np: USUM UU Stage 6.1 - Échame La Culpa (Breloom unbanned, Buzzwole remains BL)

Status
Not open for further replies.
[12:33:24] pokemonisfun: only people who can go 50-0 have relevant opinions on buzzwole

This sums up the suspect experience pretty well, in my opinion.

I'm pretty sure everybody agrees that there is pretty much zero incentive to ladder now - whats the point of wasting time if you lose one match by hax out of 20 battles and realistically can't lose another to get reqs? On paper, making a higher GXE should mean that only better people will get reqs and ensuring a fair and more "accurate" suspect result. However, it doesn't take into account how easily it is to lose a game based on matchup or even bad luck in RNG alone. I understand the mentality of "raising the bar" to ensure higher quality suspects but in reality this is doing the opposite. I'll try to explain why:

GXE is inflated by more people winning, and obviously other people simultaneously losing battles. The more battles that happen on ladder, the more GXE potential will be injected into the ladder making reqs obviously easier than say, only a bunch of low ladder people who will never get reqs just continually playing for fun or just out of sheer ignorance of how the suspect system really works.

People make new alts for suspects. They make a few, to a lot of alts, depending on the difficultly of the suspect test at hand. All these alts start at 0 COIL / an undetermined GXE obviously, so when you have these low players naturally coming in, when they win they actually just take GXE out of the system by killing mid-rage people's GXE (because when you lose to somebody lower than you it hits you harder than say, if you lost to someone way higher than you). This kills suspect runs, and then those people make new alts, and the cycle continues. All the COIL "taken" from other opponents on the ladder can be gone after a loss or two when the person restarts reqs on a different alt, the old one never to be touched again.

The suspect system is flawed in that way - on a suspect ladder with GXE based on relativity from one opponent to another doesn't take into account that what the ladder reads as a "low ladder player" and re-calculates GXE in that way after the battle could very well be an excellent player who is draining GXE/COIL from the system that might just be dropped when they lose a single game. On a suspect ladder where everybody starts from zero, the "low ladder" is not the low ladder on normal ladders - they're just as likely to be a random as much as it is as likely for a given opponent to be an accomplished and competent UU player.

Now, another factor thats coming into play is the motivation factor. Many people in UU main chat (recently including me) have been expressing sentiment about there being no incentive to ladder the suspect test, on the grounds that its impossible to get reqs with a bar this high. This, along with the previously stated point about the suspect ladder system not accounting for "hax" and other subtleties that can determine the outcome of a battle, means that less people (particularly the decent players to extremely competent players who might not want to risk wasting there time on something that has such a high chance at zero return just based off a single loss) will be even on the ladder, meaning less inflation of COIL and you're just stuck with the same two types of people:

1) People who don't understand how the suspect tests work and that its impossible for them to get reqs / people just laddering who dont know what suspect tests are at all.

2) People who are desperately trying to get reqs and are constantly taking rating out of the system by continually making alts until they get to the point where they give up, or eventually make reqs.

The suspect system before this meant that generally people were of a generally lower quality than the UU council was looking for - and I agree, previous suspects had some low quality posts on both sides of the issue being suspected, but at least there were enough people to make a consensus among the at the least semi-serious members of the community and even have a reasonable chance at getting reqs, and then discussion optimally happening before the vote (which again, is just the ideal situation and I know that some people only get reqs for that TC badge, but thats not the primary issue).

I strongly disagree with what is happening right now with the decisions made in the tier - I understand they are made in the best interests of the tier (if some of you remember I was an avid supporter of the council vote on Serperior and Azumarill because I understood they were in the best interests of the tier) but at this point its a fake suspect and now really only really lucky people (I get it, you'll need skill too, but you first off need the luck to not get hax'd out of a game or two) or people who are eligible for reduced reqs are going to be able to vote in the suspect test.

This all seems pretty ironic to me since now to avoid poor suspect test results and votes they're making it so things are a high part determined by luck and rewarding the grind instead of just somewhat based off skill, and while there are going to be exceptions to this I doubt there will be very many.

