This is an interesting article, and I agree that balance isn't always as important as people think, but I don't think is well thought through in terms of game design.
The shuffling indicates that new tactics/play styles are found and widely adopted (as well as preferences are shifting) over time. With rebalancing, the shuffling should be more or less random (or at least you would see a higher "volatility"). Without rebalancing most characters which were judged as strong stay strong but players perception for some characters changes with tactics etc.. It would be interesting to not just see the overall standings but some kind of "consensus" rating (which characters are rated similar and for which characters ratings are mixed).
By the way, your point also shows that SSBM has an extensive gameplay and is (if intended or not) well designed for competitive play - otherwise ratings should be more stable.
Balance doesn't have to do with changes in perceptions of character strength. A game is usually said to be balanced if there is a large selection of characters that are considered viable for tournament play.
In nature, there are balancing mechanisms, but there is no balance. If one organism learns how to overwhelm another, it often spells the extinction of the weaker organism.
In games, however, if one character is unviable, then the game is said to be unbalanced, which naturally leads to discontent with the portion of the player base that likes that character. Balance is a battle against the natural entropy that generates tier lists or food chains, because it is a human preference, not a trend or a law of nature.
Smash is balanced in that win rates are similar enough to allow some mobility but not enough that the developers eliminated many of the later-discovered capabilities.