Yo! How exactly are runs going to be timed on this game, given that there is now access to an auto splitter? It won't be hard to set up a load pause for the level transition, so that it splits as soon as you hit the finish, then pauses the timer until the next level starts. In my opinion this would be a better way to time the game, as it is more accurate for single-splits, as well as (possibly) more accurate for slow machines (I have no idea if loading ever lasts longer than the transition animation). The only drawback is that the current runs that were submitted would have to be re-timed, so that they match up properly with this proposed system.

I already have a script written up for load pausing, however I can not test this until I get home. I'd like to hear what your opinions are on this

I had been using IGT. Load times seem so negligible that it barely matters, but if someone was running on a toaster and/or the record became very optimized, I suppose it could make a difference.

I'm not sure if IGT already discounts load times -- the run you submitted recently was about 7 seconds faster on IGT than your splits (I had updated the time to reflect this). I would recommend testing this with the autosplitter pause to see if it lines up. If the autosplitter is indeed more accurate than IGT, then it would be worth considering, but you should take into account the possibility of someone submitting a run without using your autosplitter, and how difficult it would be to retime that manually.

The reason that my run was slightly slower on livesplit compared to IGT was because I had paused the game a couple of times, this is the only time that the IGT is paused. the main reason I'd like to use an autopause is so we have accurate IL times (for IL submissions). Also, as soon as the autosplitter gets added to the splitters.xml, it should prompt the use of an autosplitter when someone configures Grapple any% in livesplit, so I don't think it'll be much of an issue for new runners. in either case, the level transition should be roughly the same for everyone, it's only in cases where the person's machine is so slow that it actually affects loads, that you'd need to manually check the times. In most cases an average time could be used for IGT submissions, as there's always 89 transitions, that usually have a fixed time.

Edit: I'd like to add that, even though it might not make a difference on most - if not all - machines, using an autopause on transitions would at least be just as accurate as using IGT. so even if transitions were always the same, it'd just be pausing for the same amount of time for everyone.

It's understandable to not use the autopause, since it's most likely a very trivial change, but in any case I'll try to write it up so that it can be used for ILs, since the IGT is there regardless of livesplit.

Edit2: After seeing the new WR submission I'm starting to doubt the usefulness of Livesplit. There is no doubt that WR is way faster than the previous submission, but from what I can tell from the video, there seems to be a split second where the IGT pauses as soon as you finish a level. by the end of the run the livesplit timer was about 10 seconds behind the IGT.

This most likely means that IGT will be more accurate to time runs, as, to my knowledge, the only things that could affect this are issues tied to game lag. However, there is also a possibility that the IGT is tied to loads, which means that a slower machine could technically get a better time (should the IGT pause for a longer time, while the load animation stays the same).

tl;dr IGT is probably more accurate