, and then expresses how this is such a big deal, I'm likely to not appreciate the difference. It just doesn't seem to really be a meaningful thing to me. It's simply a matter of appreciation. I can however, understand that, based on the number of these crazy music nit-pickers, there is something there.

So that's what I'm saying. There is something there. There are plenty of us who feel that it matters. There's nothing wrong with not appreciating that difference, but it's kind of annoying to us to be told that it doesn't matter, when it clearly does.

Some specific points:

And once we turn our attention to non-random content generation, I also don't agree that this style implies some sort of independence from the GM's story motivations. Why does the GM, or the sourcebook, put fictional element A in fictional location B? So that the PCs can interact with it and thereby generate some story! (Some actual adventures I have in mind: the 1st ed module Pharoah, and especially the trapped Efreet; OA6 and OA7, two 2nd ed Oriental Adventures modules; the 3E module Expedition to the Demonweb Pits; the d20 Freeport modules.)

A premade story is part of the world, so it is explored during play.

(Even random content generation is not necessarily divorced from story motivations - if you look at the city encounter table in Gygax's DMG, for instance, it is clearly intended to produce exciting encounters rather than a simulation of wandering through a pseudo-mediaeval city, given the number of demon, devil, lich, vampire etc encounters it yields.)

One might say such a table isn't a very good one. Or you might say that it is a pre-established element of the world that exciting dangers are going to be there for the PCs.

There are lots of things that are uncertain but that we don't roll for. It's uncertain, for instance, whether or not a person will fall over and ruin his/her clothes walking through a city with cobblestones filthy with mud and worse - but we typically don't roll for that.

It's uncertain whether or not a person will contract illness from eating cheap gruel in a cheap inn, but we typically don't roll for that either.

It's uncertain whether or not the builder of the dungeon would really want single or double doors to his/her main chamber, but most GMs don't roll for that either - they just decide!

You don't roll for that stuff because the level of detail is too fine for most of us. We want to explore aspects of the world that relevant to our adventurous tales. Occasionally a DM might decide to randomly roll for such a thing. If the random roll is likely enough that occasionally do it makes the world feel realistic, it is still in harmony with exploration. If the random roll is a very slim chance, such that it would only feel right if you rolled it regularly rather than occasionally, then it's somewhat out of harmony.

For the doors, in the strictest form of exploration, you would roll randomly if it weren't pregenerated and there wasn't any obvious choice based on what had been pre-established.

What I'm trying to get it is that a lot of fictional content is generated by sheer stipulation. So, when it comes to the truth about the religious icon, why roll? How is it adding to the play experience for the GM to provide the backstory, or withhold it, on the basis of that knowledge check? (The question is not rhetorical, in the sense that I'm interested in what answers might be given, but I'm not pretending that I don't have an inclination - I have doubts that it does add much to the experience for the GM to withhold backstory on the basis of such a check.)

It's simply a matter of what the experience is. In particular (though I skipped the portion of this thread that dealt with the religious icon knowledge check, so I might be missing background), if the random roll determines that the player doesn't know the significance of the item, they might want to try to find out the significance. This allows them to explore and interact with the world in a different way than if they had simply had it handed to them--which would be interacting with a DM motivation rather than with the world as a third entity.

Now, in practice I don't hold to the strict extreme I described. One of things the I enjoy most about DMing is spontaneous acting and filling in details. But I try to mimic as much of it as I can by making sure content I create on the fly is informed primarily by the world--what makes sense in the setting--rather than by a story--what makes sense for my DM story-goals.

Take an actual example from my game. I was running Lost Mine of Phandelver. The party didn't feel right about attacking the Spectator in the dungeon, since it didn't attack them. They struck up a conversation with it, and I created a name for it, and when the players questioned it, I created a name for the Spectator race. (Since I was running this game in a separate continuity than my own world, I explicitly told the players that this name for the race might not carry over to my normal version of the multiverse.) The players had some interesting role-playing, during which I had to create some backstory for the Spectator. All of it was created to logically fit the lore given for Spectators in the MM, and the backstory of the mine in the adventure--none of it was created based on what I thought would lead a story in a certain direction.

