I would say that most parties know some information about most monsters they are likely to encounter. Remove those qualifiers, and it becomes a ridiculous overstatement.
First off, the discussion had trailed to “any given local yokel”, which also seems to overstate the case.
Given core rules, a party of four characters, and an on-CR monster, one party member is fairly likely to have the appropriate Knowledge skill maxed. If the DM does not allow untrained Knowledge, the players likely have taken one rank in every worthwhile Knowledge, as min/max guides recommend, to circumvent that. The odds of a character failing a trained skill check against a DC of 10 + CR are small. If the characters' level and CR are the same, the three skill points over their level, their Int mod, any other bonuses they have, and their die roll put together merely need to equal 10. If the trained character fails or if no one is trained, the party has three or four chances to roll well. I assume that if the check is made, a death attack will be part of the first level of information revealed regardless of how you parse it. I assume that if one character makes this conclusion, he can quickly relay it to others as a free action.
First off, there is the issue of CR vs Hit Dice, already discussed below. Pathfinder modified the rule from HD to CR – that may be where the error comes in, if you’ve also played Pathfinder. The question of advanced creatures also comes in – especially for advanced creatures that may have new abilities. The Medusa’s CR is actually greater than its HD (7 and 6), while the basilisk reverses that (5 and 6), with the abyssal greater basilisk (which has a lot of new special abilities) having 18 HD and CR 12. Bodaks have 9 HD and CR 8. The Banshee (Pathfinder, as it’s not in the 3.5 SRD) has 19 HD and CR 13. As CR enhances, HD seem to grow faster than CR.
So if we assume the party has every knowledge skill maxed out, that’s 9 skills. Leave out architecture, geography, history and nobility/royalty (the ones with no linked monster types) and we’re down to only 5, which seems manageable.
At L6, that’s 9 ranks + INT bonus. Wizards tip the scales high on INT bonus, and can have every Knowledge, but even a 20 INT Wizard has only 7 skill points to spread around. Make our Arcane caster a sorcerer both skill points and INT bonuses drop off. Wizards tend away from Local, Nature and Religion, in my experience, letting someone else cover those bases. Our group likes skills, so INT 12 – 14 is pretty common. Let’s assume a +2 to err on the side of generosity, but not assume a wizard. One character on the team probably dumped INT (8 or 10) too.
Our current Pathfinder team has a Cleric with Spellcraft and a Sorcerer with Kn Arcana (and the latter doesn’t cast Detect Magic), and I think no one with Planes. That’s a bit off topic, but one anecdote of a 4 character group where no one has a stellar INT, and at least one Knowledge is missing. The Cleric is INT 14, mainly due to a plan to be an item crafter.
So an average roll will be +11 (9 ranks and +2 INT) at 6
th level. That’s a 5+, so 80% likely. If we’re L13 facing that Greater Basilisk, now it’s +17 versus CR 19, so only 45% likely and those untrained rolls aren’t going to cut it. +18 on that 19 HD banshee is 50/50.
So we have a shot at recognizing its abilities. Much more helpful if rumours tip us off to what may be in there than if we have to get within light source range to ID the creature – most common sources (light spells, everburning torches) get 20’ good light and 40’ shadowy light. That’s only 10’ out of that 30’ range if we assume dim light is enough and the front guy has the light source. And this ignores doors. I think the CR’s also assume we’ll be close enough – that’s one reason why SoD monsters have low HD for their CR’s.
You keep using the medusa, where averting one’s eyes and keeping your distance allows her to fire her short bow effectively, or close for that rather nasty poison attack. Meanwhile, since you can’t see, you get a 50/50 miss chance and can’t target properly.
Still, the odds of having at least some knowledge are pretty good. Full knowledge is less so, but making the Medusa check by 5 (and getting both the petrification and poison) is reasonably likely, and you can see the shortbow.
Overall, though, I think the party has a decent chance of knowing a bit about the typical creature encountered, but often only when they are already in pretty close proximity.
In general, players should do that. I see no problem, however, with a player doing something clever on occasion. If the roll justifies it, he gets rewarded. If not...nothing ventured, nothing gained. In any case, I think it's for the player himself to decide what his character could reasonably think. How many great stories have unimpressive people making one key insight or saying one trenchant phrase? Heck, that's a JRR Tolkien special!
Very few Tolkein characters are min/max’ed combat machines. They tend to make friends easily and be pretty quick on the update. They aren’t 8 INT, 8 CHA min/maxed meat grinders. And I don’t discount the possibility that the 8 CHA character will get lucky and happen to strike the right phrasing with the right target to get a success (20 – 1 = 19 Diplomacy check, after all). But I don’t believe that the player being a good speechmaker should move him from a -1 penalty to a +3 bonus on a regular basis. If you want to play a persuasive, smooth talking character, then don’t dump CHA and put no ranks in social skills. Your character doesn’t deliver the message as smoothly as the player if he lacks the skills and stats to back it up and, by the same token, a stuttering player with no social graces does not translate into the PC with 16 CHA and 8 ranks in Diplomacy suffering similar drawbacks to the player, calling the king “that dude with the crown” and spitting when he talks.
I think it's simply a consistent enforcement of the heart of the d20 system: your bonus is how good you are, the DC is how hard the task is. If you beat the DC, you accomplish the task. The bonus and DC numbers determine the chance of that happening. I think that is perfectly adequate and does not need an exception for trained only skills. Even given the conceit of trained only skills, I think Knowledge skills are some of the least appropriate skills. As any ENWorlder knows, people can recall a lot of random factoids even outside of their expertise.
I see no issue with trained only skills – there are lots of other areas where some characters can succeed and others can’t. The d20 system (like most/all game systems) abstracts a lot, and that takes out many statistical outliers. PC’s don’t slip, hit their heads and die, and they don’t get that random smattering of odd factoids.
The discussion about knowledge checks brings up one of my pet peeves about 3E: that it's really hard to gain information. That kobold in front of you? He could be a 30th-level wizard. He could be a regular kobold. Without Arcane Sight, you don't have a way of knowing.
I'd like to incorporate a skill check to tell the players that sort of information. I know that Oriental Adventures has something about that for Sense Motive, but considering how important that information is, I don't want to place it in any one skill. And how do you know how powerful a wizard is, before they cast a spell?
Knowledge: Local should give them reputations of the inhabitants with histories -- including a power range. "That's not a kobold; that's Relnick the Capricious. It's said he killed a griffon mid-flight with but a single word. With him is Kulc the orc. Three years ago, he took down an entire company of Knights of the Golden Rose by himself."
I like the idea of removing Humanoids from Local Knowledge and moving them to, say, Nature, then making the Local Knowledge skill provide intel on creatures that are prominent locally, and the specific local residents of note. The problem with “local knowledge” is that players travel around a lot, but it seems fair to interpret this skill as including the ability to pick up a lot from local gossip, tavern talk, etc., so it doesn’t take long to acclimatize to a new locale. Geography also seems plausible for residents of note. History could be appropriate for long-lived entities, and Nobles for persons of both rank and special note.