TLDR: This suspect test isn't working.


edit: after discussion I have realized that this post exaggerates the amount of losses it takes to kill reqs - this was unintentional and was more focused on early-run losses than late game assuming you would potentially lose more later.
 
Last edited:
[12:33:24] pokemonisfun: only people who can go 50-0 have relevant opinions on buzzwole

This sums up the suspect experience pretty well, in my opinion.

I'm pretty sure everybody agrees that there is pretty much zero incentive to ladder now - whats the point of wasting time if you lose one match by hax out of 20 battles and realistically can't lose another to get reqs? On paper, making a higher GXE should mean that only better people will get reqs and ensuring a fair and more "accurate" suspect result. However, it doesn't take into account how easily it is to lose a game based on matchup or even bad luck in RNG alone. I understand the mentality of "raising the bar" to ensure higher quality suspects but in reality this is doing the opposite. I'll try to explain why:

GXE is inflated by more people winning, and obviously other people simultaneously losing battles. The more battles that happen on ladder, the more GXE potential will be injected into the ladder making reqs obviously easier than say, only a bunch of low ladder people who will never get reqs just continually playing for fun or just out of sheer ignorance of how the suspect system really works.

People make new alts for suspects. They make a few, to a lot of alts, depending on the difficultly of the suspect test at hand. All these alts start at 0 COIL / an undetermined GXE obviously, so when you have these low players naturally coming in, when they win they actually just take GXE out of the system by killing mid-rage people's GXE (because when you lose to somebody lower than you it hits you harder than say, if you lost to someone way higher than you). This kills suspect runs, and then those people make new alts, and the cycle continues. All the COIL "taken" from other opponents on the ladder can be gone after a loss or two when the person restarts reqs on a different alt, the old one never to be touched again.

The suspect system is flawed in that way - on a suspect ladder with GXE based on relativity from one opponent to another doesn't take into account that what the ladder reads as a "low ladder player" and re-calculates GXE in that way after the battle could very well be an excellent player who is draining GXE/COIL from the system that might just be dropped when they lose a single game. On a suspect ladder where everybody starts from zero, the "low ladder" is not the low ladder on normal ladders - they're just as likely to be a random as much as it is as likely for a given opponent to be an accomplished and competent UU player.

Now, another factor thats coming into play is the motivation factor. Many people in UU main chat (recently including me) have been expressing sentiment about there being no incentive to ladder the suspect test, on the grounds that its impossible to get reqs with a bar this high. This, along with the previously stated point about the suspect ladder system not accounting for "hax" and other subtleties that can determine the outcome of a battle, means that less people (particularly the decent players to extremely competent players who might not want to risk wasting there time on something that has such a high chance at zero return just based off a single loss) will be even on the ladder, meaning less inflation of COIL and you're just stuck with the same two types of people:

1) People who don't understand how the suspect tests work and that its impossible for them to get reqs / people just laddering who dont know what suspect tests are at all.

2) People who are desperately trying to get reqs and are constantly taking rating out of the system by continually making alts until they get to the point where they give up, or eventually make reqs.

The suspect system before this meant that generally people were of a generally lower quality than the UU council was looking for - and I agree, previous suspects had some low quality posts on both sides of the issue being suspected, but at least there were enough people to make a consensus among the at the least semi-serious members of the community and even have a reasonable chance at getting reqs, and then discussion optimally happening before the vote (which again, is just the ideal situation and I know that some people only get reqs for that TC badge, but thats not the primary issue).

I strongly disagree with what is happening right now with the decisions made in the tier - I understand they are made in the best interests of the tier (if some of you remember I was an avid supporter of the council vote on Serperior and Azumarill because I understood they were in the best interests of the tier) but at this point its a fake suspect and now really only really lucky people (I get it, you'll need skill too, but you first off need the luck to not get hax'd out of a game or two) or people who are eligible for reduced reqs are going to be able to vote in the suspect test.

This all seems pretty ironic to me since now to avoid poor suspect test results and votes they're making it so things are a high part determined by luck and rewarding the grind instead of just somewhat based off skill, and while there are going to be exceptions to this I doubt there will be very many.

