It's great that you have enthusiasm, but here's the thing: you need to decide how much effort you're going to be able to put into it. As the previous three posters have already mentioned, Old Pochmann is a great way to begin learning to solve BLD. It's simple, intuitive, and requires the memorization of only three algorithms (T Permutation for edges, Y Permutation for corners, and R Permutation for parity). If you want to begin- that's the place to start. Learn how to do that, it should be pretty easy, but then to become faster, turn your attention to other potential methods.
M2/R2 (or perhaps just M2) is the logical next step up from Old Pochmann. You said you didn't know the difference between the methods. In Old Pochmann, you solve the pieces one piece at a time by using setup moves, and then a T Permutation (for edges, 14 moves) or a Y Permutation (for corners, 17 moves). See the links above for more details on that. In the M2 method, instead of using a 14 move algorithm each time, you simply execute one move (M2). Things are only slightly more complicated when the piece you are trying to solve is in fact on the M layer, but it isn't particularly difficult to get used to. M2 for edges in combination with Old Pochmann for corners is probably one of the best combinations you can get at for a one-piece-at-a-time solve. R2, the M2 equivalent, but for corners, uses a single move, R2, instead of a Y Permutation for corners. However, it requires the memorization of several additional algorithms for all the corners on the R layer, which makes it not quite as efficient as the M2 method.
If, for some reason, you don't like the one-at-a-time approach, 3OP is the closest to a beginner's method. 3OP divides solving the cube into four stages rather than two (orientation of edges, orientation of corners, permutation of edges, permutation of corners rather than combining orientation and permutation into one step). However, it solves two pieces at a time rather than one by using a three cycle of edges or corners (U Permutation or A Permutation). In my opinion, however, 3OP is more difficult for a beginning blindfold solver than a method like Old Pochmann.
If you're comfortable with one of these methods, and you're still very serious about wanting to get faster, then I recommend looking into a method along the lines of TuRBo or BH. TuRBo solves two pieces at a time, while orienting and permuting at the same time, by using setup moves to get the necessary edges or corners to the U layer. It requires the memorization of 26 algorithms, if I remember correctly (18 for corners, 8 for edges). That's a good way to go about for pure memorization of algorithms. If you really want to understand what you're doing though, rather than just executing algorithms, the BH method offers an understanding of the Rubik's Cube that really changes the way you think about the way the pieces interact with each other. It is also extremely fast, as it requires no setup moves (except, of course, for setup moves within the commutators, such as the A9, Columns, or Orthogonal, but even then the A9 and Columns setup moves cancel out). It does have 378 cases, so you could think of it as having to learn 378 unique cases, or you could think of it as being able to create 378 algorithms as they are needed.
So, quick recap:
If you want the easiest possible method- look towards Old Pochmann. That's the best place to begin.
If you want to get faster after that- look towards M2 for edges and Old Pochmann for corners. That would be a good permanent method to use.
If you're willing to memorize a few additional algorithms that won't serve any purpose outside of the method- M2/R2 is a possibility that you might want to consider.
If you want to begin a three-cycle method, you might want to look at 3OP, though M2/Old Pochmann might be faster if you've practiced it more.
If you want a more advanced three-cycle method, and are willing to learn 26 algorithms (not a large number, really, only 5 more than PLL), look at TuRBo.
If you want a advanced, true understanding of the way algorithms work, and how the pieces interact with each other, how to execute commutators, along with being much faster at blindfold, BH is the method for you. Though, let me give you warning, don't jump straight to BH until you have plenty of experience with other methods and have tried other things first. You want a solid foundation to build BH on.
But let me remind you, being fast at blindfold solving doesn't just include the method. The method of solving is only half the battle. The other half is finding a good memorization method that you can do quickly and easily. I'm not going to get in depth into memorization right now- but I'll just say that using letters, numbers, or visual memorization is a fair easy way to begin solving.
Hope that was helpful. Let me know if you have questions.
