To say that this site has helped so many people discover and appreciate this wonderful subset of gaming culture, myself included, is an understatement. Not only did TASVideos show me how beautiful the world of tool-assisted runs could be, but it also allowed me to look into more niche communities that are more disconnected from TASVideos (the Mario Kart Wii TAS community, for example). However, to say that the site's original vision and still-to-this-day strict rules and processes on which runs can even be featured on this site are more and more becoming a major problem is also an understatement. I'm not saying that someone's first TAS of Super Mario Bros that's multiple seconds slower than HappyLee's TAS should be accepted (I have actually seen this exact sort of submission pop up before). There obviously has to be some sort of quality control. But the emphasis on entertainment and full-game runs at the detriment of individual level runs and runs strictly focused on time is the exact reason why the Mario Kart Wii TAS community is disconnected from this forum. It's the same reason why the Goldeneye 007 and Perfect Dark speedrunning community, AKA The-Elite, was disconnected from Speed Demos Archive. In fact, TASVideos is basically the TAS equivalent of SDA in its current state. Something definitely has to change, and it looks like we might be heading towards most of those needed changes. Being more of an archival site, preserving the history of the artform, allowing IL runs, reducing the bottleneck of the curation process. These changes are absolutely needed. However, more than anything, the strict enforcement of certain rules regarding what can even be published on the site has caused problems for multiple games and their TAS potential. As a basic example: Super Mario 64 has five major RTA speedrunning categories: 120 Star (100%), 70 Star (Any% NMG), 16 Star, 1 Star, and 0 Star (True Any%). However, two of those categories do not have equivalent TAS runs currently published on this site due to how publication rules are enforced. Now, I think we can all still live peacefully if 1 Star never gets a published run due to how similar it is to 0 Star. However, the lack of a published 16 Star run with modern optimization is very upsetting, as it's one of the most famous speedrunning categories in all of gaming. That being said, at the very least, Super Mario 64's optimization has never been held back by the rules of the site or the limitations of emulators. Unfortunately, that can't be said for every game. While some issues come down to emulator limitations (cartridge tilting, for example), there is one uniquely frustrating case where the sites rules have straight up prevented a category from using the most viable strategy, despite said strategy having similarities to what is actually used in the current published run.
Super Mario World, one of the greatest video games of all-time and a legendary speed-game in its own right. Now, unlike with SM64, all three of its major categories are accounted for in some capacity. 96-Exit is featured in its full glory, and Credits Warp is represented by the arbitrary code execution category, which focuses more on entertainment these days but has been speed-focused at certain points in time. That leaves the famous 11-Exit category, an iconic speedrunning category with a rich history in both RTA and TAS. Except...the TAS currently uses a strategy that, while it used to be the best known route, has since been made outdated in RTA. In the TAS, a detour is taken to YI1, where you take Yoshi all the way to the end, use Fire Mario to kill a plant, have Yoshi tongue the coin you get from killing said plant, and then collect the coin right as you load the Chuck into memory to trigger the Chuck-Eat Glitch, giving you the goal orb that's normally only found at the end of the Sunken Ghost Ship. This is then used to clear Iggy's Castle, and is faster than just going through the autoscroller and the cutscene after beating Iggy. However, as I said, this is not the current strategy in RTA. They instead do the Chuck-Eat Glitch in YI2, getting Lakitu's Cloud, and using it to skip a significant amount of waiting in the Bowser fight. So, why the hell isn't the YI2 Chuck-Eat used instead of the YI1 Chuck-Eat? Because the YI2 Chuck-Eat can be used to execute arbitrary code, and it's used in Credits Warp. TASVideos site rules dictate that ACE is its own category separate from regular Any%, meaning that an 11-Exit run using the Cloud Bowser strategy would be in the same category as the Credits Warp and automatically rejected for not meeting optimization standards. And making it even stupider, the YI1 Chuck-Eat is STILL allowed in non-ACE categories DESPITE being very similar to the YI2 Chuck-Eat. Has anyone ever even really tested if the YI1 Chuck-Eat could be used to execute arbitrary code in any way, or is there something about that type of Chuck that makes it okay for some reason? I know there was a huge debate about this trick years ago, but how extensively was the orb Chuck-Eat tested? If it's somehow found out that the YI1 Chuck-Eat also has ACE applications, it might straight up kill the 11-Exit TAS. You'd have to retroactively reject any run that used to orb glitch and go back to strategies so primitive that they haven't been used at the top level in RTA in OVER A DECADE! It's genuinely one of the stupidest and most glaring cases of TASVideos red tape becoming a major detriment to a category.
I think I've gotten my point across. Certain site policies have harmed the optimization and entertainment of some categories while straight up forbidding others from even existing. For Super Mario 64, that problem does have a simple solution: add a site rule where any major RTA categories for a game is eligible so long as they are significantly distinct from other categories; 16 Star would be my conservative example of how this rule should be applied. As for Super Mario World 11-Exit, that is not an easy fix in any way. A new rule making an exception for applications of ACE tricks that are useful in non-game-ending glitch categories would have too much of a grey area, so I believe that the current ACE rules would need to be completely overhauled, which would cause a major restructuring of the submission and publication process and the categorization of runs due to the red tape surrounding it. Even if all of the other structural issues the site has currently are resolved, the overly strict rules on what can be published are still big enough of an issue to potentially sink the site anyways, so I think this should be of similar urgent priority to the other issues already discussed in the main post of this thread.