TLDR: This suspect test isn't working.
Agreed, I don't want to lose all chances for getting reqs just because I skillfully work around a team I have a bad matchup against, only to lose because Bonemerang misses the last turn.
 

Hogg

grubbing in the ashes
is a Tournament Directoris a member of the Site Staffis a Community Contributoris a Tiering Contributoris an Administrator
UU & Tour Head
Wasn’t all of that true with the previous suspect requirements also? The only difference is that the bar is higher. What is the magic cutoff that makes reqs acceptable at 77 GXE but impossible at 82?

When we picked this particular number we reviewed dozens of prior suspects. Even when the GXE requirement was 77 we still had people doing exactly what you described, and grinding several alts until one of them got a sufficiently good win streak to qualify. We set the reqs high enough that it is a lot harder to qualify based off of a couple of lucky streaks. Yes, that means it will be a lot harder to achieve reqs, and a particularly unlucky run can definitely bounce you.

If you’re getting discouraged, I would recommend waiting a few days for the ladder to develop. GXE and COIL grow more slowly on newer ladders where the average rating of the players is lower. The grind is always harder at the beginning. (This, by the way, is why we only shortened the ladder to 9 days instead of a week as some suggested; we wanted to allow the ladder some time to develop to keep reqs from being too difficult.)

Also, your input on this thread is valuable regardless of whether you successfully make reqs or not. I personally read every post and I know others do too, and good discourse is always important.
 
Wasn’t all of that true with the previous suspect requirements also? The only difference is that the bar is higher. What is the magic cutoff that makes reqs acceptable at 77 GXE but impossible at 82?

When we picked this particular number we reviewed dozens of prior suspects. Even when the GXE requirement was 77 we still had people doing exactly what you described, and grinding several alts until one of them got a sufficiently good win streak to qualify. We set the reqs high enough that it is a lot harder to qualify based off of a couple of lucky streaks. Yes, that means it will be a lot harder to achieve reqs, and a particularly unlucky run can definitely bounce you.

If you’re getting discouraged, I would recommend waiting a few days for the ladder to develop. GXE and COIL grow more slowly on newer ladders where the average rating of the players is lower. The grind is always harder at the beginning. (This, by the way, is why we only shortened the ladder to 9 days instead of a week as some suggested; we wanted to allow the ladder some time to develop to keep reqs from being too difficult.)

Also, your input on this thread is valuable regardless of whether you successfully make reqs or not. I personally read every post and I know others do too, and good discourse is always important.
Well, I'm going to actually say no to this for a few reasons that are inherantly intertwined.

First off, the main issue here is margin of error. With the "higher bar" set for this suspect tests, there is a considerably narrower margin of error to deal with bad matchups, bad luck, and occasional mistakes that aren't reflective of a players true ability. This, right off the bat, takes players out of the pool on sheer lack of motivation, and the less people playing on the suspect ladder means that it takes more time for the ladder to actually settle.

This further compounds with the result of the narrow margin of error, meaning more people are restarting and effecting taking alts out of the ladder and deflating the value of GXE. You also have people giving up, bringing people out of the ladder pool permanently, often taking from people who are still trying to get reqs.

This even further compounds when you include the people who are good but are getting unlucky, constantly taking out mid level people every time they restart an account.

All the while less and less people are playing the ladder - whether because they've gotten bored, given up, or even say, they actually got reqs. This means that the less people in the pool, the wider the gap between players on the ladder can be, meaning that the closer you get to reqs the more dangerous an unlucky loss can be as well as you're having to fight more and risk more since you're gaining less coil every battle then on a more active ladder.

A mistake I made in my first post was not making it clear enough that I didn't believe the old suspect system was perfect or even good - I believe the suspect system is inherantly bad, but when you raise the bar like this you're just magnifying the issues with the old way in an attempt to fix a smaller issue that was previously there.
 

pokeisfun

UU player and community member
is a Community Contributoris a Tiering Contributor
It was explicitly mentioned these requirements were supposed to be significantly harder, to the point where possibly less than 10 would qualify. We still have several more days after, as Hogg just mentioned, the requirements will get easier because the pool of points will increase (more stronger opponents). This is a new system so obviously it will have kinks, but it may be more prudent to let it run its course first. Nobody knows just how much easier the reqs will be after a few more days. I still have high hopes.

