Mark Langsdorf took a look at the DF thief. My experience with them is that a) my group had one and he was useful, then died and b) they didn't really NEED need to replace him with another thief.
That's despite me using:
- some magic-proof locks (locks with meteoric components)
- effectively bash-proof barriers (doors of metal)
- traps you can't easily disarm or "soak" - either they're too lethal, never run out of charges or energy, or they're effectively impervious to brute force "disarming."
The fact is, even if a thief is needed to get at everything, it's sufficient for most groups that they get at most things, and bypass most obstacles. And a combination of magical detection (See Secrets, See Invisible), magical disarming (Lockmaster), and brute force (bash the door, set off the trap) plus careful use of those and other resources (clever players), and you get past most of these. A magic-proof lock on a bash-proof barrier just means they move on, if they can't tunnel around or through.
Add in the stuff that Mark mentioned - the relative weakness of the template, and the amount of points invested in what work out to be duplication of effort.
So what to do about this?
GM, In Play
As a GM it's incumbent upon you to use those obstacles I listed above, otherwise the thief is useless. Or at least less fun, because everyone else can solve your niche problems, often better than you can. If you, the GM, create situations that can be solved by anyone, expect them to be. If you create ones that require the full complement of available skills, a thief will be useful. This isn't unfair or rewarding a player for choosing a thief - it's a baseline assumption of the game that these things will be in the world and you will need the thiefly skills to get the treasure behind them. If this isn't true, it's bad form to allow a player to make a character centered on skills that are undermined or reduced in utility by everyone else's.
Allowing a thief, and then providing no obstacles that really demand thief skills is like letting people play wizards and not providing things only magic can deal with, clerics but no unholy things to sanctify or exorcise, or barbarians and scouts with no wilderness challenges at all. You're allowing someone to play a character type and not making that type a valid and necessary solution to in-game problems. Again, this isn't about molding the world to the players, it's about not offering them useless choices, or totally redundant. "Oh, you made a thief? Bad choice man, the wizards and fighter-types can do everything you're good at already. Sucks to be you."
As a GM, On the Template
The thief template, in my opinion, would benefit from some changes to make it a little better in my current sort of game - old school dungeon/megadungeon bashing. Mostly it's beating down what works out to be internal redundancy. Rescuing a few points from places that probably don't need them is a good way to get some points to flesh out your chosen niche.
Here are some skills I think you can pare down.
Filch and Pickpocket fill a very similar niche - taking something without being spotted doing so. In DF, it makes sense to drop Filch, and simply say that it's a function of the Pickpocketing skill. Just use the modifiers and contest from Filch when doing the things described there. This frees up 2 points, as well.
Smuggling is one of those skills that doesn't seem to come up that often - and it overlaps thematically with Holdout. It's worth considering dropping this one and either making it a Background skill, or saying that in DF Holdout covers both. Or vice-versa. Merging them or dropping one will free up 2 more points. Personally I think "Smuggling" sounds better, but more rules point to Holdout so I'd keep that name in a merge.
Shadowing is a waste of points if you aren't going to have a lot of town adventures; even if you do, it defaults to Stealth-4 on foot, and you've got Stealth-18 on the template. I think it can go, and drop to optional. That's another 2 points.
Finally, Urban Survival is useful, but a 13 is probably good enough for most uses. Remove 1 point from it and reduce it to Per from Per+1.
All together, this nets 7 points back. What to do with them? Opinions will vary, but it won't hurt to suck less at combat and have more advantages to choose from. Raise the players-choice advantage pool from 30 to 35. Put the other 2 points in weapons - raise the main melee weapon to  DX from  DX-1 and raise the missile weapon from  to  as well - bring them from DX to DX+1, DX-1 to DX, or DX-2 to DX-1 respectively. Your thief is now just a little more combat-worthy, and have some more flexibility to choose advantages.
Another option is to squish down the Background skills pool from 7 to 4, and have 10 more points for advantages - this is better if you still intend the thief to be basically a non-combatant.
As A Player
As always, try to leverage your own skills and abilities. It's not up to the GM to hand you opportunities, you have to make them or take them. Use your Stealth to scout (and if the party wizard or cleric has spells that help that, use them too.) Use your Streetwise in town and your Stealth in the dungeon.
Leave the heavy fighting to others, but learn the backstabbing rules from DF2 by heart, and take advantage.
For more Grey Mouser or Liane the Wanderer (or even Gord the Rogue), you're better off not being a Thief in the first place. You're better off with the Swashbuckler template plus convincing the GM to let you spend 50 of your 60 discretionary advantage points on the Swashbuckler-Thief lens, plus a few points from your Background skill points going to thieving skills.
Thief-Swashbuckler works even better for more of a thief, but the Thief only has 30 points to throw around and the lens costs 50 points. Even a cut-down version of the lens would cost more than you have, just to really keep up. The first seems like a much better option - and it's not like another Weapon Master with a sword and a knife won't be useful to a group. Especially if you can also steal a bit, pick locks, and disarm traps. You can easily improve them, especially since picking locks in DF (unlike in reality) is primarily a question of manual dexterity.
Those are my quick suggestions for tweaking the DF thief to make it a bit more dungeon-worthy, even if it's just a little less rules-compliant than before (in terms of merged skills) or less versatile (if you just drop those skills). It still needs a GM who places challenges that require (or are best solved by) thiefly skills. Llike I've said above and before on this blog, the GM is responsible for making sure the templates and character-types offered in the game actually are required for challenges in the game. If you game lacks sneaking, scouting, stealing, and deft lockpicking action, or makes it so a specialist isn't as necessary as a generalist for doing them, then it's better to just forbid the specialist. Don't punish the player that chose what you offered!
I hope this helps folks who find the thief needs just a little help to be truly useful in DF.