Knowledge Checks In-Game

[AnimeTheCat]

So, just curious to know how others have used Knowledge Skills, and Checks, in-game. I've personally used and seen them used in a variety of different ways.

One way, and the most common way I've seen them used, is where a player asks the DM "What knowledge check do I need to roll to get X information?". This utilization is primarily bus for a Q&A session between the player and the DM and less an interaction of the Character and the world. Nothing wrong with it, and clearly it works well enough for most groups. In these situations, the DM usually selects the most applicable knowledge skills in a range that would grant information, and adjusts the information based on the check result and the skill used. This is similar, but not as extensive as, the second method below.

Another way I've seen them used, and my favorite way to use them personally, is to have the player tell me what knowledge skill they are using, and I determine what they know based off of their check. All things correlate and reciprocate in nature. An advanced Architect can easily determine the structural integrity of buildings and infrastructure, but they can also apply that knowledge to armor, and natural armor plating to determine strengths and weaknesses of that particular part of a creature, but that's where their knowledge falls off. A wizard, highly trained in the field of Arcana, could likely draw comparisons and conclusions about non-magical creatures with relative reliability, but would likely only be able to draw minor conclusions based off of their knowledge with regards to structures or non-natural outsiders. Because of this, I feel that using knowledge skill in this way enhances both the creativity of the player AND the DM, while simultaneously enhancing the interactions between the Character and the World. This also allows esoteric texts to be found that describe foreign creatures or substances in a certain way that may seem obvious to someone reading them from one body of knowledge, but not to another, and only allow correlations to be drawn when somebody observes the text with knowledge of both fields. This allows the DM to shroud even very simple and well-known creatures and places in mystery only for the player to figure out through their interactions with the environment via their character's knowledge. Imagine describing a Kobold from the perspective of an individual that has extensively and exhaustively researched and learned about planar creatures, including some legends and myths that describe interactions between outsiders and dragons (or other Draconic creatures). They may not know extensively what dragons are or how they act, but they could certainly draw similarities between the limited descriptions of draconic creatures in their planar texts and these similarly draconic creatures, possibly creating a false identity for them and shrouding the entire situation in mystery.

A third way that I've seen Knowledge skills used, and I think my least favorite way, is where the DM looks at the knowledge skill list, looks at the object being questioned, and if the player can't draw conclusions because they don't have the knowledge skill, they know nothing and can not speculate based on other knowledge skills. I'm sure there are reasons for this, and maybe this is the standard way for adventure league or similar, but I don't feel like this method creates a living world.

Please comment on what you think of these methods, or if you use them in a different way please share that way.


I'm using Knowledge skills in a way that you haven't listed, so I thought I'd add it. Admittedly it's a houserule, but it works fine for me.

I use passive checks for a number of skills that you can use without consciously trying to. Perception, Sense Motive, Knowledge. It lets me be fair to players, and valorize their skills, without slowing down my game. Now the way I do this is that I always assume they take 10, and tell them what they know in consequence. So if the DC to recognize a Smurf is Knowledge (Nature) 13, and the character has a score of +8, upon encountering the creature I straight-up tell them "you recognize the characteristic blue skin and white headwear of a Smurf. They're tiny Feys living in loosely organized communities in deep forests. They are pacific, but have a particular distaste of Wizards and can get aggressive towards them. Sometimes, however, they fall under the influence of a Redcap, who turns them into warriors. This may be the case here, because this one is looking at you and drawing a minuscule, but very sharp dagger. Roll Initiative". Or for an example relating to more general information, "Doctor Gargamel? You recognize the name as that of a somewhat famous local wizard, specialized in Necromancy, but you don't know much more about him."

Now unlike Perception and Sense Motive, I don't allow active uses of Knowledge, except in a research situation (library, books, asking informed people). You either know something or you don't, and if you don't, thinking about it some more is not gonna help. What that means in practice is that your Knowledge scores are constant and reliable: 10 + your bonus. If you've studied the Planes enough, you will always, without fail, be able to tell an Imp from a Quasit. But until you study some more and have enough ranks, you have no chance of knowing what SLAs Balors have.

Advantages: - Players know that their Knowledge investment will pay off each and everytime it comes up, without having to ask for it. It feels fair.
- I can figure out in advance what characters will know and include relevant knowledge in the preparation of my game, because I know their scores.
- It also lets me sneakily include flavor text and details to make the world feel more alive, or include hints to help the players ("that creature is not acting typical of its kind, why might that be?"), while giving the characters credit for it, which makes the players feel awesome.
- The players don't waste any game time wondering what Knowledge to roll, or rolling it.