As of now though, there are already some things that can be improved, in my opinion:
  • Be explicit in the OP about the new reqs - if you were just starting UU, you wouldn't know these are harder reqs compared to every other suspect in every other tier
  • Be explicit in the OP about when to ladder - I doubt sincerely that some of the council could achieve reqs in early ladder. Many of us ladder early because we love UU, tell us we almost literally can't achieve reqs like this
  • Be involved in NP threads - many of us, including myself, are complaining now but both tier leaders told us this was coming a couple weeks ago. We dhould have been more proactive.
  • Be explicit in why we are moving to this new system. State our goals, in the OP, on what should be expected compared to previous suspects. Stating the goals on the onset is a way to ensure we fairly assess how well we met these goals after the test is over.
None of this seems very controversial. What likely will be controversial is after we finish this test, we see exactly how hard these reqs were. And if they had unintended consequences, like forcing a more stall perspective on voter(s) because it's easier to ladder with.

As I said I'm still hopeful. I don't see any reason to ask for a new system already when we have yet to see the complete effects of this system. But if the system really turns out to be as bad as Durza envisions, I hope we definitely work together to at least make improvements.
 

Ransei

Mr. Worldwide
is a Forum Moderatoris a Community Contributoris a Battle Simulator Moderator
Moderator
[12:33:24] pokemonisfun: only people who can go 50-0 have relevant opinions on buzzwole

This sums up the suspect experience pretty well, in my opinion.

I'm pretty sure everybody agrees that there is pretty much zero incentive to ladder now - whats the point of wasting time if you lose one match by hax out of 20 battles and realistically can't lose another to get reqs? On paper, making a higher GXE should mean that only better people will get reqs and ensuring a fair and more "accurate" suspect result. However, it doesn't take into account how easily it is to lose a game based on matchup or even bad luck in RNG alone. I understand the mentality of "raising the bar" to ensure higher quality suspects but in reality this is doing the opposite. I'll try to explain why:

GXE is inflated by more people winning, and obviously other people simultaneously losing battles. The more battles that happen on ladder, the more GXE potential will be injected into the ladder making reqs obviously easier than say, only a bunch of low ladder people who will never get reqs just continually playing for fun or just out of sheer ignorance of how the suspect system really works.

People make new alts for suspects. They make a few, to a lot of alts, depending on the difficultly of the suspect test at hand. All these alts start at 0 COIL / an undetermined GXE obviously, so when you have these low players naturally coming in, when they win they actually just take GXE out of the system by killing mid-rage people's GXE (because when you lose to somebody lower than you it hits you harder than say, if you lost to someone way higher than you). This kills suspect runs, and then those people make new alts, and the cycle continues. All the COIL "taken" from other opponents on the ladder can be gone after a loss or two when the person restarts reqs on a different alt, the old one never to be touched again.

The suspect system is flawed in that way - on a suspect ladder with GXE based on relativity from one opponent to another doesn't take into account that what the ladder reads as a "low ladder player" and re-calculates GXE in that way after the battle could very well be an excellent player who is draining GXE/COIL from the system that might just be dropped when they lose a single game. On a suspect ladder where everybody starts from zero, the "low ladder" is not the low ladder on normal ladders - they're just as likely to be a random as much as it is as likely for a given opponent to be an accomplished and competent UU player.

Now, another factor thats coming into play is the motivation factor. Many people in UU main chat (recently including me) have been expressing sentiment about there being no incentive to ladder the suspect test, on the grounds that its impossible to get reqs with a bar this high. This, along with the previously stated point about the suspect ladder system not accounting for "hax" and other subtleties that can determine the outcome of a battle, means that less people (particularly the decent players to extremely competent players who might not want to risk wasting there time on something that has such a high chance at zero return just based off a single loss) will be even on the ladder, meaning less inflation of COIL and you're just stuck with the same two types of people:

1) People who don't understand how the suspect tests work and that its impossible for them to get reqs / people just laddering who dont know what suspect tests are at all.

