This is my first ever "Maximum Score" TAS speedrun I finished in one day made with. I was having fun making this TAS, I think I'm getting better at making these compared to my most true first and disliked one here ( https://tasvideos.org/6841S ). Today, I attempted to make the TAS reach the highest score limit on Office Yeo In Cheon Ha score in 14:20.50.
Darkman425: Claiming for judging.
Darkman425: This submission was made with a prerelease versions of BizHawk 2.9, which isn't acceptable at this time. While that can simply be remedied by having this work on the proper release there's more notable problems with this submission.
As pointed out by Spikestuff, it's faster to end minigames with two players crushing the third player/CPU, rather than wait on one CPU to beat the other CPU. This is demonstrated by the Salary Man Champ publication on this site.
Even disregarding the time saves from using two players, I unfortunately still found suboptimal inputs that would apply to a single player run. The first major one is that the transitions between minigames can be made faster by using precise inputs at the right times rather than using rapid fire. This saves a small amount of frames per transition and adds up across the entire TAS. The second major one is that even in minigames the inputs can be done faster. For example, the face mask minigame can be made faster to remove the last mask earlier. This in turn makes the CPU rubber band and finish 39 frames faster than this submission. Thus even if the one player input was copied to another player it'd still have suboptimal inputs.
If you're willing to try this again when a proper BizHawk 2.9 release, I definitely would recommend using 2 players to not have to wait on CPUs finishing up. I suggest researching how other Bishi Bashi type game have been TASed before as well with regards to inputs. I would also recommend using TAStudio to the fullest to make sure every input is done as soon as possible rather than relying on rapid fire inputs alone. For now, this submission is mainly being rejected for optimization issues.