I guess I fall under the "play to learn" camp, but I'm mainly writing this to try to clear up some problems with semantics and application of terms I think people may be having.
I think "playing to learn" has two different ways it can be applied: in terms of technique and in terms of strategy/mindgames.
It is completely obvious (to me anyway) that "playing to learn" should be applied literally in the simplest sense in friendlies when you're talking about technique. In this case, you should play to learn even at the direct expense of playing to win. I'll give you an example:
Say I suck at chainthrowing with marth (which I do, incidentally). Say I miss my grab 85% of the time when my opponent < 35%. So in a tournament (playing to win), if I grab a space animal at 20something percent, I uthrow to utilt to (unsweetspotted) fsmash or something of the sort, because it's guarranteed damage -- better than I would've done with a throw or two before I drop the ball.
But if I'm playing a friendly match (less emphasis on winning), I will attempt the cg, even though I will likely mess it up. Why? Because I know that cging is a superior technique and when else am I gonna practice it? (the comp can't DI; even in training mode on "evade" it follows a predictable formula). This applies to most/all effective techniques. Fsmashing a noob is more effective than missing a short hop and flying over his head, but you need to learn how to shffl eventually and it's best to learn it against real opponents.
The other way to play to learn applies to mindgames/strategies. I think what's confusing here is that playing to learn the mental game and playing to win are usually very similar, and at the highest levels of play are pretty much the same thing. "Playing to learn" is a little less simple and obvious here than it was when you were learning techniques.
For a noob/mediocre smasher, developing the a mental game must involve a certain degree of experimentation. This is because there is much that you don't know about the game. Imagine two noobs, both playing sheik. Noob A always approaches with a dash attack. Noob B always couters by either sheild grabbing or rolling behind noob A. To these two, those are the best options they have (that they know of) and since they're the best known options, it's all they ever do. (This may seem like an extreme example or like one of them will eventually counter the other one and up their level of play. Sadly it probably isn't that unrealistic. We've all encountered people who have been playing this game since its release, think they're great, and do nothing but dash attack and rol and smash. FLT recently beat one such player.) Playing to win involves using the best known strategy at a given time. But playing to learn (mindgames) involves more experimentation. What would happen if, just for the hell of it, noob A decided to jump up into the air and throw a needle during his approach? If he stopped before attacking and tried to see if he could hit noob B after his roll had ended? If he freaking grabbed? Noob A at least would start to improve by experimenting instead of getting stuck in the same infinite loop forever.
Mookie's advice of playing to learn suggest that you don't go for chain grabs, because you know it works.
You can either follow Mookie and play to learn, or follow Azen and play to win.
Now we get to the good or even great smashers. They understand their options very well in most situations. I don't seen anywhere in mookie's advice that advocates avoiding tactics that work. The problem is that playing to learn at a very high level involves playing against high-level opponents and forcing them to react to you thus forcing you to react to them. Your experimentation will come in when presented with a unique or especially deadly strategy that can be countered in several ways, all of them with ups and downs. You will probably have to try a few different ways to play best against said strategy. To do that, you MUST use effective tactics and push your opponent to force you to play in a new or different way. Playing to learn and playing to win at this level look almost exactly the same, with only a small degree of experimentation separating them. If you're playing to learn, you try out all the counters to a new strategy that you think would work and see what works the best. If you're playing to win, you mostly stick with the first one you find that's effective (why risk giving him the upper hand even briefly?) until your opponent pushes you off of it.
Azen is a fine example of not always playing to win -- if he did always play to win, he would counterpick more and abandon many of the lesser characters.