2) People who are desperately trying to get reqs and are constantly taking rating out of the system by continually making alts until they get to the point where they give up, or eventually make reqs.

The suspect system before this meant that generally people were of a generally lower quality than the UU council was looking for - and I agree, previous suspects had some low quality posts on both sides of the issue being suspected, but at least there were enough people to make a consensus among the at the least semi-serious members of the community and even have a reasonable chance at getting reqs, and then discussion optimally happening before the vote (which again, is just the ideal situation and I know that some people only get reqs for that TC badge, but thats not the primary issue).

I strongly disagree with what is happening right now with the decisions made in the tier - I understand they are made in the best interests of the tier (if some of you remember I was an avid supporter of the council vote on Serperior and Azumarill because I understood they were in the best interests of the tier) but at this point its a fake suspect and now really only really lucky people (I get it, you'll need skill too, but you first off need the luck to not get hax'd out of a game or two) or people who are eligible for reduced reqs are going to be able to vote in the suspect test.

This all seems pretty ironic to me since now to avoid poor suspect test results and votes they're making it so things are a high part determined by luck and rewarding the grind instead of just somewhat based off skill, and while there are going to be exceptions to this I doubt there will be very many.

TLDR: This suspect test isn't working.


edit: after discussion I have realized that this post exaggerates the amount of losses it takes to kill reqs - this was unintentional and was more focused on early-run losses than late game assuming you would potentially lose more later.
I'm not a part of this UU suspect but I have to say...



20/log2(40*87.5/2650)

If you have a dude on the top of the ladder with 87.5 GXE and 1568 elo with 35-0 and they aren't even close to reqs, there is a problem.


Edit: Isn't a suspect test meant to gather the opinions of many of our good players for whether something should or shouldn't be banned?

As for what pokemonisfun said, if suspect was so hard that only less than 10 people could qualify, why is this a suspect test, especially when it affects a large community?
 
Last edited:

Hogg

grubbing in the ashes
is a Tournament Directoris a member of the Site Staffis a Community Contributoris a Tiering Contributoris an Administrator
UU & Tour Head
He wouldn’t be close to reqs on the previous standards either. The COIL requirements never changed, only the game limit. His low COIL is largely a result of laddering in a new ladder where most of his opponents didn’t have a high rank. Again, COIL rises faster once the ladder is more developed.

EDIT: To clarify, we are using the exact same formula with the exact same COIL requirements we've always used. Literally the only thing that has changed is that we now we are only accepting people who achieve the COIL requirements in 65 games or fewer, whereas previously we accepted people who achieved those requirements in 90 games or fewer. The ladder mechanics and rate at which COIL accrues have not changed.
 
Last edited:
I think getting reqs will prove too difficult and doubt that half of the uu council even has the ability to meet those reqs. Anyway, thoughts on suspect:<br />
<br />
Breloom: Stay UU, because there are so many good grass pokemon and pokemon that resist both grass and fighting. If you have a grass pokemon like celebi or amoongus, breloom is very easy to play against, since you can just switch celebi or amoongus in to breloom. If you do not have a grass pokemon, you will have to let breloom put one of your pokemon to sleep. Once one of your pokemon is asleep, you can just use your grass/fighting resist, like crobat or togekiss to switch in for the rest of the game.

Buzzwole: Ban, because it has zero safe switch ins, and is very hard to scout due to its high attack. Buzzwole has an extremely good movepool that can cover the entire meta, and can use subsitute to ease prediction. This enables it to get a kill at least half the time it comes in. Ive been abusing buzzwole a lot to show people how broken it is, and here is an example: http://replay.pokemonshowdown.com/gen7uususpecttest-696275697

In this game, buzzwole easily killed crobat, since my opponent was predicting me to use substitute or a fighting move. Later in the game, buzzwole was able to get a sub on cobalion due its immense defense stat, knock out cobalion, and then one shot suicune due to beast boost. It's obvious that buzzwole is too powerful for uu.
 
