Everyone who's played Braid seems to agree that it's pretty hard. I myself was able to finish it without too much trouble, although there were a handful of puzzles I found frustrating. There are others I've spoken to who are having a much rougher time with it, complaining that it demands too much of the player too quickly.
Braid is definitely a hardcore game. It is brutally difficult at times, both in terms of puzzles and platforming. On the other hand it is a very beautiful, unusual game that one imagines might appeal to a diverse audience. The intro especially seems to suggest a casual aesthetic, with minimal instructions and simple controls that ease the player into the experience. However, it betrays this simplicity quickly by escalating difficulty at a steep rate.
It's not simply that Braid is hard. It's that Braid reverses player expectations so continually and so rapidly it gives less determined users almost no time to build a stable foundation of competency. Put another way, it teaches the player a solution once and then immediately undercuts that solution in the next puzzle. For example, in World 2 there is a puzzle where you must rewind time in order to open two separate doors with the same key. This puzzle is clever and takes some figuring out, but once you get the principle it feels rewarding. However, the next puzzle actually punishes the player for employing the same strategy. In the new puzzle using the same key on two different doors actually breaks the puzzle, forcing the player to restart the level.
The second puzzle is a devious riff on the first, a trap of sorts set by the designer that forces the player to question their existing mental model and adjust. On one hand this is excellent game design, since it keeps astute players on their toes. On the other, it expects the player to comprehend, internalize, and adjust to new mental models with zero iteration.
Whether or not this is a flaw of Braid is an interesting question to consider. From the point of view of my own game experience I can't say that it is, but from the point of view of many players it might be. I certainly don't think a gentler difficulty curve would have hurt the game, but perhaps Jonathan Blow felt he didn't have the luxury of easing players into his puzzles as much as one would in a longer commercial game? I can easily imagine how an indie developer, having lived with a game design for several years, would want players to experience the true depth his core mechanics afford. It would be disheartening to spend all that blood and sweat and then just give players 25% of the complexity you know your system can support. On the other hand, an indie game of modest scope means the game will likely be short, which further means that the ramp up from easy to hard puzzles will be extremely steep. Given that Braid has only 35 screens, it has no choice but to up the ante significantly between puzzles in order to reach its peak by the end. There are only two ways to alter this curve: lower the peak or lengthen the game.
As an indie developer functioning on scant resources, Jonathan Blow perhaps didn't have the option of lengthening the game. It seems fair to assume, though, that he had the option of lowering the peak and chose not to... assumedly to give users the "full experience" of Braid. While I can understand this, it is a decision that may have cost him some players. Some people just can't scale a cliff that steep.
All this makes me wonder whether there is a paradox in indie development. Indie games in some ways can take risks that bigger games cannot, and they can attract game designers who have strong personal statements to make. But these strong personal statements may bump into a wall when the limited scope of indie games "forces" developers to choose between depth or accessibility.
Even if this paradox exists (and I'm not convinced it does, but it's intriguing to consider) I think there are some clever ways around it. Braid in particular I think would have benefited greatly had Blow taken a page from Miyamoto and made only a percentage of the levels mandatory. Mario 64, Mario Sunshine, and Mario Galaxy all employ an excellent system by which players only need to finish about 60% of all the levels in the game in order to unlock the final level. This means that, if players choose, they can skip the harder levels and play mostly the easier ones and still finish the game. Going back and finishing all the levels afterwards becomes a more hardcore task that only players of a certain level of dedication will do. These games successfully appeal to both kinds of players without flatly sacrificing content.
Braid is so modular that it seems like a similar system could have been implemented with virtually no change to the current game. I don't understand why Jonathan Blow felt it was necessary to force players to gain every single puzzle piece in order to even have the option of finishing the game. Braid basically forces players to 100% the game on the first play through, which seems needlessly demanding. To my mind, making some of the puzzle pieces optional for completion would have not only made the game easier on more casual players; it would have enhanced the invitation to interpretation that characterizes Braid, motivating players to go back, after the ending, and find more puzzle pieces to unlock the mysteries of the story.
Perhaps not all these choices were up to Jonathan Blow? I'd be interested to hear what he has to say about why Braid was balanced the way it was. I can only speculate of course. Still though, I think this is proof that Braid is an extremely useful game to analyze. It seems like future indie developers have alot to learn from it, both good and bad.