Thanks for your reply, p4wn3r.
1)
Sure, TASes like Brain Age are not particularly deep, because they basically repeat the same glitch over and over . But TASes that aren't deep but are flashy can be an onramp for TASing interest. Look at how popular the earlier TASBot blocks have been, even though they featured a lot of ACE/playaround TASes, where you do one glitch and then do whatever you want after that. Personally, the International Superstar Soccer Deluxe TAS, despite not solving some convoluted optimization problem related to score or speed, has been one that I've watched and laughed super hard at several times since I found it, have shown it to other people for similar reactions, and has an incredibly high view count on nicovideo (in the millions IIRC - on youtube it's not bad either, 232k). Once you're more familiar with TASing and your tastes have matured, you might not appreciate it as much as something finely optimized and incredibly detailed, like say Dawn of Sorrow "all souls" or YI 100% (the two highest rated TASes on the site ATM), but these TASes still serve an important and useful purpose, despite being 'non-competitive'. Not every TAS has to be all things to all people, and indeed it can't be.
2)
To have an unambiguous set of rules that is set in stone, you have to handle corner cases and borderline cases with an iron fist, and the rules would have to never ever change over time. I think this is simply a case of misplaced priorities - it is more useful to TASVideos overall to accept a TAS if it is incredibly good, but bends/breaks some of the rules. This shouldn't be seen as a sign of being weak - it is simply the compromises we have to do in an imperfect world, where things aren't always nice and neat. Plus, this kind of rule bending is permissive - rules are always bent in favour of publishing a TAS, not rejecting it. What does that leave to complain about, unfair treatment? Well, maybe if you TAS Ocarina of Time 100% and it takes 5 years to complete, you too can get dinner and back massage of your choice :)
Anyway, so you want TASVideos to reach out to the speedrunning community for each game (so I guess www.speedrun.com?) and ask them what the definition of each branch for that game should be? Which side is the pro-active one here (what event triggers the branch definition discussion between TASVideos staff and speedrunners)? RTA speedruns ban many things that TASes allow like L+D/U+R, do we now need to strike all TASes that use these from TASVideos? What about cases where the TAS and RTA rules for a category explicitly do not overlap, like SMW any% no ACE, which allows cloud in RTA but not in TAS (because TASers consider it to be as powerful as ACE)? Do we strike that TAS from TASVideos because it's not a SMW speedrunner approved category? I just want to get a better idea for how you would expect this to work and why.
(I personally think that the correct response to 'this TAS branch is slightly different from the RTA branch it's most closely related to, and I don't like that' is to create and submit the TAS that does the branch the way you want it to be and better optimized. Obsoletions of slightly different criteria can and do happen. It's not an un-competitive act, because all the similar aspects of gameplay can be compared to verify that the level of optimization and game knowledge is the same or better everywhere.)
3)
Just out of curiosity, I checked and there were 143 moons TASes published in 2016, vs 47 so far in 2018. We would expect about 42 so far if we were on 2016 pace, so we are a little ahead of 2016 pace. So if there is an impending 'TASpocalypse' where TASvideos slides into obscurity, the numbers do not bear out for it yet.
If you personally don't feel TASVideos is right for your TAS, you aren't obliged to submit it. (That's why you can't submit someone else's TAS on their behalf, for example). And in fact there are a ton of Japanese TASers who put stuff on nicovideo and never submit it to TASVideos, so TASVideos can't ever really be a collection of all TASes worth recording. But things seem to be going fine overall, and TASVideos' collection continues to grow year after year, with improvements and new games alike.
I'm not even sure if I have any point here, I might just be rambling now.
(4: I also agree that TASing is not necessarily a competitive endeavour. Of course speedrunning has many co-operative elements like glitchhunting and routing, but TASing even more so, as the input itself can be worked on by many people, rather than it being a contest of personal skill. We also give co-authorship when multiple people contributed to a TAS, which is great.)