I watched this and the [movie 1590]current NES run[/movie] back-to-back.
Now, there may be subtle differences in how the game runs, but I find it telling that you're four in-game seconds slower at the end of the 1-1 compared to the NES run (ending time of 293 compared to your 289) despite the use of the glitch.
To continue the comparison:
Level | 1590M | You |
1-1 | 293 | | 289 |
1-2 | 294 | | 292 |
1-3 | 287 | 285 |
1-Fortress | 292 | 292 |
8-Tank1 | 167 | 162 |
8-Battleship | 200 | 197 |
8-Floating Ships | 239 | 235 |
8-1 | 292 | 289 |
8-2 | 293 | 288 |
8-Fortress | 380 | 376 |
8-Tank2 | 211 | 209 |
8-Castle | 375 | 373 |
Thus, I doubt the level of optimisation that's been put into this, particularly in comparison to the long history of refinements SMB3 has on this site. There are a few instances such as the large obvious slowdown to pick up the mushroom that point to this being the case.
I also would like to dispute the level of entertainment present in the autoscrollers of - the current SMB3 run, as with many of its predecessors, take great pains to chain together long jumps and acquire large numbers of lives, which, for them, has been a key selling point (so to speak).
I therefore posit that this run fails to live up to the technical and entertainment merits of the run it intends to obsolete, and must vote No.