Good catch on level 13. ("How did I miss that?!")
@Pepsi added it on a whim, without interacting with me, shortly after I added him as moderator. Removing it is certainly an option.
Would have been nice if you waited for a signing off on glitchless% or included video footage for your debut submission (!!), but I suppose you'll be vindicated if we get some interesting or competitive runs out of it.
Oops, I had looked at the guy's other run and had gotten mixed up. Yeah, you can certainly use a tab-switch to clip into the wall in level 2 and end up on the ceiling. I remember deciding against that path when planning my run.
If we can find differences in the game as it appears on different sites I certainly agree that we would need to decide on a standard, but no differences have come to my attention so far; I suppose that only one version of the game has been released. If a "definitive" link is necessary, I'd suggest https://www.newgrounds.com/portal/view/471144
On the subject of differences in the game, it is worth keeping in mind that different browsers and flash plugins may have different behaviors---especially concerning the tab-switching gimmick. For example, I haven't been able to use tab-switching tech on Firefox 16.0.1. It has also been suggested on this forum that clips are sometimes easier to make when the system is lagging. I think that sounds very plausible.
When I did my run way back when, I downloaded the flash object and embedded it on an empty page to avoid distractions and speed up resetting. As an entertaining aside, I discovered that the game will happily draw outside of its usual container range. Ever wonder what's going on in level 4? 
Presumably, "elevator skip" is due to the koopas clipping through their boxes and falling sufficiently far down to trigger the level end.
.....
The run is spliced. I'm extremely skeptical that the "skip" is possible---the run proves exactly diddly squat to me---but that's beside the point.
But hey, I'm a guy to compromise. If @TigeroBill can splice a run that doesn't look like something a second grader would make, I'll give him the record.
Edit: Thank you @Rallis for transferring super mod privileges. I look forward to verifying a run that legitimately obsoletes my own!
I created a simple TAS engine for eddynardo's current series of games, initially aiming to TAS "Sticky Sorcerer" (see: https://www.speedrun.com/Sticky_Sorcerer/thread/hwkw6 ). Today, I finished a TAS for Choppy Orc that improves @Neinet 's excellent RTA record by just under 10 seconds, reading 114.752 on the end screen. I worked carefully to plan and optimize each level and would be very surprised if so much as another half-second could be taken off the run.
I hope you enjoy: https://vimeo.com/310082708 . Gametime passes very slightly slower than normal in my replay but the in-game timer can nevertheless be trusted.
My most significant save is on level 14, where the TAS executes 4 frame-perfect jumps off the edges of deadly surfaces to save about 5 seconds over the strategy used on the leaderboards, but the TAS's optimized movement and "double-jump" tricks over the preceding 13 levels would still be a bit more than enough to push @Neinet or @LazyTitan below the 2 minute mark. I have no idea whether this goal is within human ability.
I'd be happy to answer any questions or send my tool, in the form of a patch to the game's JavaScript source, to any interested parties.
Thanks! I'm thinking of doing Choppy Orc next. The runners on that game have already done some clever innovation, so I'm excited to see how well I can improve the run.
lol
It was disappointing to be notified about a new WR in this game and then see some ineptly doctored run but hey, no rush. I do ask that @Rallis take a look at this when they get the time.
There are a few pretty obvious discrepancies in the recording. I'm actually surprised @TigeroBill managed to fit so many mistakes in; it would be hard to do better if your design were to be caught.
After ironing out an ad-hoc TAS engine for the game, I completed a respectable TAS that improves over our RTA record by 25 seconds IGT. I hope you enjoy: https://vimeo.com/309156733
I'd be happy to answer any questions, and if anybody is interested in taking their own shot at TASing I'd readily share my tools. A basic familiarity with JavaScript would come in handy.
@SiriusAshling: I used the strategy you found to charge an extra dash after slamming into a wall in a few places. You suggested in your videos that is was necessary to change directions after slamming into the wall before you jump, but this seems to not be the case; your horizontal directional input seems to not affect your window.
In general, the game is very casual about letting you jump after you leave the ground in any way other than jumping. For example, on level 1 I charge a dash by jumping shortly after I leave the center platform. Just pressing x after my dash animation dies above the chasm wouldn't have worked, but pressing up and x together on the same frame does. It seems like you always have a ~ 100 ms window to jump (5 frames for me) after leaving a ground surface.
Obviously, it is necessary to reset your progress when you are starting a new 100% attempt. Less obviously, level completion progress may confer small advantages in Any%. Therefore, Any% runs that do not reset their progress can expect to be rejected, as per the rules. (Non-competitive submissions---like the 10:50 on the Any% board that sees no dialogue popup on level 4---may be treated with a certain lenience in practice.)
I'm aware that standing below the low-hanging vertical wall and jumping up while defocusing will, in particular conditions, clip you in and eject you out with enough force to fly up to the exit. In the most favorable scenario this will save 2 seconds over my run though; did you find something else?
I'm not aware of any fool-proof ways to do this. However, although you can see in my 11:44 run that I am still struggling, I am improving with practice. This is something that must ultimately be learned on your particular controller -- for me, it was useful to lift my finger from the key and pretend that I am playing in staccato.
It is worth mentioning that failing this trick the first attempt or two may still yield a net win on many levels.
Heads-up that I changed "Hat wizard" in the name and URL of this game to "Hat Wizard". I didn't do this a long time ago because I wasn't finding a way to change the name without changing the URL -- some sort of technical issue with the site apparently -- and I didn't want to change the URL because it would break links used in certain then-recent Twitter posts.
I'm changing both now because I don't think there are important standing links to the game. If there are any complaints posted here, I'll change it back.
When Bowser falls down to the screen border during the fight, he simply respawns.
I'd assume that some people have already gotten the best times possible for a level, [...]
What an odd assumption.
My own run has the fastest segments up to the boss fight. The fastest boss fight stands in one of the other sub-1:00 submissions.
The tab-switching tech works because some browsers -- in particular recent versions of Chrome -- will heavily throttle Flash apps running in inactive tabs. (See: https://saumyaray.wordpress.com/2011/07/14/flashplayer-throttle-solving-a-problem-or-creating-one/ .) Because operating systems do not usually throttle unfocused applications, I do not expect that this behavior can be emulated with a standalone Flash player.
If you grab the flagpole with Tetris mode active you will be able to place tetrominoes during the cutscene. Since movement inputs are killed when the mode switches, it's most reliable to do this by screen-scrolling yourself into the flag. Using the screen scroll to clip big Mario into the block at the bottom of the flag will work.
This glitch alone will let you get a few blocks into the castle before you're supposed to. However, damaging Mario with tetronimoes during the cutscene also has interesting effects. When big Mario takes damage, he becomes invulnerable during the castle construction. When little Mario takes damage he disappears from the screen entirely as if it were a game-over, but the game-over trigger is delayed for some reason.
