I think you might be conflating a number of different things. Particularly with what was published and not. In terms of wild variety in playstyle, you are definitely correct.
First of all, Thieves had percentile success since they were first printed in oD&D Supplement I: Greyhawk (see p. 11). The
proto-thief playtested at GenCon VII also used percentile success. The Gary Switzer version of the thief, from which the D&D version was cribbed, may or may not have used percentile success, but this isn't particularly germane since exactly one playgroup ever used it.
Secondly, you are correct that the thief is not an original class (coming out in Supplement I) -- and by the time it came out, people had already determined for themselves/their own gaming group which characters could do what thief-like actions (and how those were adjudicated). What happened when the thief was introduced occurred at a case-by-case level, especially for groups that had already determined that some of the things a thief could do could already be done by anyone (perhaps with a higher chance of success than was listed for a thief). Some would say anyone could do it, but only a thief could do it when challenged (or up against magic, or as you say move quietly vs. move silently. Others simply said that they already let fighters do what a thief could do so why would anyone play a thief and moved on with a thief-free game. Others went on with whatever rules they had made, but if your thief character failed using those rules, could try again using the thief rules. As with everything oD&D, it really was a free-for-all. And the rules were truly vague ("Basic abilities are:" and percentile chances, with the only clarifications being for when a re-try was possible, and chance of pickpocket attempt being noticed).
AD&D certainly did change this -- giving all sorts of rules and parameters and giving all sorts of situations where only a thief could attempt something (and many where even a thief couldn't -- or should be punished for trying).
