Premade Files and the New Owl Save Exploit

Edit: As of 4/17/17, a community vote has been held and option 3 has been decided by a majority.  The common rationale cited among respondents was that all benefits of premade files should be treated equally, and that speedruns should start with no prior advantage when possible.  Runs completed prior to this date are considered grandfathered; their times on the leaderboard will remain valid and unmodified.

---

There is a recently discovered trick in Majora's Mask which abuses the owl saving system.  It turns out that, on the file screen, the 2nd file's owl save is (partially) loaded into memory.  When the first file is loaded as a non-owl save (this includes creating a new file), not all of the loaded addresses are overwritten.  This means that some data from the 2nd file's owl save can persist into a new game created in the 1st file slot.

Currently, the most relevant address that is not cleared is the rupee accumulator.  This means e.g. that if one withdraws a large number of rupees from the bank and then owl saves before their rupee count has fully increased, the difference will carry over to every non-owl save loaded from file 1 (see https://www.youtube.com/watch?v=bNADU5f9nHE for a demonstration).

To explain again and more plainly, if one sets up their 2nd file correctly, every time they load their 1st file (aside from owl saves), they will automatically receive a few hundred rupees (up to the maximum their wallet can carry).  This means starting a fresh file with a full 99 rupees, and then getting another free 99 upon beginning cycle 2.

This would obviously be beneficial to speedruns.  Also, once again, note that it is exclusive to versions with owl saves (mainly English VC for speedrun purposes).

The question at hand is whether the time investment required to set up such a glitch should be included in a speedrun timing.  The main complicating factor in making this ruling is the long-standing use of premade files in 100% speedruns.

In 100% categories, it is currently permitted to copy an unused file with known bomber code and Oceanside mask order.  This clearly has both similarities and differences with the new glitch and file manipulation techniques in general.  On one hand, every 100% run is done on a fresh file which has not been in any way manipulated.  On the other hand, there is clearly effort and time expended to obtain and/or set up a file with known codes, and that is not currently taken into account in the timing of a speedrun.

With the introduction of this new exploit of the owl save system, the community has been looking at the effects of premade files with a renewed scrutiny.  There are, as it stands, 3 main choices:

1. Do not require any sort of timing for anything set up before starting a run (of course with the caveat that you must still begin runs from the start of the game).

This would mean that the new rupee glitch, along with any subsequent uses of this exploit, would be fair game.  It would of course also mean that the current use of premade files for 100% would also remain in place.

2. Require anything that affects memory values in a new file to be part of speedrun timing, but not anything that grants the runner knowledge of memory addresses while not affecting them.

I worded that as best I could -- basically, this option maintains the status quo.  Premade files would still be allowed for 100%, but the new glitch would be subject to a significant time detriment (effectively invalidating it).

3. Any file other than the one created for the speedrun which affects the run should be included in the speedrun timing.

This option would require all future 100% runs not to begin from a file with known codes, and would also effectively invalidate the new glitch.