Quote Originally Posted by snowblindfrog
    It's very possible that Chicobo used Swiss for Week 2 in S20 Gold, it certainly looks that way except for Nine Angry Men getting a Bye in Week 2 after getting a free forfeit win in Week 1. That part has me confused. As I eluded to in the other thread we may not always have been super consistent on the Week 2 approach. I can recall a few instances where an admin had forgotten about using TER in Week 2 and just used Swiss (since the Swiss ranking is what we show on the admin scheduling page). It's hard to say for certain at this point. Silver definitely did not use Swiss for Week 2 in S20.

I'm 95% certain that for every season I've mained in after s11 (where I wasn't familiar with the website/scheduling system), week 2 was done by Swiss, at the very least in the division I was in. Maybe I'm misremembering, but as far as I know every week 2 I've played has been done by Swiss, which is why I was completely blindsided by the schedule this season and even more so at the fact that it's supposedly the default for scheduling. I'm going to go check week 2 schedules for s12/13 silver, s14-16 gold, s17 plat now as well. If you ask most players that have been around since then, I'm fairly certain they'll say the same.

s12/s13 silver

i had something typed up for this and accidentally closed it but every team within the first 10 I checked all had week 2 matchups against teams that had the exact same result as them in the week before, be it 2-0, 4-3 or forfeits

s14 gold
92 (1-0, 4-2 w1) vs LDS (1-0, 4-2)
muma (1-0, 4-1) vs JEW (1-0, 4-0)
IFA (1-0, 4-2) vs lambo (1-0, 4-1)
smash (4-0) vs PvP (4-0)
mtn (4-2) vs spacebears (4-2)
ski (1-4) vs more (1-4)
the (2-4) vs swak (2-4)

again, definitely swiss

s15 gold
ss (0-2) vs Tv (0-2)
aeiou (2-1) vs TW (2-0)
dk (2-0) vs GR (2-0)
bp (0-2) vs 2003 (1-2)
infms (2-1) vs bonus (1-2)
wcw (missed week 1) vs tlb (0-2)
GN (2-0) vs DOOR (2-0)

not as perfect but still definitely swiss, lots of 2-1 games means there's going to be some record variance

s16 gold
bonus (0-2) vs wykd (2-0)*
pepe (2-0) vs pingpong (2-0)
mtn (0-2) vs spacebears (0-2)
qaK (2-0) vs kanye (2-0)
ifa (2-0) vs wlly (0-2)*
gR (0-2) vs XD (0-2)
qt (0-2) vs wu^ (0-2)
glory (2-0) vs does it (missed w1)
WWE (2-0) vs catbean (2-0)
Hnggg (0-2) vs Shh! (0-2)

Two matchups here are opposites, so if there was a season that used TER, this was it. Regardless, there's still a large majority of matches that are between even-record teams. Not sure what the deal with those two MUs was, I want to say there was a reason but I can't remember.

s17 plat

alec (2-1) vs DOOR (2-0)
sheen (1-2) vs pingpong (1-2)
cK (1-2) vs bp (1-2)
qaK (2-1) vs BR (2-1)
mM (2-1) vs ww (missed w1)
92 (0-2) vs TBA (FFL)
Bv (BYE) vs aMw (FFW)

easily swiss

s18 gold
appel (1-2) vs meme (1-2)
quail (0-2) vs ifa (0-2)
sjw (2-1) vs aW^ (2-1)
ts (2-0) vs wu^ (2-0)
.cfg (2-0) vs No.2 (2-0)
pepe (2-1) vs does it (2-1)
shadowball (1-2) vs u$a (1-2)

swiss

I've already covered s19 plat and s20 gold.

If you'd like I can go and check other divisions in other seasons not listed here, but I think it's pretty safe to say that even if it's been intended for week 2 to go by TER/ER scheduling, it's almost never (if at all) been put into practice before s20 silver (maybe s16 gold?? maybe).

Coincidentally, with this being the 2nd time clear, widespread implementation of this system, we've had scheduling issues that I don't think have ever been seen before (the issue with s15 plat was due to a near-full season of bye weeks in a <16 team plat, meaning that top half teams received a bye that unfairly implicated playoff seeding). I think a lot of the issues were due to the extreme circumstances of the skill divide in the division this season, and sheer bad luck beyond that, but door/bv/mcm all starting out 0-2 and OG/swift/chugger starting 3-1 and ayylmao 4-0 reveals a pretty serious flaw in week 2 TER scheduling that definitely has the capability to be repeated.

Straight from the wiki,
The first round is either drawn at random or seeded according to some prior order, such as rating (in chess) or last year's performance. All participants then proceed to the next round in which winners are pitted against winners, losers are pitted against losers, and so on.
Breaking down Swiss a little bit, it's generally so accurate due to the fact that each week adds another skill subset and adds another chance for a team to move subsets. After week 1 (assuming all superior team will always beat an inferior team), assuming matchups done at random, the swiss system assumes that all winners inherently belong in the top 50% of the competition. After week 2, Swiss presents a 25/50/25% split. Week 3, 25/25/25/25, week 4 20/20/20/20/20, etc. The purpose of this is to naturally let teams rank themselves through the inevitable increase in accuracy of a (well-functioning) system as a sample size increases. Intentionally messing with the system by allowing top half teams to be paired with bottom half teams not only delays the proper settling of teams but expressly increases the time it takes for it to happen. For what it's worth, most swiss systems that don't RNG their first matches use a bracket style seeding: 1v16, 2v15 etc to ensure that top half teams have an early advantage in being in the top half and to prevent free wins from going to bottom half teams. Personally, I believe RNG is the best way to play week 1 (with some possible admin intervention in the case that the two perceived best teams end up playing each other week 1), but using TER/ER for it is fine. Beyond that, letting the system take it's course is 100% the best way to ensure accuracy.

tl;dr the system that was supposed to be in place wasn't actually in place, and putting it into place completely borked it and we should go back to the old way.