I have ran NES Paperboy on a front loader NES with an everdrive, and I also TASed it using Bizhawk, this would be a new TAS to surpass the old one. However, despite the fact that the tas editor allows you to modify frame by frame my run of Monday was 4272 frames, but TAS required 4285 frames. There is no lag frames on this day and it is the fastest the day can be accomplished. After TASing it I compared the frame count to the frame count of my recent world record run, and the amount of frames it took me to complete each day ran almost exactly 99.7% of what was required in Bizhawk, using NesHawk. This is very strange. I first thought that maybe my NES was consistently running faster but again I have compared my runs to other player runs and they are comparable. I used three forms of verification to get the frame count of my run on original hardware: 1) Viewing the timer value in LiveSplit on first and last frame, 2) frame analysis of OBS capture of the game, and 3) frame analysis in Yua after deinterlacing the direct feed recorded by Amarec.
Here is the BK2 file for the first version of my TAS since my next version is not completed yet. You will also find in here the direct feed of my world record run and a video of the BK2 file. Placing these side by side you can see what I am talking about:
https://drive.google.com/open?id=1k6oFIsPaWaBeFjnNrkib-_yhxXwFhRU3
Also, here is the run the TAS is based on:
https://www.youtube.com/watch?v=4sESTUcQM10
Any thoughts?