I've been watching the ladder and I've only seen 3 people actually get reqs despite the Suspect Test nearly ending. This is also a problem because not everyone who has reqs will vote, and not everyone who would have voted before can vote now. I'm going to be honest here in that I have to agree with everyone else here, the Suspect Test here really isn't working. I argue that you're going to have a lot more biased results than you anticipated with less voters because if only 3 people get reqs and two of them vote to ban, then it's banned by technicality. And you could also argue otherwise, if only one person votes ban but the other two don't vote, then it's not likely going to get banned. I believe that having an extremely small sample size (<10 voters) can actually have negative effects on a suspect test.

I geniunely believe this has been taken too far. I don't believe making the reqs harder is necessarily bad per se, but I feel it was made so much harder out of reaction of the Mega Latias suspect. Ever since the Mega Latias Suspect Test, you were council voting on almost everything up until you implemented this system, and so few people are actually getting reqs that this is effectively no different from a council vote to begin with.

I do feel like the council votes were in good faith after all, I just don't ultimately believe that less than 10 people should have an effect on a Suspect Test. I mentioned the votes because I think the Mega Latias suspect upset the council and lead to this overly restrictive Suspect Test we have now. I agree after all that a higher req threshold may help improve the quality of suspects, I also think that a lower amount of voters will decrease the quality of suspects as well if there are too few voters.

This is just my thought on the matter.
 

Zarel

Not a Yuyuko fan
is a member of the Site Staffis a Battle Simulator Administratoris a Programmeris a Pokemon Researcheris an Administrator
Creator of PS
EDIT: To clarify, we are using the exact same formula with the exact same COIL requirements we've always used. Literally the only thing that has changed is that we now we are only accepting people who achieve the COIL requirements in 65 games or fewer, whereas previously we accepted people who achieved those requirements in 90 games or fewer. The ladder mechanics and rate at which COIL accrues have not changed.
I don't understand the math here.

COIL stands for Convergent Order-Invariant Ladder. Why are you using a convergent ladder, and then intentionally prevent it from converging by having a maximum number of games? Antar did you intend for COIL to be used this way?

I feel like if your goal is to have a "true GXE" requirement of 82, someone like pokemonisfun whose UU GXE is 92 should not have trouble qualifying.

When we picked this particular number we reviewed dozens of prior suspects. Even when the GXE requirement was 77 we still had people doing exactly what you described, and grinding several alts until one of them got a sufficiently good win streak to qualify. We set the reqs high enough that it is a lot harder to qualify based off of a couple of lucky streaks. Yes, that means it will be a lot harder to achieve reqs, and a particularly unlucky run can definitely bounce you.
I'm not an expert on COIL, but I think you should be solving this problem by increasing your COIL B-factor (which increases the minimum number of games), not by instituting a maximum number of games. Giving a maximum game count makes it easier to qualify off win streaks and harder to qualify off skill, which is the opposite of what you want.
 

Hogg

grubbing in the ashes
is a Tournament Directoris a member of the Site Staffis a Community Contributoris a Tiering Contributoris an Administrator
UU & Tour Head
I'm not an expert on COIL, but I think you should be solving this problem by increasing your COIL B-factor (which increases the minimum number of games), not by instituting a maximum number of games. Giving a maximum game count makes it easier to qualify off win streaks and harder to qualify off skill, which is the opposite of what you want.
Correct me if I'm wrong (and I very well could be, math isn't my strong suit), but wouldn't raising the B value mean even more games would be required in order to qualify? So in this particular example, pokemonisfun would be even further away from qualifying to participate in this suspect test. If, for example, we raised the B value to 25, by my math it would mean it would take over 80 games with a GXE of 82 to qualify, as opposed to the 65 games it should currently require. Nor would this prevent people from qualifying with a lower GXE: someone could still qualify with, say, a 75 GXE, as long as they were willing to grind out about 140 games. (That definitely happens, by the way—before we instituted a game limit, we once had someone qualify by playing over 400 games).