Disadvantages: - There are no miracles. Since I know what the players' scores are, I can't be surprised by a very good roll that makes me go "okay, you know everything on that Monster manual entry" and makes the player go "wow, did I tell you I'm an expert on this particular obscure topic?". Such things are fun for everyone when they happen. But I feel it's not that big of a disadvantage, and the advantages are well worth it.

[Falontani]

the way you describe them the second one does seem like the best, however that is describing them in a highly personalized way. I personally believe the third option is the way the rules were intended to run with only a single knowledge skill working for any given topic. I believe the first one is easiest on the players and the DM, but isn't too kind to the in game lore, while the second one is the best one for lore, however the hardest to use effectively as a player, and probably difficult for a DM that has to answer in the moment if they don't have any descriptions ready

[Ellrin]

The way I prefer to use them is the player just rolls once when an appropriate subject comes up. That roll is added to all his knowledge skills, and the DM determines what the PC knows about the subject based on the results.

E.g., player has +17 to Knowledge (arcana) and +10 to Knowledge (history). He encounters a magical artifact and rolls to see what he knows about it. Roll is 12, so DM determines what he knows about the magic of the artifact based on a result of 29 for arcana and what he knows about the history of the artifact based on a result of 22.

I prefer to do it this way because Knowledge skills are pretty much always passive, and no one knowledge skill typically covers all the possible information a PC could know about a given subject, so choosing a single skill, or having one chosen for you, doesn't really feel like it reflects what's supposed to be going on in game.

[Dr_Dinosaur]

I typically tell my players (the ones with the highest relevant skill or relationship with an NPC) any information they actually need to progress without a check, with more accuracy and detail for higher bonuses. In the case of rolling, I’ll accept suggestions from players regarding alternate knowledge skills to roll, but they get a penalty for the degree of separation.

[Climowitz]

I made a few changes to knowledge for my games. I took out knowledge local because it interacted weirdly with new places or required massive pointa invested to actually be useful, instead i use gather information. Knowledge geography and nature are both combined into knowledge nature while half of knowledge nature ended in knowledge biology. Knowledge nobility is now knowledge society and works like cultural and popular knowledge instead of being only about kings and the like.

Every knowledge is used either to find information of a creature or an unknown matter to the player. Anyway the player asks what knowledge can he use or if the ones in which he has ranks work

[Darth Ultron]

I tend to avoid Knowledge checks in my games. I would much rather role play out the characters getting information then just have a player sit there are roll some dice. Knowledge checks are simply bad game design and do not lead to fun gameplay. It seems sorta absurd to gate factual information behind luck

I just about always let a player choose the Knowledge skill rolled, though most clever players just roll them all.

A knowledge skill roll does just give quick trivia information, and in my game it is not absolute fact...and might be wrong or incomplete. So it is always best to not rely 100% on knowledge checks.

I much more enjoy spreading knowledge everywhere in the game. It's a good reason to talk to most NPCs. Most NPCs know something of value, if you ask the right questions. And you are the right person asking. Just about any PC can get knowledge by money or force...maybe. Most can get knowledge by role playing...maybe. And there is always to peer to peer effect.

For example a commoner type PC can talk to the commoner NPCs and learn things. The same way warriors have a bond. Bards, wizards and clerics do mostly like to share information. The clever PC collects infromation to be used to trade.

Also, I'm a big fan of OOC knowledge.....just not during the game. I encourage players to read the rule books. If Amy has an elven cleric, I will recommend she read the Complete Book of Elves. And other rule books. I also recommend reading novels, though a 'novel' is just a 'story or tall tale' in my Game World. Still, if Amy reads some elf novels, she will learn a lot about elves. I'm also open to having players ask questions about lore in e-mails, and I do keep a Website of lore too.

Also, I have a game wiki

[liquidformat]

I took out knowledge local because it interacted weirdly with new places or required massive pointa invested to actually be useful, instead i use gather information.

Ya knowledge local always bugged me, it seems like it would be a slew of different knowledge subsets like knowledge local (waterdeep), knowledge local (Sword Coast) and so on. But it also is used for knowledge local (humanoid biology) which is just weird and in most of the games I have been in its only real use was with knowledge devotion to kill things. Everywhere else gather information is normally used.

