In fact, if you took that twist mentioned and used it in another adventure with the same players then you have done it.
I don't know why you keep insisting I've done it. I've sadly never been in a position where I stayed in one area long enough to run two campaigns for the same group. One of the reasons I've never gone much beyond 12th level or so in any edition is I'm not even around long enough usually to bring a campaign to a complete close. I've moved around maybe 11 or 12 times in 36 years.
The fish event, did you have the NPC already written as a connoisseur of fish? If not then you've done it, right there.
With the name of an NPC being 'Fishbandit', it really wasn't much of a stretch.
Would that have broken the trust? Are you saying that it would have been a bad thing? Would it have broken the game?
As something that has no real bearing on the games structure or balance, then probably not. One very important difference, I'd be writing into open undetailed space. I'd also be making arbitrary and roughly trivial choices. And as I've said, winging it on a blank page doesn't strike me as being the same class of thing as changing your ideas on the basis of what you overhear the PC's saying.
I think you are veering out of the thread by extending the idea to things other than 'chumming'. If a player says, "I want to find a dealer in furs, is there a furrier in this town?", and I've made no decision on that, then either answer is acceptable provided that I make that decision in a neutral referee stance and not adversarially. There is no 'chumming' involved in this, and if I think to myself, "This is an area that likely has alot of fur bearing creatures, it would make sense for there to be a fur industry in the area.", I'm not 'chumming' in the sense it is being used elsewhere in the thread. But if for whatever bizarre reason I've made the decision that there is or is not a furrier in town, then the answer had better be whatever I'd previously decided because the only reason I could possibly have for changing my mind is some sort of metagame one. And any time you start changing things for metagame reasons, beware.
While I haven't fudged in this way, I have fudged before and every time I do it, no matter how good I think my reasons are, I find I almost always have cause to regret it later.
Now, compare these trivial matters we've veered off into with the selling point provided by the OP, and the writers he quoted, and at least half the people in the thread. The inspiration for the very name 'chumming' is 'blood in the water'. The idea behind chumming, the very reason it is being recommended, is the DM gets to use the players devious, scheming, dastardly speculations against them. The metaphor is the DM as the bloodthristy shark, chasing down the players and shredding them. If you can't see how that is an adversarial metagame stance, then I don't know why I've bothered saying anything in this thread.
Don't get me wrong. I'm absolutely not dismissing your DM style. Not one bit. What I'm failing to see eye to eye with is the sense that you seem to say that others only use such a thing to paper over some mistake or that it is somehow, even used lightly or just once, a break of the trust between DM and players.
I think every time the DM retcons something he prepared, he risks breaking the trust of the players. I think this falls into the broad category of 'DM fudging', which includes railroading, fudging dice roles, having NPC's suddenly suffer drops in IQ to keep players alive, and what we are now calling 'chumming'. Every kind of DM fudging depends to a certain extent on maintaining the pretence that it isn't happening. For example, there are times when I might pull a railroad out of my toolbag. An example would be I have this hexcrawl exploration game going on, but there is something on the map which I always no matter what want the PC's to find. I might decide that the best way to handle this is to not put it on the map. Instead, I might decide that the encounter is always on the third day of exploration. That might be a good idea to balance the pure simulationist game I'm doing with some sort of narrative consideration to give structure to those players who want more story and clearer hooks. But I'm also risking breaking faith with the players, who have a reasonable belief that when they turn right that choice will result in something different than if they turn left. So, I'd recommend pulling out a railroad very rarely as well.
Pretty much anything that requires you to engage in a high level of illusionism or out of game deception is not something I'd recommend as your normal and expected method as a game master. You won't break trust with your players every time you chum and there might even be times when its the right thing to do, but I don't think it should be upheld as some sort of clever 'best practice'.