A sudden thought visited my head.
Some time ago I had an argument with adelikat and Zeupar about obsoletion obsession. I was told by those guys that no matter what, a faster movie MUST obsolete a slower one, ideal solution would be to keep just one speed-oriented branch for each game, but "the poo-shield isn't strong enough" to apply it right away.
This point is about setting a record, which is quite sane (at first glance). You use a trick you knew as the most optimal, kept a record (published run), then someone either improves this trick (making your old aplication of it suboptimal), or finds a way to skip the whole segment where this trick is needed. That means your run is overall suboptimal from this point, and once an improvement is submitted, it must obsolete your movie. This perfectly fits into the old politics (keep amount of branches as little as possible).
However, until now, no strict border was drawn to separate obsolete-worthy runs from legit goal choices. But this is quite important, because adding a new branch usually means the adjacent branches still use tricks that may be proven suboptimal by such new run. And this situation kinda contradics with the record-oriented politics explained above.
Example:
[1908] SNES Super Metroid "ingame time" by Saturn in 39:15.30
contains tricks that prove suboptimal the tricks used in
[1368] SNES Super Metroid by Taco & Kriole in 38:41.52
but none of them obsoletes each other. Why? Because the goal was considered
different enough. No one minded SM had 5 branches from that point (it was cool, and 2 more published branches to come).
But in
some cases, the branch wasn't considered different enough. And one movie obsoleted another one. Despite of having noteably different flow. Why? Some tricks used in one branch were proven subopptimal by a new branch. And it got obsoleted.
Examples:
[1564] GB Super Mario Land 2: 6 Golden Coins by andymac in 21:43.52
[1721] GB Super Mario Land 2: 6 Golden Coins "game end glitch" by andymac & MUGG in 02:28.08
[450] SNES Chrono Trigger by hero_of_the_day in 4:21:11.83
[1285] SNES Chrono Trigger "save glitch" by inichi in 21:23.98
[1920] NES Battletoads "warps, 2 players" by feos & MESHUGGAH in 11:04.72
[2100] NES Battletoads "game end glitch" by TheZlomuS & DyLaX in 01:02.68
I was saying adelikat that a fastest branch is a gameplay-based TAS, and a glitched is
different enough to become a side branch. He said that the trick used in any% branch was considered optimal when any% was published, but was proven suboptimal when glitched was submitted. He said the author might be trying to cut some more time with some trick in any% run, but couldn't, and the glitched run utilizes it better, thus must obsolete any%. He also suggested to obsolete all 3 Toads runs with one.
Why such double standard exists at all? There might be a sudden border between a trick (SM cases) and a glitch (SML cases) - but no one ever mentioned that. It might be a personal decision of adelikat, but what it was based on back then? To keep branch amount minimal? This restriction was terminated. To avoid suboptimality? More or less suboptimal movies of same games have to co-exist (still, co-exist based on what?). To always keep only the fastest branch? It was never done officially. To satisfy the viewers who found the obsoleted run too outdated? SML2 and Toads still don't fit. Why then? Tell me someone?
And now, when we no longer aim for branch minimum, we revive arbitrary goals as long as their submissions were not really sloppy. That system was successfully runned in for quite some time. But the obsoletion obsession still doesn't want to be revised. WHAT FOR we keep it?