I like the second version of knowledge check but normally see something closer to the first or third. Part of the issue seems to be a question of game design. If you are playing a premade module often the dm has limited info making one or three easier. Whereas, someone who designed their own campaign or world has a large breadth of knowledge to work with making it easier to use the second version.

[AnimeTheCat]

I use Knowledge (Local) a ton with my NPC's. I also have it apply to Kingdoms or Nations rather than towns or cities.

In a few games, I gave everyone max ranks at character creation in the Knowledge (Local) skill that corresponded to the nation or kingdom they were from. It really helped to push some narrative prior to the party's arrival when the party's Barbarian was able to describe the interworkings of the tribes and some of the local tribal legends to the party, as well as some customs that would be known to the common tribes people. Knowledge (Nobility and Royalty) was similarly broken down, but had to be taken with skill points, and that marked more specific knowledge about economics, international (or in the above example inter-tribal) governance, and specialized tribal customs (such as interacting between chiefs or key leaders).

I feel like Knowledge (Local) and Knowledge (Nobility and Royalty) are complementary to each other, and knowing one more than the other can help a character define their backstory. A sheltered Noble kid growing up to be a paladin without knowing the struggles of those on the street might have very high ranks in Knowledge (Nobility and Royalty) but no ranks in Knowledge (Local) and thus would only know the "Noble" tales of heros and heroines, whereas the Bard in the Tavern may have learned much about both the local setting and those who rule it, even performing in some keeps and at some parties, resulting in high ranks in both Knowledge (Local) and Knowledge (Nobility and Royalty).

In some circumstances, I've granted stray ranks in certain knowledge skills to people to indicate the growth their character has had. Other times, I leave it as an untyped bonus to certain skill checks as well. It all really depends on the game and the group.

[Seto]

I'm using Knowledge skills in a way that you haven't listed, so I thought I'd add it. Admittedly it's a houserule, but it works fine for me.

I use passive checks for a number of skills that you can use without consciously trying to. Perception, Sense Motive, Knowledge. It lets me be fair to players, and valorize their skills, without slowing down my game. Now the way I do this is that I always assume they take 10, and tell them what they know in consequence. So if the DC to recognize a Smurf is Knowledge (Nature) 13, and the character has a score of +8, upon encountering the creature I straight-up tell them "you recognize the characteristic blue skin and white headwear of a Smurf. They're tiny Feys living in loosely organized communities in deep forests. They are pacific, but have a particular distaste of Wizards and can get aggressive towards them. Sometimes, however, they fall under the influence of a Redcap, who turns them into warriors. This may be the case here, because this one is looking at you and drawing a minuscule, but very sharp dagger. Roll Initiative". Or for an example relating to more general information, "Doctor Gargamel? You recognize the name as that of a somewhat famous local wizard, specialized in Necromancy, but you don't know much more about him."

Now unlike Perception and Sense Motive, I don't allow active uses of Knowledge, except in a research situation (library, books, asking informed people). You either know something or you don't, and if you don't, thinking about it some more is not gonna help. What that means in practice is that your Knowledge scores are constant and reliable: 10 + your bonus. If you've studied the Planes enough, you will always, without fail, be able to tell an Imp from a Quasit. But until you study some more and have enough ranks, you have no chance of knowing what SLAs Balors have.

Advantages: - Players know that their Knowledge investment will pay off each and everytime it comes up, without having to ask for it. It feels fair.
- I can figure out in advance what characters will know and include relevant knowledge in the preparation of my game, because I know their scores.
- It also lets me sneakily include flavor text and details to make the world feel more alive, or include hints to help the players ("that creature is not acting typical of its kind, why might that be?"), while giving the characters credit for it, which makes the players feel awesome.
- The players don't waste any game time wondering what Knowledge to roll, or rolling it.

Disadvantages: - There are no miracles. Since I know what the players' scores are, I can't be surprised by a very good roll that makes me go "okay, you know everything on that Monster manual entry" and makes the player go "wow, did I tell you I'm an expert on this particular obscure topic?". Such things are fun for everyone when they happen. But I feel it's not that big of a disadvantage, and the advantages are well worth it.

[AnimeTheCat]

Oh yeah, I do recall a DM running it that way before. They came from 5e and were big fans of passive perception so they did something similar with all of the knowledge skills, and combined listen and spot to make perception. I'll add that to the OP.

[BassoonHero]

I don't like the Knowledge skill. That is, I don't think that the D&D 3.5 skill system is a good mechanical fit for in-character knowledge.

