When you make the decision about who all the goblins have captured, would you make the same decision to say that this particular mother-of-a-wizard is captured, if the wizard in question had been an NPC rather than a PC?
Of course not! The mother is of no interest as a topic of the fiction except in her capacity as a mother of the PC.
If all of the PCs was an orphan, I wouldn't worry about whose parents the goblins might have kidnapped. I'd write goblins with other sorts of motivations do to other sorts of things relevant to the PCs (and hence the players, and hence the game).
Are you letting your out-of-game knowledge about which characters are PCs and which ones are NPCs affect the choices which your NPC goblins make?
Well, I'm letting it affect the motivations and actions I author for the goblins. The effect is not on the goblins (they are imaginary, and are only affected by imaginary things). The effect is on me - it makes me author the goblins in a certain way.
NPCs are perfectly allowed to make decisions based on what they know about the world, even if that includes the PCs; they just can't make decisions that take into consideration that the PCs are PCs.
It's meta-gaming whenever any character (PC or NPC) makes use of knowledge which it should not otherwise possess
<snip>
When a goblin decides to kidnap the PCs mom instead of some unrelated person, or even to attack the PC's home village instead of some other village, that is a meta-game decision which the DM is not allowed to make.
The goblin doesn't make use of knowledge it doesn't possess (in the fiction).
But I, as GM, use information that I possess in real life.
Hence, I can easily write into the fiction a goblin who has some reason or other to kidnap the PC's mother. (In my campaign, the goblins had kidnapped a whole lot of villagers and townsfolk, and the mother happened to be one of them.)
Sometimes I don't even work out the details of the fiction in advance. Rather, I follow
Paul Czege's practice:
I frame the character into the middle of conflicts I think will push and pull in ways that are interesting to me and to the player. I keep NPC personalities somewhat unfixed in my mind, allowing me to retroactively justify their behaviors in support of this.
For instance, if it seems like it might be interesting to have the goblin notice the family resemblance; or, in some other way, to have or use knowledge about the relationship between the NPC and the PC; then I might do that. If not, then not.
Do the consequences follow logically from the action? Could they have predicted that this would have been the outcome, by asking the right questions and possibly performing the right tests?
To the fist question, yes. To the second, dunno. That's not very important in the way I run the game. It's not mostly an investigation game; it's a game in which it has turned out that the main issue confronting the players is how to react to the pending Dusk War, and the different choices as to whom to support (primordials, the Raven Queen, other gods who want to build the Lattice of Heaven, no gods at all, some sort of balance of forces, etc).
Given that this choice can't even be posed unless most of the information is on the table, the notion of "asking the right questions" or "performing the right tests" doesn't have much work to do.
As long as all of those traps and monsters existed in both rooms beforehand, then that's fine. When it becomes pointless is when you start changing things, after the fact, in order to mess with the players.
After
what fact? Before
what? Gygax - who was a pioneer of the "skilled play" paradigm - didn't write all his dungeons before starting the campaign! (I mean, how could he?)
He wrote ToH deliberately to test certain players. He wrote the Fraz-Urb'luu room in Castle Greyhawk because he thought a particular player would find it fun.
If the behir in room 1 dies in the first round due to a lucky critical hit, or if the hidden parchment detailing the Baron's nefarious dealings is incinerated before anyone can notice it; then adding in another monster in room 2 where there was previously none (whether or not that had yet been revealed to the players), or putting in additional evidence to implicate the Baron, specifically to negate the unexpected outcome of room 1, would be shenanigans.
With the monster example - it depends entirely on what the "unit of play" is. If it's
one dungeon expedition, then introducing the new monsters after the PCs withdraw is fair. If plotting changes in the dungeon population is itself a part of the game, then making such changes because they're "fun" would be an issue. This might depend on such things as what sort of detection magic, or conventions around scouting and rumour gathering, are in play - hence it will be highly campaign- or table-specific.
As far as the baron example goes - that is so far from how I run my game I can't really comment.
Some particular group of adventurers can be sent off to explore a certain region, or complete a specific quest, based on the estimated difficulty of the task at hand at hand; you don't send a bunch of rookies out to map Dragon Island, but maybe you send them to get rid of a goblin infestation.
As a DM, what you cannot do is to replace all of the dragons on Dragon Island with kobolds, just so you can send the rookies there, because that would be meta-gaming.
I don't understand. Where in Plato's heaven do I look to find the true nature of Dragon Island?
In my 4e campaign, I had planned to run the
Demon of the Red Grove scenario the first time the PCs went to the Feywild, because I thought it was interesting in itself and it would drive home certain aspects of relevant campaign story around Corellon, Lolth and the Raven Queen. I first wrote it up for low-paragon tier PCs (with the demon somewhere around 13th level). By the time I actually got a chance to run it, the PCs were mid-epic, and I reworked the level and capabilities of the demon appropriately.
I think that counts as GM meta-gaming.
I don't understand why I
can't do it: the Red Grove and its Demon have no existence or nature outside of my authorship of them as part of the campaign backstory, and when I introduced them into the shared fiction they were what they were.
I also don't understand why I
shouldn't do it. It produced interesting and somewhat important play - besides fun colour, like the sorcerer getting to show off his bardic skills by singing a song to defeat the evil cries of the demon, it forced the PCs (and thereby the players) to consider whether and how to bargain with a demon. The meaning of all this is not any less because it occurred when the PCs were 10 levels higher than they might have been.
If the point of play was to explore some imaginary world pre-authored by the GM, then there might be an issue (given that I have not pre-authored any such world). But that's not why my group plays RPGs.