I kinda think the depth comes from recognizing a missed l-cancel and punishing ur opponent for it, especially in a situation where ur shield is being pressured. Its always best to l-cancel, but having the awareness to punish in situations where a missed l-cancel is possible/likely (such as when ur opponent does an aerial on ur shield) adds some depth in my opinion.
Also, I-canceling is a skill in the way that it challenges a player's dexterity. I cant support ur argument that simply because the other player gets better reads he should be better. Its kinda like saying that a baseball player that can recognize a pitchers pitches better should be a better hitter. Or a hockey player that alway's knows where to position himself to should be good. Mental ability is part of all sports, and so is physical execution. Esports are sports because execution matters. Chess isn't a sport because its only a competition of strategy.
A baseball player who knows where the pitches are going will be a better player if they apply themselves to the game. That means learning to swing that bat and aim for the ball. That's why people learn to use attacks, space, recover, etc. But L-cancelling would be equivalent in your analogy to making a batter give a thumbs up after every pitch to reduce the time it takes the catcher to throw the ball back - it's totally unnecessary and could be avoided through better design [just have the man throw the ball back instead of lagging if you don't thumbs up in an acceptable way].
You can also "double tap" L-cancels to avoid missing L-cancels literally ever - Kadano made a video on it [I'm sure he disagrees with my position on being in versus removing itl, but he provides the frame data to support this and recommends use of this method]. Optimum L-cancelling involves a double buffer [tap L and then R lightly, or vice versa, in a certain window] and precludes missing any and all L-cancels (even Ganondorf's stomp on something vs on nothing). Thus at the top level, if we all had all day to practice, there would be no missed L-cancels ever. It's simply a barrier to separate those who have time to practice properly mashing shield triggers when they land vs those who don't, which is dumb, when so many people are just so bad at the mental aspects of the game or certain MUs, which definitely need practicing, and such practice is going to lead to more interesting and better quality matches much faster than miring someone down with trying to hit triggers [you can go to a tournament and look at skill gaps - PPMD destroyed M2K at Apex 2014, who if I remember correctly bodied Leffen, who got beaten by Mango [who lost to M2K], and all of those people could destroy numerous people below them, who could in turn destroy people below them, who could destroy countless other players out there, and the chain extends well beyond that]. Separation of skill level by ability to double buffer or correctly time an input that could be completely obviated is ridiculous.
And yet another instance of someone criticizing L-cancelling for lacking qualities it never meant to fulfill. I suppose we should criticize the fact you have to correctly press A and a direction on the control stick to perform an attack, or that we have to let go of the jump button within x amount of frames to short hop.
Honestly, the failure of your argument lies within two regards:
1. Depth is not limited to strategy and decision making. Technical depth has merits and contributes to overall depth. The technicality of an instrument such as a guitar or piano can enhance the meaning/creativity of music.
2. The technical depth L-Cancelling adds does not overpower mental strategy. While it seems arbitrary, its binary nature is rather complimentary in a game full of abstract concepts and exceptions. It is actually a great introductory technique since it gives newcomers immediate feedback with their progress.
Technical depth only has merit in as much as it is a requisite to a different action. Angling attacks adds depth. Smashes vs tilts vs neutral moves adds depths as well because it allows for more inputs on a limited set of commands [and it's integrated efficiently in my opinion, and I'd find aerial smash attacks an interesting but welcome edition to any potential Smash 5 (since Smash 4 has already come out)]. DACUS adds depth because it allows characters new options [Ganondorf's DACUS in PM is terrifying, and yes I'm aware we're in a Melee forum but the argument holds]. Wavedashing adds depth because it allows new options as well, out of dash, out of shield, etc. So does b-reversing, shielddropping and spotdodging, jump-cancelling [it matters because not all JC grabs are better, and a slow JC grab is always slower on startup anyway, therefore it is a decision], and other actions. L-cancelling is not a prerequisite to a different action, it's an arbitrary input, and therefore could be obviated easily without loss of the impressive skill on display in a wide variety of other forms. Technicality in music can also sound terrible, but people in music don't obsess over technicality, they are consumed with the desire tor beauty. I'd much rather see high-level players make masterful reads, complimented with good technical skill in the areas that open up options [wavedashing, jump-cancelling, etc.] than see players who think they're good because they can multshine but walk into the same stupid things the other does over and over again, but that's what happens at low levels to players who focus on technical skill - it's a diversion and distraction from what's ultimately going to affect how far they can progress as a player [and if you need an example of this, look to good ol' Borp - very little technical prowess, but he can keep reasonably close to Toph and beat various other players who can do flashy things because he chose to be good at playing Smash instead of choosing to be good at pushing buttons fast with various usually strict timings].
Funny enough, I tend to have L-cancel rates no higher than the mid-80s in tournaments [often the 70s], but I beat people who have L-cancel rates at 100% and punish my whiffed L-cancels. At best, L-cancelling leveled the playing field a bit, but it's more likely that they've spent significantly more time getting good at practicing hitting the L-cancels while I've spent time honing my style, working on reads, and making sure I hit at least a few L-cancels on important aerials [as Link in PM, mainly dair], to the point where I can outplay them at Smash, even if they outplay me at Super press buttons bros. Also, the human reaction time on average is about 16 frames or so [I think it's like .28 seconds or something], so unless you're expecting a whiff [and against a good opponent, you shouldn't], punishing a whiffed L-cancel on reaction is basically impossible anyway.
Given that bit about human reaction time, at the top level, players pay enough respect to each other that whiffed L-cancels are harder to punish simply because you expect the opponent to hit their L-cancels and punish around it - that, combined with the L-cancel rates that are probably at, or very near, 100% [at least in a game by game basis, obviously not lifetime], means that L-cancelling doesn't add depth to the game, since people are going through an input that could be obviated without changing the outcome on screen and won't be looking for whiffed L-cancels, but rather for openings in technically proper [but not necessarily smart] pressure/zoning/whatever.
I’m also sure that people will still take issue with this, either from an “agree to disagree” standpoint (as in, "I don't want to argue and I'm not changing my mind") or [more likely] a “this is wrong” standpoint, but all these issues seem to start with people over-simplifying what I say or attempting to compare it to something with a flawed example. So I’m sure this won’t be the last of it [though we really shouldn’t turn another thread into one of these discussions] but I’m also fully prepared to continue to defend this position.