Many intriguing questions. Many good points. It seems in this discussion that whether any modified rom should be allowed is the major issue. To date it has not been allowed. Do we say no simply on principle that anything modified, whether cosmetic or not, is not "US Set 1" as it was originally released and therefore should not be used? There is a strong argument for this position. When I wrote "the ROMs must be US Set 1" it was written indicating the original code with no alterations in game play. That way the list was intended to only track a single code that is identical in all respects from the first line to the very last. That is its original intent. While we did not have such a rom with minor cosmetic changes such as this at that time means that it is not explicitly excluded, but yet its rejection is implied in the intent of the rule.
It was certainly intended to exclude US Set 2 which is different in the game code itself and hence its exclusion because the game play is altered. So far the annotations included on the HSL only denote hardware and source of video or inp files. The HSL has not included any altered rom files. While there is certainly a difference between US Set 1 and US Set 2, from US Set 1 to modified US Set 1, since one changes the game code and the other is a cosmetic difference, should we ever allow any modified rom at all? Obviously Pauline is yet another case but here the graphical interface between player and game is not absolutely the same as Jumpman and is part of the game play although the functionality of the game is unaltered.
I think that the scale can be tipped in favor of its acceptance with annotation on HSL that said modified rom was used if we could have absolute certainty that game play is not influenced at all. While the way the game normally functions may not be changed, does the modified rom influence RNG, the timing between boards, etc, so that the outcome is different than it otherwise would have been, although within the limits and terms of the processing code. Willms once noticed that if you finished the barrel board timed with the flashing 1Up that it would eliminate the delay in the 5th fireball. At least this is how I remembered it, and assuming that this idea is still valid, would this still work on the modified rom? Is the cutscreen still the same set of frames and no lag or extra stress placed upon the Z80 would have influence outcome in anyway?
There are very few people who understand Assembly and can verify with absolute certainty that there never will be found some anomalous, unintended result. No one knew that NES Jaws begins to function differently at 10 million points until one actually achieved that score. While I don't think there is anyone who questions the competency and integrity of the one who modified the rom, unless the community agrees very strongly (perhaps unanimously?) with one perspective or another it should not be allowed for the time being but the discussion should continue.
While the HSL has functioned with normative and auxiliary rules, it has become clear that in certain ways the sense of the normative rules, or at least some of them, has taken on a more absolute character in how they are generally applied such as the 3+1 setting. (Thanks for Jeremy for this insight) And perhaps the same can be said of a strict interpretation of "the ROMs must be US Set 1" to mean nothing short of a single code that is identical in all respects from the first line to the very last.