Our goal is to establish a minimum GXE in order to qualify to participate in this suspect test. If possible, we'd like to do it without making it an absurdly long process to qualify either. (So, arbitrarily raising the B value to the point where it is practically impossible to qualify with a GXE below 82 is impractical, as it would mean drastically increasing the number of games required even for those with a GXE above the current standards.) The only way that I know of to accomplish this is by establishing a game limit. If there's a way to modify the formula to get this end-result WITHOUT requiring a game limit, we are absolutely open to it.

We're open to modifying the standards we've established for this suspect test if there's a better way to reach our end goal. This isn't set in stone. It's also been about 16 years since the last time I took a math class (I was an English major ffs), so if there are issues with my math please let me know and we'll modify things appropriately.
 

Hogg

grubbing in the ashes
is a Tournament Directoris a member of the Site Staffis a Community Contributoris a Tiering Contributoris an Administrator
UU & Tour Head
Alright, I talked this over with Zarel for a while and discussed things with Hikari.

We will be changing the current requirements to the following: 2650 COIL with a minimum GXE of 82. There is no longer a game limit.

This is functionally the same as far as end-results go, in that we are raising the bar for what qualifies. On the other hand, by removing the game limit, we avoid a situation where winning is ever a bad thing, we promote a more stable ladder by making it less advantageous to constantly restart from a new alt, and we put early ladderers at less of a disadvantage (since you accrue COIL more slowly in a new ladder, and therefore you need an even higher GXE to qualify within the game limit as compared to someone who waits to ladder until later).

So, to reiterate: there is no longer a game limit for the suspect. To qualify, you must demonstrate that you have at least 2650 COIL and a minimum GXE of 82. Those are your only requirements.

We have also been discussing a new suspect method that does not involve COIL at all for future suspects (the general suggestion being a non-COIL ladder with both a minimum game and minimum GXE requirement, similar to our current council reqs). However, that will take a bit more time to figure out what the right numbers are, so if we decide to implement it, it will not happen until a later suspect.

Thanks to everyone for their feedback in this process.
 

warzoid

wreck into a poll like a Raichu volt
is a Tiering Contributor Alumnus
COIL can be used to require a minimum GXE, but the trick is to change the COIL requirement, not the B-value. Here is the formula for games needed to qualify:

N = B / log2(40 * GXE / C)

Since the logarithm of anything less than 1 is negative and you can't play a negative number of games, the minimum GXE required to qualify is described by:

minGXE > C / 40

So, for instance, we could set the COIL requirement to 3250, which would allow only players with over 81.25% GXE to qualify. With a B-value of 2, players with 82% GXE will need to play at least 151 games, players with 83% need at least 65, players with 84% need at least 41, and so on. The nice thing is that there's no required minimum number of games, but even if someone grinds out an absurd number of games, they'll still need more than 81.25% GXE to make the cut.
 

Zarel

Not a Yuyuko fan
is a member of the Site Staffis a Battle Simulator Administratoris a Programmeris a Pokemon Researcheris an Administrator
Creator of PS
I already convinced Hogg on Discord about why a max gamecount is bad, but in case anyone else wanted answers to his questions:

Correct me if I'm wrong (and I very well could be, math isn't my strong suit), but wouldn't raising the B value mean even more games would be required in order to qualify?
Yes. Requiring more games means lucky people with get lower scores (the more games you play, the closer your GXE will be to average). If you had problems with people getting lucky, requiring more games would be the correct solution.

So in this particular example, pokemonisfun would be even further away from qualifying to participate in this suspect test. If, for example, we raised the B value to 25, by my math it would mean it would take over 80 games with a GXE of 82 to qualify, as opposed to the 65 games it should currently require.
That's true, but without the minimum gamecount, his path to qualification is much more straightforward.

Previously, you were asking him to thread a needle: hope his COIL increases above a specific number before his gamecount increases above a specific number. This is very frustrating to do because, among other things, practically no one understands the math behind COIL (I do but the more time I spend talking about it, the more I realize no one else does).

