Knowledge checks are a bit problematic for me, any ideas?

[Jon_Dahl]

I'm a big fan of knowledge checks. That is why I'm usually very strict on withholding information from the PCs unless they have the relevant knowledge skill. My insistence on using knowledge skills has led to many situations that are far from ideal. There are at least three main scenarios that cause problems.

Scenario #1: Second-hand information followed by witnessing firsthand. First you hear about a lizard-person who has a blue skin. You try to recognize it from the (poor) description. This calls for a skill check with a hefty penalty. Then you see the blue half-dragon for yourself. Does this mean two separate rolls? Or one roll for each? What about the bonuses? This example may seem simple, but sometimes this can very annoying and complex. There are so many examples that I don't even know where to begin... For instance, a villager gives you a poor description of a blue half-dragon and you roll well. Even with the penalties, you get 18. Then you see the blue half-dragon and you roll very badly: 9 with all the modifiers. What does this mean? And what if the blue half-dragon tells you that it's a blue half-dragon and someone you know confirms that "Yes, it is! It is half a blue dragon, half a man!". Do you roll again? Or not? Why? Why not? Or what happens?

Game-mechanically speaking, how does this work? First have poor data to work with, then sufficient data and, after that, very good data to work with? In my games, I ask the players to save their knowledge checks: "Okay, this is your knowledge check on this particular monster. Save it. It will be like this until you invest more ranks in that knowledge skill. You may get additional bonuses to your knowledge roll later on, and those bonuses will be added to THIS roll. You just have tell me the current total. Okay?" This gets kind of nuts because saving your roll is a lot of work if we play once a month!

Scenario #2: One-trick ponies, multi-ability monsters and high knowledge checks. High knowledge check on a dire boar just isn't worth as much as a high knowledge check on an ogre mage. I don't know if this is a major problem, though. I just find it a bit ridiculous when I give four bits of useful information about a dire boar and quite often this information just confuses players. For instance: "Okay, dire boars have a pretty low AC, but it's not like super low!" and then they focus on that detail because officially it was a bit of useful information that was given to them by the DM.

Scenario #3: What evidence is important? If you see a village that was burned to the ground and the only survivor says that it was a flying creature that burned everything to the ground in the middle of the night, is it easy or difficult to know what that creature is or isn't? Do we go by the HD in this case or by the really easy/basic/hard categories?

What do you think about these problems and how should they be handled? Whatever happens, I would like to have knowledge checks in my games and my players genuinely like them. Throwing them out of the window or diminishing their importance is not really the solution...... Maybe?

[Duke of Urrel]

Every dungeon master has their own rules about the Knowledge skills, because they are very world-dependent.

I don't have lots of advice that's transferable from my world to yours, but I may be able to help with this bit.


In my games, I ask the players to save their knowledge checks: "Okay, this is your knowledge check on this particular monster. Save it. It will be like this until you invest more ranks in that knowledge skill. You may get additional bonuses to your knowledge roll later on, and those bonuses will be added to THIS roll. You just have tell me the current total. Okay?" This gets kind of nuts because saving your roll is a lot of work if we play once a month!

My general rule is that all granted checks take 10 by default. I often grant skill checks that take 10 secretly when I believe a PC may have Knowledge of something and it doesn't occur to the PC to make a skill check at all. Of course, this secret granted skill check succeeds only if it can succeed by taking 10. If it succeeds, I give the PC some free information and tell them that they have it because they have Knowledge skill, and I name the Knowledge skill that they have to thank for it. If the secret Knowledge check that I grant to a PC fails, I don't even tell the PC that I have just granted them a Knowledge check that failed. Instead, I keep one of my secrets until the PC chooses to make a Knowledge check, which of course the PC may choose to roll.

When I'm the dungeon master, there is never any need for a PC to save a former Knowledge check, because every piece of knowledge requires its own separate check. It makes no difference if you have two or more questions about one and the same creature. You don't know EVERYTHING about a creature automatically just because you made ONE successful Knowledge check. If you make a Knowledge check that scores some multiple of five higher than you need, you get to answer more than one question (because you get one or more free answers to follow-up questions), but if you make a Knowledge check that merely succeeds, you get (following my house rules) only one piece of information in addition to the creature's name. If your Knowledge check fails, you don't get to make another Knowledge check unless you ask a completely different question about the creature, though you may ask as many questions as you like and may make a separate Knowledge check for each. Moreover, if the answer to a question belongs to two or more fields of Knowledge that you possess, you get to make two or more Knowledge checks, one for each field, to answer one and the same question. But I never allow you to retry a Knowledge check of one and the same field to ask one and the same question about one and the same creature more than once – until, of course, you get another skill rank in Knowledge of this field.

Not every piece of information is old enough to belong to any kind of Knowledge skill whatsoever. If a piece of information is recent news, it may be just too recent to get with a Knowledge check. In order to find out this news, you must make a Gather Information check. In order for some fact to belong to any field of Knowledge at all, it has to be old enough for somebody to have observed it and for somebody to have written it down in a book that you may have read at some time before now.