What a character knows is a vital part of their characterization and their connection to the campaign world. In a sense, it's too important to be trusted to the roll of a d20. It's true that even a particularly knowledgeable character could have surpising gaps in their knowledge and that even a dabbler might have heard an obscure fact or two. But the extreme swinginess of a d20 roll prevents a player from developing coherent expectations for what their character would know. In an actual game, players invariably assume that their characters possess all sorts of common knowledge that the rules say ought to be rolled for. The player of a cleric with Knowledge (Religion) will assume that their character automatically knows the important details and rituals of their faith without bothering to roll every time. The player of a rogue with Knowledge (Local) will assume that their character knows the streets of their home city. To be clear, this is a good thing; rolling for all of these things would be disruptive to the game and to the characters. The fact that it is usually better to ignore the Knowledge skill mechanic is, in my mind, a strong indictment of that mechanic. While having a Knowledge skill on your character sheet can usefully inform roleplaying, actually using the skill's mechanics would tend to hinder as much as help.

As an aside, it's frustrating as both a player and a DM when a Knowledge check is required to notice or understand details that exist to enliven the setting. For instance, if the party sees a mural depicting a historic battle, and it's plausible than someone in the party would recognize the scene, then there's no harm in just telling the party what it is. What is there to be gained by interrupting the exposition so that a player can roll dice, all so that the DM knows whether they should finish describing the room? The DM doesn't need the dice's permission to deliver exposition.

What, then, should Knowledge checks be used for? If not for characterization and exposition, then perhaps for weightier matters. It's a common feature of fantasy stories for the plot to turn on a forgotten piece of lore. The problem is that it's really, really problematic for the outcome of a quest in an RPG to depend on the outcome of a single skill check. For one thing, adventures should not generally depend on the players having one particular ability. When that ability is "trapfinding" or "divine spellcasting", then perhaps most parties will supply it, but when the ability is, say, Knowledge: the Planes, it's very dangerous to assume that the PCs will be capable of making the check. And even if they are, it's just as dangerous to assume that they will make it.

Other d20 rolls are swingy, but the randomness is usually mitigated by other factors. In most cases, characters can take ten on a skill check when not under stress. Taking ten on a Knowledge check, if allowed by the DM, raises serious logical problems. In other cases, individual d20 rolls usually aren't decisive. In combat, a character can expect to make many attack rolls, so success or failure in the combat does not hinge upon a single roll. Save-or-die spells are a notable exception, and those spells are the continual subjects of complaints from players and DMs alike. But Knowledge checks are always "save-or-die"; one roll counts for all. If you fail a Climb check, you may suffer an unpleasant consequence, but in the end you can usually give it another try. If you fail a Knowledge check, then there's nothing to be done.

Knowledge checks are particularly unpredictable, even relative to other skill checks, because the DCs are inherently subjective. The DM must decide on their own whether a question is "really easy", "basic", "really tough", or something else. This is not something that most people can do in a consistent or predictable manner. Even worse, most of the time the question itself will be purely fictional in nature. If a character wants to identify a noble's banner, then the DM must decide how prominent that noble is in the setting and translates to an abstract scale of difficulty with few or no useful reference points. The degree of subjectivity involved often makes the DC, and therefore the result, seem arbitrary. When a DC is derived from concrete in-universe details, there's clear link between skill, difficulty, and success or failure, even though the dice provide an element of chance. When the DC is ballparked by the DM on-the-fly, then it feels less like your character has succeeded or failed and more like the DM and luck conspired to determine the outcome.

All of this means that if the characters really need a piece of information, then the DM has to provide other means for them to get it. There is no shortage of options here: the characters could visit the Imperial Archives, consult an old sage, or raid a crypt. But these solutions are inherently more interesting than rolling a Knowledge check in the first place! If a character succeeds at making their Knowledge check, then they haven't overcome a challenge -- they've skipped a side quest, not by choice but purely by chance. It's fine and good when a quest allows the characters to choose between several paths, but here we have the worst of both worlds: the DM must prepare for both outcomes, but the characters aren't offered a meaningful choice.

What about information that, while not critical to the campaign, may nevertheless give the characters a significant advantage? The consequences of failure here are less dire than above, but the root problem remains. Knowledge DCs are arbitrary, the roll is swingy, and the result is final, so the outcome feels random and disconnected from the character's efforts. Neither success nor failure feels properly earned. On the other hand, when the characters uncover important information as a result of their decisions -- taking a risk or finding a clever strategy -- the players justifiably feel like they have earned the fruits of their characters' success.

