Elder, you are still missing the boat a bit here. Let me quote a bit from the skill:
A Bluff check is opposed by the target's Sense Motive check. Favorable and unfavorable circumstances weigh heavily on the outcome of a bluff. Two circumstances can weigh against the character: The bluff is hard to believe... <snipped as it relates more to feinting rather than lying>
also
Generally, a failed Bluff check makes the target too suspicious for a bluffer to try another one in the same circumstances.
A successful Bluff means the character
believes the bluff. That includes PCs.
Now, if there is independent evidence that the bluffer is lying, then a heavy circumstance penalty should be imposed.
In the present case of a PC lying about magic items, eventually the evidence will pile up to the point that the circumstance penalty become something like -100 the the bluff check (simply impossible for normal mortals).
In the case of a group chasing down a suspected murderer, if the DM sets the cicrumstance penalties appropriately and he
still succeeds, then the PCs believe him (at least until new evidence comes forth).
That's the way the skill works. It's up to the DM to decide on extenuating circumstances and set circumstance penalties (or bonuses) appropriately.
Remember:
" Favorable and unfavorable circumstances weigh heavily on the outcome of a bluff. "
But the outcome is
not "he seems to be telling the truth." It's "your PC believes him." Of course, that just means your PC believes him
now.
If the players then remember some other fact that makes his story fall apart, fine. That could be considered new evidence, and changes the circumstance under which the lie was believed.
If you water down the skill to the point that a successful bluff doesn't really mean anything, then why take the skill? No, that's the wrong way to go.
This skill can be troublesome - it's
heavily reliant upon the DM setting appropriate modifiers.
If you are trying to influence behavior, then:
If it's important, the DM can distinguish between a bluff that fails because the target doesn't believe it and one that fails because it just asks too much of the target. For instance, if the target gets a +10 bonus because the bluff demands something risky of the target, and the Sense Motive check succeeds by 10 or less, then the target didn't so much see through the bluff as prove reluctant to go along with it. If the target succeeds by 11 or more, he has seen through the bluff (and would have done so even if it had not entailed any demand on him).
A successful Bluff check indicates that the target reacts as the character wishes, at least for a short time (usually 1 round or less) or believes something that the character wants him to believe.
So, in the case of the murderer trying to redirect the group's investigation, it's entirely possible to get the group to believe him and yet fail to influence the group to investigate elsewhere - which leaves the group in the immediate area to either stumble upon new evidence or simply mull over what he said and realize they've been duped.
All within the skill as written.
Of course, successfully bluffing an entire group is hard in the extreme - only one person has to be unconvinced to start the whole thing unraveling. This can be represented by making new Sense motive checks with a new circumstance bonus.
Again, all this can be done without watering down the skill one bit - but it requires pretty active DM involvement.
Finally, it is well within the DM's discretion to simply state that the attempted bluff is impossible. Though I tend towards allowing it
if the player can come up with a story that is at least plausible. Such a circumstance may
require 10 ranks (as a nearly impossible skill) with a circumstance penalty of 20 on the bluff - nearly impossible, but still could succeed against a select few.
Those that max out bluff along with the charisma to support should have a chance of pulling off the Big Lie - even if its a samll one.
Even at that, a 10th level characer coudl normally have, say 13 ranks plus 4 for charisma for a total bonus of +17. At -20, they would roll with a -3 modifier against porbably a o or +1 modifier.
They will succeed a good portion of the time. Of course that's for something that started with an at least barely plausible story. For the totally implausible I'd simply declare it impossible to succeed.
The current problem in this party will indeed self-correct, but it might be good to have a talk with the players before that - the players may start to get upset with each other over this kind of behavior.