Because this didn't give them an easy way to resolve the dilemma (and after fruitlessly asking people in Phandelver), they went to the nearest place they knew where information might be found: Neverwinter.

In preparation, I spent a bit of time scanning info about it. They actually stayed at that inn mentioned, and I did describe it a bit (though not to that detail), referencing Volo's Guide. One of the players was playing a priest of Kelemvor, and we had established (created) before the game began that there was a temple of Kelemvor in Neverwinter. There was player-DM backstory collaboration on that, and because it assisted in his backstory (that's why he was in Neverwinter at the start of the adventure), I allowed the player input. That's all pre-game collaboration, and I have no problem with that. Had that not been decided before the game began, I would have probably rolled randomly, or made a judgment based on what I know about Neverwinter and what I know about temples of Kelemvor, as to whether or not there was such a temple there--rather than based on whether it would help or hinder the story.

I created a personality for the priest at the temple on the fly. Since he didn't happen to know the answer to their questions, and wasn't of high enough level to provide the needed spells (knowledge roll for priest, world-consistent decision on his level). He recommended visiting the temple of Tymora, where a higher level priest may be found. They did so. I created the architecture of the temple, the personality and name of the priest, and a variety of details entirely on the fly, informed by the details already established about the world. The priest was able to help them out somewhat. He traveled to Phandelver with them, where he knew the acolyte of Tymora at the shrine (since that made the most sense in-world).

Due to the way things worked out, the players ended up convincing the Spectator (I can't recall if they cast a spell to cure his insanity or not) that his contract was over and he didn't need to be there, and he then left, allowing them to plunder the building of its treasure.

Now, during the course of the adventure, there were two things that I did that weren't consistent with my general described DMing style. The first was to say that the drow found a silver dagger and a few copper when she searched a room full of skeletal corpses, without rolling. The adventure said nothing about treasure in the room, and I wanted to reward her for searching. I'm still not entirely satisfied that I did that. The other one was assuming that Traveler Petros (the priest of Tymora) was also associated with the Harpers and knew about the acolyte's mission regarding Agatha and the spellbook. That one I'm happier about, because it was a quite likely detail based on what had been pre-established (including, at this point, the existence and personality of the priest), and it served a meta-game purpose (encouraging the party to pursue side-quests) that I liked.

As I said, I'm not 100% on the strictest end of the scale, but hopefully that will demonstrate the feel of what I'm doing and the sort of experience it supplies the players, and the role/desirability of the two instances in that long sequence of exploration where I did not follow my general rule.


TL;DR Exploration where the players interact with pre-established elements of the world as a third entity (including appropriate random tables to fill in details) and in the moment content creation is primarily to fill in fine details (not the same as pre-established material for the purpose of a fair challenge), is a significantly different experience than one in which the world is created by DM and/or players on the fly to fulfill story desires. Neither experience is inferior, but the exploration style has an inherent appeal derived from allowing and respecting the world's consistent existence as a distinct entity from the DM's spontaneous world creation.

 

[The Crimson Binome]

If you are concerned about the PCs seeming to be the luckiest people ever, then you ration: in Trail of Cthulhu, finding more than the most basic clues is rationed; in Burning Wheel, finding boxes that the GM didn't specify might require a Perception check, with an adverse consequence for failure, just as in my angel feather example above.

That's part of it, but a larger part of it is that such things are not for the GM (or player) to decide, based on such meta-game context. Whether the GM decides that there is a box (or is not a box), that decision cannot possibly depend on whether or not the players ask, if the GM is to maintain the illusion of an objective reality. For a GM to maintain neutrality - for the GM to avoid intentionally empowering or disempowering the players - the decision would need to be based on the GM's prior knowledge of the world, or else it would need to be determined randomly.

As for rationing systems, that necessarily invokes meta-game resources. I'm not terribly familiar with the games you cite, but any resource which the players spend on behalf of their characters (or to further the story) is a meta-game mechanic which requires breaking from character stance in order to invoke. If the players have a limited pool of points to spend on adding elements to the game world (spend a Fate point or a Benny or something to add a box where none was described), then you're adding a gamist element (spend points now, or save for later) over narrative control (player defining the environment, beyond the scope of character ability), which is a double-whammy of illusion-shattering.

