Would this REALLY do anything about combos at earlier percents?
I mean, he said it himself, the vectoring is more significant at higher than lower percents. Let's say for the sake of simplicity that vectoring is only going to have a value equal to 1/20th of your current knockback value.
A good chunk of combos we've seen so far, (tilts guys, tilts!), have very very little knockback. Were they to attempt vectoring breaking out of an early combo would still be extremely difficult since the knockback would be nearly negligible. Assuming that knockback is a discrete, noncontinuous variable then it may be the case that at low percents it's completely nonexistent. (For example, 1/20th of 5 is .25, which would simply be rounded to 0. It could also be the case that the change in knockback of .25 isn't enough to break out of said combo.)
A second thing, probably most important, is that this does not reduce hitstun. So if at higher percents someone vectors in order to prevent themselves from dying, they may potentially set themselves up for another hit.
It's definitely a gamechanger, but I don't quite think we should start rioting just quite yet.