Well to start with, talking to Chi, Kevin, and Charles is pretty much getting anything I'd tell you second-hand. I frequently share ideas with all of them usually by text, and you're likely to encounter a good portion of rephrased redundancy.
I do not believe that Marth has a weakness. A lot of modern gaming revolves around the idea that <entity> is good against X and bad against Y. This is simply not the case for many older designs of interaction in gaming. In a lot of these examples, the <entity> loses to more subtle types of exchanges that are based around experience, finesse, and working around a margin. For an easy example, the older Pokemon games are this way; many Pokemon do not "lose" to something, but those Pokemon can still be defeated for many other reasons. As you've probably guessed, Marth is also this type of character.
Marth in Melee is a character that is based strongly on his strengths. In much the same ways as other more obvious characters, Marth can be played such that he makes the match totally degenerate and unplayable for the opposing character. This is done with those ideas of experience and finesse rather than through an overt tactic, but the strategy is degenerate regardless of its execution. Properly played Marth is very aggressive, non-committal, and removes options from his opponent amazingly fast.
In Project M, this is simply not the case. Marth can still do those things, but they are not the certain death that they were in Melee- and that death was very certain. Rather, his lack of strengths means that he is no longer able to leverage the same critical advantages that he had in Melee. To be clear, Marth in Melee is a fair character, except that he has a broken dashdance, that leads into a move that creates instant positional advantage in his upthrow, or down/forward throw in more pocket cases. From the point of the conversion at the upthrow, Marth can choose to engage the aerial opponent and win 100% of the time. This ability to use a broken tactic, and then convert into a flawless air-to-air interaction indefinitely until KO is fundamentally degenerate. But for Project M, dashdancing is not as good of an option because the movement speed of the cast is normalized to be much faster. The conversion with upthrow is still solid but Marth no longer wins air-to-air interaction 100% of the time even when it is done correctly. Finally, characters live longer due to bigger stages and better recoveries, so there's lots of margin to lose out on. The last part is probably the most important and that's why I covered it in that thread, but all of it is applicable.
I only feel that Marth is mediocre because I'm under the assumption that players will eventually learn that they do not have to respect Marth from poor positions to nearly the same extent that they did in Melee. Simple shield camping bad Marth players and waiting for a slow dedicated swing is actually acceptable method to beat him when you shouldn't be able to. More directly, I'm telling you that it's usually okay to expect Marth players to be bad in Melee. In Project M, the same still holds true, except Marth can't dashdance to the same extent, and the nature of the game actually encourages the Marth player to blind swing more rather than less, which we've already established to be "bad" overall (you can still lose to a bad strategy, but that doesn't make it less bad). As players learn to finesse through Marth's no-longer-perfect juggles, as they learn that he's easy to defend against, and as they learn that his swings in neutral are hardly threatening, Marth's relative goodness as a character can only go down. I'm simply a master of the character from 10 years of experience so I don't need to learn it from scratch like most PM players do.
I'm sure all of these ideas create more questions than they answer. Ask anything you want and I'll do my best to answer it. When you are satisfied, I would like one of us to post this exchange somewhere to the public as a reference.