Another factor in the perceived arbitrariness of Knowledge checks is that, unlike most skills, they don't clearly correspond to an attempted action by a character. When a character tries to climp a wall, persuade an NPC, or spot a distant object, it is clear what the character is doing and what it means for them to succees or fail. A knowledge check is much fuzzier in this regard. From an in-universe perspective, either a character knows something or they don't, even before a knowledge check is rolled. A knowledge check is inherently retroactive: when the check succeeds, it means that the character knew the information all along, and when it fails they never knew it to begin with. It means that the character learned or did not learn the information at some nebulous time in the past, off-screen. Compare to a character trying to climb a wall: if they fail, then they missed a handhold, or their foot slipped, or something else happened in that moment that caused the failure. (You could suppose that the final cause of the failure might stem from an error in learning or insufficient practice, or something else remote from the attempt, but there is nevertheless a proximate cause. Success or failure on a Knowledge check has no proximate cause.)

One counterargument to the above point is the idea that a Knowledge check represents an active mental effort to remember information that was learned long ago. This is an understandable attempt to find a sensible interpretation for knowledge checks, but it has several flaws. First, it assumes that the character knew the information in the first place. For this to make sense, the character would have to have learned everything for which they could possibly make the check DC, then forgotten a great deal of it. Second, if a knowledge check represents an attempt to remember something the chatacter does truly know, then they should be able to try over and over until they do remember. It's not as though failing to think of something you know actively erases the fact from your memory.

Another possible counterargument is an analogy to the Spellcraft skill. A major use of Spellcraft is to identify spells as they are being cast. Like a Knowledge check, a Spellcraft check of this sort is purely mental and relies greatly on a character's accumulated knowledge. But this use of spellcraft is not to know or remember something, but to interpret new information in light of what is already known. A character with Spellcraft observes a spellcaster's gestures and speech and attempts in a brief moment to figure out what spell they correspond to. If the spellcaster casts the exact same spell in a subsequent round, the character with Spellcraft will make a new attempt to identify it based on a new set of observations. In this way, Spellcraft is no different from Climb: although the character relies on their knowledge and experience, each skill check represents a specific attempt to apply that experience to a particular situation; a clear in-character interpretation is available for either success or for failure; and the character is free to try again if the situation should repeat.

One more counterargument is that, while admittedly most Knowledge checks don't have a good in-character interpretation, the skill can nevertheless be used in ways that do have such an interpretation. For example, suppose that the characters need to find some specific piece of information in a library. A relevant Knowledge check could represent the process of research. This is fair; unlike a typical knowledge check, a roll for research represents an identifiable course of action and has identifiable reasons for success or failure, even given that the information is there to be found. But from a gameplay perspective, this scenario is problematic in several ways. First, and most importantly, if a piece of information is in the library, and the characters are determined to find it, then there is no reason they shouldn't keep trying until they do. That is, they should nearly always take 20. On one hand, this is not compatible with the standard rules for the Knowledge skill (which prohibits taking 20), and on the other hand, it means that once again no actual roll is necessary. Second, such a Knowledge check (or series of checks) might represent days of in-character time during which only one or two characters are really capable of contributing. The party barbarian is probably limited to carrying the wizard's stuff around and occasionally growling at other visitors eyeing the same books. This sort of scene could nevertheless be a great opportunity for roleplaying and world-building, but rolling dice is likely to make the process less interesting for the barbarian rather than more interesting for the wizard.

A common piece of advice for running an RPG is to roll dice only when there is an objective, a course of action, a chance of failure, and a consequence for failure. The Knowledge skill nearly always fails to supply a course of action, and the chance of failure is retrospective rather than following prospectively from the state of the world. This makes the skill mechanic a poor fit for modeling in-character knowledge in D&D. In other sorts of games, however, the situation may be different. In particular, in more narrative-focused games the flow of information between the world and the players is fundamentally different. Using a "knowledge" skill might allow a player to specify a convenient detail of the setting. The world depends explicitly on the players' decisions (as distinguished from the actions of their characters). In such a system, a knowledge ability represents the player's power to affect the narrative in certain ways. The mechanics of that skill or ability don't correspond to an in-character action, nor are they meant to. This is a perfectly fine way to run a game, and there are many fine systems that provide detailed mechanics along these lines. If a group wants to run D&D in a more narrative-oriented fashion, that group could come up with house rules for the Knowledge skill that reframe it in narrative terms. Such house rules are outside the scope of this discussion.