As a matter of historical fact I have doubts about that. Games like RuneQuest and Classic Traveller, for instance, weren't derived from Gygaxian play but reactions against it, and in part reactions against its contrivances (like dungeons with convenient levels, and combat with convenient hit point buffers, etc).

That's pretty much what I said in the second half of the sentence, which you didn't quote. Other games added in their own contrivances, because they didn't like the contrivances which Gygaxian play required.

Really, the whole matter of GNS theory is about priorities and personal preference. We'd all love to immerse ourselves in a great game where the story flowed naturally in an interesting manner, but that's not going to happen anytime soon. As soon as you build a game to focus on one element, you need to make compromises within the other elements. Personally, I can't stand narrative contrivance, and find that its inclusion actively detracts from any roleplaying experience.

 

[Tony Vargas]

TL;DR Exploration where the players interact with pre-established elements of the world as a third entity (including appropriate random tables to fill in details) and in the moment content creation is primarily to fill in fine details (not the same as pre-established material for the purpose of a fair challenge), is a significantly different experience than one in which the world is created by DM and/or players on the fly to fulfill story desires. Neither experience is inferior, but the exploration style has an inherent appeal derived from allowing and respecting the world's consistent existence as a distinct entity from the DM's spontaneous world creation.

D&D 'worlds' are imaginary. You can create an illusion that they have an independent existence, but it's just an illusion. The prep-in-advance/random-generation approach creates that illusion in a particular way that works well for some players, and might even let the DM get into that illusion a little bit, if he doesn't think about it too hard. Creating the world just ahead of the PCs as they interact with it can also result in a very effective illusion, that, again, might work better for some players. It does require a different set of talents and a different agenda from the DM, as there's no room for self-denying the DM's creative role. Both approaches create artifacts that can spoil the illusion, as well. Prepared settings tend to be static, they don't change over time the way a real place would - unless the DM goes ahead and fills in such changes continuously (which is using some of the alternate technique of filling in the world as you go) - random tables can give oddball results that stick out and break the sense that there's something real there. Build-as-you-go settings can easily fall into inconsistencies - the mountain pass heading into town where you were blinded by the rising sun, a few adventures latter, is dead north of town - some players notice such details, and some of them even care about the inconsistencies and find their illusion spoiled by them.

The thing I find amusing is that a DM can run a prepped world that he has run many times and is deeply familiar with, without resorting to notes or random tables, and look like he's making it up on the fly, while, conversely, a DM making up everything has he goes along can shuffle papers and roll dice behind his DM screen, and look like he's running an (amazingly detailed) fully-prepared adventure. Even the illusion of how you get your illusion can be illusory.

 

[jeffh]

I think I have a different solution to this problem than the ones being discussed here. So as not to crowd this thread, and because my solution isn't D&D-specific (I originally wrote it up for a system that uses a d6-based die pool), I've given it its own thread here.

Anyone who's interested is welcome to comment over there. I hope having a separate thread isn't too much of a pain for anyone. I really don't think my solution belongs in the 5E forum, though, considering the completely different context it originates from.

 

[pemerton]

Good examples. Actually, I wish things like that appeared more often in published adventures - I can recall a few instances, but only a very few.

They are pretty typical in 4e.

 

[pemerton]

why roll?

For the same reason you might roll to see if a PC can bash down a dungeon door to get to some treasure - if they succeed then the story goes one way, and if they fail then it goes another.

<snip>

If the PCs fail the roll then they have to find some other way to pursue the story. Or, indeed, they can abandon it, in which case it wasn't the dice that made that choice.

<snip>

there are many potential fun adventures out there, where the choice of which one actually plays out dependent on emergent gameplay. If the Cleric identifies the icon right away you get one particular story; if the Cleric doesn't then you get a different story - which may well be just as much fun.

It seems to me that there are a few things going on here. Some of what I am about to say I think might be a rehash of some of [MENTION=1207]Ristamar[/MENTION]'s points, but I'll have a go anyway in my own words.

First, to me there is one important difference between the "open door" roll and the knowledge check. The "open door" roll is not, typically, determining backstory. Rather, it is determining an ingame causal process: does the character hit the door hard enough to open it?

Most combat rolls are like this too.

