My Traveller GM at the time (one of the rare periods when I regularly got to *play* instead of GM) converted part of our heavily-modified CT game over to a heavily-modified MT game, because he loved the MT task system. It quantified all of the tasks into a definable system. (The fact that he's a computer programmer probably contributed to his affinity for the system.)
As a *player*, at first I kind of liked the task system, because I knew what I needed to roll to succeed. The GM no longer needed to improvise numbers - and, frankly, he was inconsistent from session to session on what those numbers *were* - and everyone was "happy."
He also modified the system for "middle" difficulties, which was nice, and it went something like:
3 = Easy
5 = Simple
7 = Routine
9 = Hard
11 = Difficult
13 = Very Difficult (hey, we didn't have online thesauri in those days)
15 = Formidable
17 = Very Formidable
19 = Incredible
21 = Stupendous
23 = Stupid Player Don't Even Try It
Anyhow, as time went on, I found myself becoming.... je ne sais pas.... nostalgic I guess for the non-task system, and that's because the "mystery" of a really good roll [or a really bad one] was gone.
You roll, you compare to the target number, you succeed, fail, or have a mishap. Period.
In the pre-task system, if we rolled a really good number (say, an 11 or 12), the GM would usually have it mean something "extra." An 11 with Engineering-3 and an Int of B is a 16, and if you're shooting for a Very Formidable task, you missed. Nothing special here, move along.
Over the years, though, as I ran various Traveller campaigns, I saw the task system as a very helpful tool. However, I stopped
sharing that tool with the players. They might know, if they had read the books, what the general difficulty classifications were, but they didn't get to see them.
Oh, I'd let them know ahead of time if the task was (Fateful) or (Hazardous), but if the consequences were hidden (the alien device was just *waiting* to explode) I might not. And if they were in a position to judge difficulty, I'd certainly give them an idea. They also knew that to shave time raised the difficulty and taking longer lowered it (usually), but the specifics remained behind the screen.
That's how I run D&D and how I'll *be* running T20 - I don't tell them what they need in more than general terms.
This allows for several controls, as the Game Master, that I lose if I use a straight-up task system that's "public" information. First, I can fudge. Yes, yes, I know, that's a sin against humanity, but as the GM I'm interested in moving the story along, not in killing someone just because the dice say so.
Now, as some of my players could attest, if it's important for someone to die - or be made an example of - they die. But I digress, so back to fudging rolls (or target numbers). By modifying the result, I get to have things happen the way I want them, and that builds a good campaign story arc. If the players are constantly rerolling characters [exaggeration], that's not good for the arc.
The second control I gain is uncertainty. The players are more wary of a task - which heightens tension - than if they know what the difficulty is.
What you, as a GM,
***DON'T*** want to do if you're working like I do, is create a feeling in the players that they have no control whatsoever. They should be confident that their highly skilled character knows what he's doing and won't step on the twig right before he clobbers the Stormtrooper. It still might happen - everyone flubs a roll sooner or later - but they shouldn't expect it to happen. The bad result - just like the spectacular result - should be a surprise when it comes, and create an emotional response.
"Oh, crud muffins!" instead of "Well, it happened again; I hate this character, he's so pathetic."
To sum up: The task system (UTP or otherwise) is a useful guide and tool, and I am generally using it behind the screen, but it stays there. More important is to create a believable pseudoreality that makes it fun for the players, and doesn't drag them into the game mechanics.