In addition to all of the objections described above, there are some mechanical problems with the Knowledge skill. The most obvious is that the skill just isn't very good. With the exception of identifying monsters (which is addressed later), Knowledge skills just don't do very much to justify investing skill points into them. It is rare that a successful Knowledge check will provide any substantial advantage; overall, Knowledge skills are more comparable to Profession than to (say) Tumble, Spellcraft, Use Magic Device, or Bluff. When a player chooses a Knowledge skill for their character, it is usually because they need it as a "tax" prerequisite or because they want to use another ability that refers top the skill (like Knowledge Devotion) -- or, in many cases, because they feel that the skill is important for characterization. This is a major problem with the skill: it imposes a concrete mechanical cost on reasonable characterization decisions without providing a commensurate benefit. If a player wants their character to be an expert in (say) arcane theory, they are obliged to spend skill points on it with no expectation that the skill will be useful. In other words, the existence of Knowledge skills forces players to sacrifice character effectiveness for effective characterization. This is a toxic dynamic that can lead to out-of-character resentment, and it's not justified by any mechanical concern. Some DMs address this problem by giving characters a separate pool of skill points to be used for "characterization" skills like Knowledge and Profession; this is a good start.

The other major mechanical problem is that making Knowledge a skill ties it to the level system in a way that can lead to logical and narrative awkwardness. D&D 3.5, as a system, is justifiably centered on adventuring PCs, but non-adventuring NPCs follow the same basic rules, including the skill rules. This means that PCs of all types have a tremendous inherent advantage over non-adventuring NPC not only in "active" skills like Tumble, Hide, and Sense Motive, but in "background" skills like Craft, Profession, and Knowledge. Or, flipped around, the only way to represent that an NPC is an expert craftsman, professional, or scholar is to give them class levels to accummulate ranks. (You can delay the problem slightly by giving NPCs Skill Focus, but only slightly). That is, as long as skill ranks and net skill bonuses are the measures of knowledge, then the only NPCs worth listening to will be adventurers with real class levels. In D&D 3.5, the standard way of dealing with this is giving noncombatant NPCs levels in Expert or even Commoner, a mechanical kludge that puts an ugly special case in the leveling system.

Admittedly, this is a subjective issue. If it makes sense that anyone sufficiently good at archery must also have a lot more hit points than the average person, then why not anyone sufficiently knowledgeable about arcane lore? To make a distinction, one might appeal to genre conventions: a high fantasy character who is very good at swordfighting, archery, spellcasting, or even tracking or climbing is likely to be a generally competent character to whom "heroic" conventions apply (with a name, likely to recover from injuries, an active player in the plot), but when the hero seeks obscure knowledge they will often encounter a sage lacking further plot-relevant function or ability.


If you've read this far, you'd probably be disappointed if I didn't have some sort of solution to suggest. My suggestion is the culmination of several conclusions drawn from the above.

The first is that the game would be better off without knowledge checks. And if you don't make a check with a skill, then it shouldn't be a skill in the first place.

The second is that given the arbitrary nature of Knowledge difficulties and the newfound lack of a mechanic that interacts with them, it's not important to finely differentiate levels of knowledge about a subject.

The third is that knowledge and other "background" expertise (that rarely provides a concrete benefit when adventuring) is still a valuable part of character-building and deserves a place on the character sheet.

The fourth is that background expertise should not compete for existing resources that provide substantial concrete benefits while adventuring.

Therefore, a simple solution would be to remove all Knowledge skills and allow every character to choose a couple of "backgrounds". The old Knowledge categories would be reasonable backgrounds, as would Craft and Profession skills, Speak Language, and even Perform. Unlike skills, you do not make checks with backgrounds. Unlike 5e, backgrounds don't give concrete bonuses or substantial abilities. Instead, backgrounds are a reference for characterization, a vehicle for exposition, and a target for plot hooks. If a Knowledge skill would have entitled a character to make a check to learn certain information, then instead the DM should just give that information to a character with an appropriate background.

Characters can periodically pick up new backgrounds by (say) spending a couple of skill points. Optionally, each background could have an additional "expert" level (costing a total of two backgrounds) to distinguish a competent student from a master. Further levels of granularity should be unnecessary. NPCs use the same rules as PCs, though in practice you can handwave most of the time. You can assume that an NPC has whatever backgrounds are relevant to their profession, or you can roll a random background to add detail to an otherwise-featureless townsperson.

I'm working on a detailed writeup of this background system, but the summary above is the essence of it. I'd appreciate any feedback or suggestions, especially if anyone's tried something similar in a game.