But the knowledge check is determining backstory, namely, "Does my PC know this fact?" The earliest example I can think of like this, in D&D, is the theif's Read Languages skill. As Gygax explains on p 20 of his DMG,

This ability assumes that the longuage is, in fact, one which the thief has encountered sometime in the past. Ancient and strange languages (those you, as DM, have previously designated as such) are always totally unreadable.​

In other words, the Read Languages roll isn't adjudicating the results of something the thief is attempting in the fiction; rather, it's randomly determining a question of backstory (is the thief familiar with this particular fragment of this particular language, or not?).

Some rolls that are typically treated as adjudicating in-fiction attempts could, in fact, be interpreted as randomly determining backstory: for instance, a roll to hit could be interpreted as "Has my character been trained in the fencing moves that will counter this guy's manoeuvres?", and a roll to pick locks could be interpreted as "Is my character familiar with this sort of mechanical lock design?" - but in my experience to hit rolls and pick lock rolls are not normally interpreted in such a fashion. They are interpreted as determining how quick, strong, agile etc the character was in applying his/her physicality to the situation at hand.

I think the reason for rolling to determine the outcome of an application of physicality to the situation at hand are fairly well-recognised: it avoids railroading, produces surprise etc.

What are the reasons, though, for rolling randomly to determine a PC's backstory? In the case of rolling for starting money, a reason can be fairness combined with verisimilitude: everyone has the chance to be the richest, but not everyone is equally rich. In the case of rolling for handedness, in systems which care about facing, shield vs weapon hand, etc, again it can be fairness combined with verisimilitude. In the case of 4e monster knowledge, the roll means that the players have a chance but not a guarantee of getting the monster stats - so its a type of rationing device.

In the case of the religious symbol, I don't see any issue of fairness or verisimilitude. But I don't see any reason for rationing either.

The reason that you (delericho) give is that we want to randomly determine whether the PCs go on adventure A ("talk to the sage") or adventure B ("track down the cult"). I guess I don't understand why this is randomly determined. Most of the time a GM wouldn't randomly determine which module to buy - s/he would choose one that looks fun.

I'm not saying you must roll for such things. I'm saying that you might roll for such things, and especially in those cases where the outcome is relevant to the story.

I can see that the outcome is "relevant to the story" in at least two senses: (1) the PC backstory is different depending on whether the roll dictates knowledge or ignorance, and (2) the events the PCs undertake following the roll might be different depending on whether the roll dictates knowledge or ignorance.

But (2) would be true even if the roll wasn't made. As you have posted:

this isn't an encounter. It's a gating point between encounters. And, no, I don't think it should be much of a choice whether you examine the item (and therefore roll) or not - of course you should roll.

The interesting question (to me, at least) is not whether to roll or not - it's what to do after the roll. If you now know what you need to know, how are you going to deal with the cult? If you don't, what's your alternative plan of action?

That is, this isn't an encounter - it's not a resolution of action declarations by the players for their PCs. If the players know the backstory about the holy symbol, they have choices to make (ie (2) is true). If the players don't know the backstory about the holy symbol, they have choices to make (ie (2) is true). Which is to say, (2) is true independent of (1).

Hence, the desirability of the players having choices to make, and thereby affecting the content of the shared fiction, has no bearing on whether or not we should roll to determine whether or not the PC backstory includes knowing the religious symbol backstory.

And I'm still unclear what the benefit is of rolling randomly to determine what challenge confronts the PCs, rather than the GM just stipulating - as s/he does in so many other cases (eg when s/he buys a module by choice rather than by random selection).

Because while I want my PCs to succeed on their quests, I have to allow them to fail.)

Having the PCs fail because the players make silly choices is one thing. Having the PCs fail because the players get unlucky in action resolution is another thing. (Gygax recognised the difference between these two things when he said that, if a skilled player's PC is killed in combat, the GM might instead adjudicate it as unconsciousness or maiming: DMG p 110.) Having the PCs fail because the GM doesn't give the access to the backstory they need to succeed is yet another thing.

The difference between silly choices and access to backstory isn't always clear. For instance, if the party includes a high level cleric then maybe the players should be using a Commune spell to get access to the backstory that they need. Similarly, one way to get information is by talking to/interrogating NPCs, and that typically involves action resolution.

