I'm not sure anyone thinks the abilities should always work. The issue is what that failure rate should be, and what the bar for reasonableness is.It feels like the thread is stuck in something vaguely like the following (please edit in your head as needed). I wonder if we'll get an AI someday that can enforce a stalemate for a thread when it finds an (essentially) repeated set of moves?
A: X always works
B: No, it doesn't, that's ludicrous
C: Why wouldn't it work?
B: No, it can work sometimes but certainly not always.
C: Why can't it work the vast majority of the time?
B: Why should it work in crazy examples!?
C: Why are you hung up on crazy examples?
<iterate>
B&C: So X should always work when it's reasonable, in iffy cases when there's a reason, and should not be expected to work in the virtually impossible cases
A: X Always works!
<go back to line 2, possibly with B and/or C subbed out for B+1 and/or C+1>
Or is there some general disagreement (beyond a person or two) about "X should always work when it's reasonable, in iffy cases when there's a reason, and should not be expected to work in the virtually impossible cases"
Personally I can almost always find a way to justify such abilities, or a way to play through the justification of such abilities. And I think the GM should be looking for a reason to say yes rather than no. The failure rate I would expect to see is south of 1 in 10, and it's where we're in extreme edge cases like 'you're in a locked box with no way to contact the outside world' - stuff that any decent player wouldn't try anyway.
I think the 'other side' defines what's reasonable differently. They assign such abilities a much higher failure rate. They apply a much stricter reading to the conditions in the text. They require the justification to have been pre-agreed as part of the character's history rather than revealed on the spot. They require the usage to meet with their own ideas about cultural attitudes, shipping charts, underworld etiquette, bird migrations, weather conditions, etc. They aren't looking for a reason to say no, necessarily, but they are at best neutral about it. If the player's idea of what should happen isn't a match with the GM's, then the GM's ideas should automatically take precedence.