[GoodbyeSoberDay]

If one does allow multiple rolls to identify the same creature, the PC should take the better result. The worse one simply doesn't help. I use a similar form of "save your roll." If we forget, eh, re-roll. No biggie. Something I add to this: Knowledge skills confer knowledge to the players, but not the only possible knowledge. If the PCs properly identify [creature x], e.g. a half blue dragon, and they observe that creature using [ability y], they are free to make generalizations about half-blue dragons in character doing [ability y], with the caveat that this particular half dragon might have something else giving it that ability.

For scenario 2, the goal of a successful knowledge check isn't to confuse players. Give the most useful stuff first. The explicit expectation of the players should be that, if they're hearing about things like AC ranges instead of special attacks and abilities, the creature has run out of useful abilities to describe. Sometimes, with such basic creatures, I just say "your character basically knows the monster manual entry for Dire Boar." Then again I often have super-knowledge-skill characters in the group.

With scenario 3, again we run into the issue of second hand knowledge; in general I wouldn't call for specific identification checks after second-hand descriptions. Instead, the PCs already know what they know about various creature types (including using your saved roll rules), and they're using the villager's description to make deductions about what the creature could be. If they haven't rolled knowledge checks on some creatures, well, this is a good reminder that they should have done that, so go ahead and do it now. With such a vague description it would be hard to pin a particular creature down; for instance, a pyromaniac sorceror with Fly as a spell known is consistent with what the villager said. But if he has more details, such as huge-sized, red scales, wings and tail, etc., the PCs can start using their knowledge skills ("hey, red dragons have red scales, and all the adults are huge sized") to start narrowing down possibilities.

[zlefin]

#2 sounds touhg; some monsters in particular just don't have much to say about them.
maybe just straight up give them the numbers instead? like just say what it's AC is? (for monsters that otherwise lack useful things to say about them, fluff it as you being able to very accurately estimate it's hides' thickness)

or, you could give a partial benefit similar to knowledge devotion in cases where there just isn't much useful to say.

saving knowledge checks is a good idea; but if people forget/lose track of the results, then just roll again (ofc that's somewhat abuseable, but if your group won't abuse it that's not a problem).

for your case 3; I'd go by categories, and give some value to non-precise identifications that can just exclude some possibilities.
while HD may be the listed metric, its' really not a very good metric for deciding knowledge checks. makes more sense to use CR than HD; and even then there's a lot of other modifiers that should apply. I wonder if someone has made up a better table for knowledge checks.

[BowStreetRunner]

In TV shows and movies you will often have the heroes go off to consult some sage who provides them answers to their questions. This can be anything from a one-time visit to a reputed expert on the subject (so effectively a Gather Information check to learn about the expert followed by a mini-quest to locate them and convince them to answer their questions) to a regular supporting character who is always on-hand to do the research for the party (cohort or NPC hireling perhaps). It can even involve the party going down into some dusty library archive and looking through old books (Search checks to first find the book and then locate the proper reference within the book, with possibly a Decipher Script check if the book is in another language).

The key to these sorts of situations though is a play-style that accepts that monsters may need more than one encounter to defeat. While this happens often in stories, it does tend to be more rare in RPGs where for some reason players often seem to act like every battle must be a win-or-die-trying contest. Scouting out the encounter first, fleeing from an apparently unwinnable encounter in order to regroup with a better plan, and other strategies like this may not be something your players do very often. But if you can get them to work that way, then you have the option to allow them to put in a little work to either get someone else to make the Knowledge check for them, or to find resources to give them a high bonus to their own check.

EDIT: It's also worth keeping in mind that people in the real world make a huge number of decisions based merely on assumptions. If they fail a check, instead of giving them no information you could sometimes give them choices. "You're pretty sure this monster has some sort of defense against magic - either spell resistance, energy resistance, or possibly ridiculously high saves - you just aren't sure which."

[Glimbur]

For scenario 1, just have a cap on how much they can deduce from the bad description. Even if they blow the check out of the water the best they get is: blue scales could be one of these three things. Then make an uncapped check when you see the thing for yourself.

[Segev]

As has been suggested, characters can only leverage their knowledge against things they perceive/hear about.

A really high K:Arcana check might tell you that "blue-scaled humanoids" include blue half-dragons, spellscale humans of blue dragon heritage, some Dragonfire Adepts, etc. It's Arcana, so it's mostly going to be about dragons. It will tell you this even if you're actually hearing about somebody's description of a blue-scaled Lizardman or the flash of a tail vanishing beneath the waves that actually was a mermaid.

K: Nature is more likely to bring up actual lizards and other creatures with blue scales.

K: Local will get the various humanoids and monstrous humanoids.

The point being that, barring more information to narrow it down, "The thing walked on two legs and had blue scales!" can only get you so far. The Knowledge checks can give lists of things that fit that description, and little more.

When encountering more evidence, I usually just let people roll again. If they roll better, great! They get more information about the new evidence. If they roll worse, well, they get new information about the new evidence, but less than otherwise. Or they may fail to figure anything out, if they roll too poorly. Sure, now they see some scortch marks where the blue-scaled humanoid was fighting, but they just don't recall that blue half-dragons can breathe lightning, so they don't pick up on that as significant. Who knows what caused the scorch marks, after all?