But I don't see how this generates a reason to make the Religion roll. If the GM wants the players to engage in action resolution in order to get the backstory, then the ability/skill roll is counterproductive because it might thwart that. If the GM thinks that action resolution in order to get the backstory is tedious or a timewaster, then the roll is counterproductive too. And if the GM can't make up his/her mind whether having to actually engage in action resolution to get the backstory is worthwhile, then we're back to choosing the challenge via random selection, and I still don't see how or why that's a good thing.

Unless the roll is literally without consequence then it always has stakes

<snip>

Why do they need to know the stakes?

If the players don't know the stakes, then the roll doesn't seem to serve any purpose other than titillation for the GM.

To give a caricatured example: suppose at the start of the campaign the GM asks the players to roll a d6, and the GM has mentally noted that, on a 1-3 s/he will start the campaign in FR, on a 4-6 in Greyhawk. That's a roll that would have consequences for the content of the shared fiction, but I don't think that shows the GM was being sensible in calling for such a roll. How does it add to the game? If the GM can't make up his/her mind, and isn't going to ask the players, why not flip a coin?

Back to the holy symbol case, if the GM doesn't know whether it would be more fun to run a "visit the sage" adventure or a "track down the cult" adventure, why not just flip a coin? How is it better to make the players roll? And why should the outcome of the coin toss depend on whether or not one of the players has built a PC with a high religion skill bonus?

Apart from anything else, if the players have built PCs with low religion skill bonuses then, on the approach of checking religion they are more likely to end up having to visit the sage, but at least in my experience the sorts of players who enjoy having their PCs visit sages are more likely to build their PCs with high religion skill bonuses.

There is another reason that knowing the stakes, or not, matters, which comes up in response to the following:

They want the information that's gated by that knowledge roll. Deciding not to attempt the roll just means they go straight to the "fail" case.]

the alternative that's being suggested is that the PCs might have found this clue and yet may choose not to examine it.

One way <snippage> is to frame the consequences of success and failure clearly: succeed, and you get the info for free and quickly; fail, and you get the info at a cost and requiring the passage of time.

Sorry, I find that entire construction absurd. Because the Cleric can't personally identify this icon we have to go consult a sage?

The first two of these three quotes seem to be making an assumption which is exactly what is being questioned - namely, they seem to assume that rolling Religion is the way to adjudicate the PCs examining a clue.

I take it for granted that PCs who find a holy symbol at the scene of the crime will examine it. But that doesn't necessarily mean a Religion check is to be rolled. [MENTION=97077]iserith[/MENTION], in this thread, has already emphasised that within the 5e framework whether an action declaration (such as "we examine the symbol to see if we recognise it") is resolved via a die roll or via GM stipulation of the outcome is something subject to GM discretion. The same is true in many other systems too (eg those that use "say yes or roll the dice; or the example of reading languages in AD&D, that I quoted from Gygax's DMG above, where the GM is entitled to say no with no die roll).

As to the framing suggested by me in the third quote, no one is saying that because the cleric doesn't recognise the symbol the PCs have to confront a sage. I am assuming that the players would agree to those stakes. The point is that, if I as GM have decided that the backstory is not going to be automatically available, then I want the players to confront that - what are they prepared to do to acquire the information? If the players don't want to stake a visit to the sage, they are free to nominate something else - eg they visit their unsavoury contacts in the world of dark cults. The point is that if they're not ready to commit to something, then they don't get to make the roll!

To put it another way: if I'm not going to force the players to make some sort of commitment, to stake something, in order to get the information, then why as GM am I not just giving it to them?

This also helps show why it is useful for the players to know the stakes: it leds them decide how much they care, and hence how much they should invest in the roll. Eg if the players really don't want to have to visit the sage (eg they know the sage will chastise them, or charge them) they can devote resources to the religion test (eg go to the local monastery's library in order to get a circumstance advantage, or summon an angel in order to get a helping advantage, or whatever).

When the players are making rolls that only the GM knows the consequences of, the players don't know what they are staking, hence don't know what the worth is of succeeding at the roll, hence don't know what degree of resources they should commit. The game becomes closer to a lottery, and often also to a railroad shaped more by the GM's choices than the players'.