One of the interesting peculiarities about the Supplement I (and Holmes) thief class is that, while it possesses a "remove traps" ability, it does not possess a "find traps" ability, which wasn't introduced until the AD&D Players Handbook and also adopted by the Moldvay rules. (As an aside, it's worth noting that the Cook/Marsh Expert Rulebook lists only "remove traps" as a thief ability on its class tables, although one presumes this is just an editorial glitch). I surmise that the thief lacked an explicit "find traps" ability, since it was simply assumed they had the same chance to find them while searching as any other character class (1 in 6). Thus, thieves were adept only at overcoming traps once found but lacked any special skill in finding them.
In playing my Dwimmermount campaign, what I've noticed is that what I enjoy most about traps is not the question of whether the PCs will find them -- they generally do given enough time -- but how they go about avoiding them. The players have been very ingenious in the methods they've employed to defeat many of my traps. They've really gotten into the spirit of things, describing precisely what their characters do, often employing common adventuring equipment like iron spikes, 10-foot poles, and rope to sidestep hidden dangers. When they can't do this, they look for ways around the traps, by taking other passages or doors. It's really amazingly gratifying to see and a vindication of my experiment in old school archeology.
Consequently, I'm starting to think that it's not the thief's "find traps" ability that bugs me, but rather its "remove traps" ability. As I said, with only a couple of notable exceptions, the players largely intuit when traps are afoot; perhaps my dungeon designs telegraph the presence of traps too easily. Regardless, I don't find much fun in having the characters harmed through random trap damage because they failed a 1 in 6 dice roll. That's not to say I don't do it, of course. Several hirelings have died in Dwimmermount because of undetected traps, for example, but it's not something I derive any satisfaction from. On the other hand, watching the players grapple with a mysterious trap, trying to figure out what it does and how they might avoid its effects is something I do greatly enjoy, particularly when they figure out how it works only after they've accidentally sprung it.
I'm not entirely sure what all this means just yet. Although Brother Candor's henchman Gaztea is a thief, complete with a "remove traps" ability, we've actually not been using that ability at all, sticking with the traditional method of having the players describe what their characters do to avoid the traps. Even after 16 sessions of this approach, we haven't tired of it and, I think, prefer it to the traditional AD&D/Moldvay method of handling trap removal with a percentile roll. Granted, it means that the vast majority of the traps I place in Dwimmermount have to be explicable, which is to say, I have to be able to describe their components and context in such a way as to enable the players to find a way to defeat them. That's limiting on some level, but it also ensures that I can't produce "traps" that are little more than game mechanics waiting to interact with the "remove traps" ability. Dwimmermount's traps have to work, if you get my meaning.
So, while my stance on the thief class continues to soften, my opinion is still very much in flux and I'm finding a lot of my biggest dislike of the class, purely from a mechanical standpoint, centers around the "remove traps" ability.