I'm debating whether or not Mystic Powers should have Difficulty Level Types, such as "Helper", "GM Fiat", "Player Gambit", etc.

When Mystic Combat occurs we know that it runs like regular combat and in the case of combat skills we use DL Type: Combat, but in the case of Mystic Powers we use Mystic AC and Mystic AL instead of AC and AL. So that's a DL Type of Combat. But not all spells and miracles use Mystic AC as the Difficulty Level. For example Invisibility.

There's several ways to approach Mystic Powers like Invisibility.
1) Don't roll for success, just grant it.
2) Roll but only to see if it's a critical success or a fumble. Otherwise it simply works.
3) Roll using one of the Difficulty Level Types that applies to the Power. In the case of Invisibility it would be a DL Type: Helper.

The only reason I'm considering using DL Types for Mystic Powers as a default (GMs can of course choose to not use it for their own Worlds) is to keep the system cohesive. In other words, Mystic Powers would then work exactly like Skills, and I think that might be easier to understand. On the downside, it also makes using Mystic Powers a bit more complicated as people need to understand the different Difficulty Level Types. Though they are reasonably clearly written, and some are quite easy to understand (such as GM Fiat), others are more complicated (such as Player Gambit, and Helper), and so it might take people a little more time to get used to it.

Thoughts?

I think the best systems are cohesive. AD&D was fun in its time, but it had different methods to determine success for different situations:

-Better armor class values were negative. Combat and saves were resolved on a d20. Damage used various dice.

-Thief skills were resolved on percentage dice, and there were no modifiers to the % chance for success.

The closer a system comes to one golden rule, the better it is, because players can focus more on play.

The DL for any conflict resolution should always be whatever makes sense according to the gaming group. Ideally, I don't ever want to look up a rule when I'm playing. Although categorizing DLs as you have is helpful for explaining how they are selected, I would not dwell on them.

As far as rolling for success goes, I want to see less rolling in role-playing for trivial tasks. Rolls should only be made for critical/interesting situations. It takes up too much time otherwise.

I think the best systems are cohesive...

As far as rolling for success goes, I want to see less rolling in role-playing for trivial tasks. Rolls should only be made for critical/interesting situations. It takes up too much time otherwise.

Yes, I tend to agree with you about cohesiveness of the rules. For one thing, if the rules remain consistent, even if they are slightly complicated, you at least only have to learn the one way it's done, and after that you don't need to look the rule up again. The plethora of different rules mechanics in the original D&D drove me nuts back when I first encountered it in 1978, which is why Elthos was designed with one centralized GRM to begin with, and I've tried to make everything that can function the same way, do so. Hence non-Combat Skills and Combat Skills work in essentially the same way.

As for your second point there ... yes, I think that's a darn good point, too. Maybe one way to handle that is for the GM to simply arbitrate when a roll is significant enough for the Player to roll it, and otherwise he rolls it himself. An example might be for Listening where the GM knows that nothing significant can possibly be heard, he simply rolls to see if anything at all is heard, and if so it's nothing much anyway. In these cases the Character would not get Experience Gains for the success. However, if the GM knows that it's critical to the story whether or not what is being said is heard or not, then the Player rolls, and if they make it, great for them, and they get Experience.

Would something like that work better?

When I run a game, I make many crucial rolls behind the scenes for the players. Telling a player to roll a die to listen, or rolling a die to listen because the player asks to do so breaks immersion. This argument is jumping to the concept of hidden die rolls, which is a little off topic, but I'll write about it anyway.

If it's a passive (the player doesn't have to choose to use it) skill, such as listening, I will hide the die roll; the player may not even know what was checked. If it's an active die roll, I will let the player roll. I do it this way for two reasons. The first is that if the DM is checking to see if a PC detected a thief shadowing them, the player is not tipped off. The reason why the player sees the active dice roll is because the player should receive information from the content of the roll. For example, if the character is climbing, the player rolls a 6, and the character still fails, its clear that the task is daunting enough to look for another solution.