Afterwards, the process is a lot more straightforward. Increase COIL by winning games. That's a lot more doable.

Nor would this prevent people from qualifying with a lower GXE: someone could still qualify with, say, a 75 GXE, as long as they were willing to grind out about 140 games. (That definitely happens, by the way—before we instituted a game limit, we once had someone qualify by playing over 400 games).
If you think someone with a 75 GXE shouldn't qualify, your minimum COIL should be above 3000. (If this makes it too hard for other players to make reqs, B could be decreased accordingly.)

I think the biggest mistake here is definitely misusing the math behind COIL. Having spent the last hour talking to tier leaders about COIL, I think the correct solution is probably to stop using COIL entirely.

Our goal is to establish a minimum GXE in order to qualify to participate in this suspect test. If possible, we'd like to do it without making it an absurdly long process to qualify either. (So, arbitrarily raising the B value to the point where it is practically impossible to qualify with a GXE below 82 is impractical, as it would mean drastically increasing the number of games required even for those with a GXE above the current standards.) The only way that I know of to accomplish this is by establishing a game limit. If there's a way to modify the formula to get this end-result WITHOUT requiring a game limit, we are absolutely open to it.
By this point, I'm nearly convinced about what the correct solution should be:

Require a minimum GXE of 82 to qualify for suspect voting.

That's it. That's the only requirement you need. You don't need COIL. You probably don't even need a minimum game-count; you already need so many games to reach a GXE that high, anyway. Every other requirement is just confusing and frustrating players.
 

warzoid

wreck into a poll like a Raichu volt
is a Tiering Contributor Alumnus
I agree with almost everything Zarel said in his last post, except this:
By this point, I'm nearly convinced about what the correct solution should be:

Require a minimum GXE of 82 to qualify for suspect voting.

That's it. That's the only requirement you need. You don't need COIL. You probably don't even need a minimum game-count; you already need so many games to reach a GXE that high, anyway. Every other requirement is just confusing and frustrating players.
In my experience, a good player with a good team hits 82% GXE in 18 to 25 games. The first 10 of those games are low ladder curbstomps. I don't think that a simple GXE limit requires the top players to play enough games to get a feel for the suspect meta.

And doing away with COIL may not be necessary. Raising awareness of the fact that COIL has a built-in adjustable minimum GXE requirement may be enough.
 
We will be changing the current requirements to the following: 2650 COIL with a minimum GXE of 82. There is no longer a game limit.
You changed the requirements... in the middle of the suspect test... right after I went full tryhard to get it under 65 games...



May as well post my thoughts now I guess. Buzzwole and Breloom both have solid counters. Checks like Amoonguss and Sap Sipper Azu counter Breloom quite hard, Buzzwole also counters Breloom. Pokemon like Crobat, Chandelure and Moltres check Buzzwole quite well. Honestly, from my encounters on ladder, I didn't feel particularly threatened by either of these two mons, so I don't think that they warrant a ban. There is still some negative stigma surrounding Buzzwole from people's experience with it during its previous drop a few months ago, but the meta has changed significantly since then, and Buzzwole is no longer the same threat it once was.

The team I used, http://pokepast.es/2117e117729a464b
 
Last edited:

explodingdaisies

What's the point of talking if nobody ever listens
is a Battle Simulator Driver
Checks like Amoonguss and Sap Sipper Azu counter Breloom quite hard, Buzzwole also counters Breloom
Don't forget the Lansat NG flying breloom set that's been floating around. It nails amoon and buzz after +2. It also gets worry seed, which most people don't realize or haven't mentioned. I'm still testing out, so stay tuned for my thoughts on both.
 

Zarel

Not a Yuyuko fan
is a member of the Site Staffis a Battle Simulator Administratoris a Programmeris a Pokemon Researcheris an Administrator
Creator of PS
I agree with almost everything Zarel said in his last post, except this:
In my experience, a good player with a good team hits 82% GXE in 18 to 25 games. The first 10 of those games are low ladder curbstomps. I don't think that a simple GXE limit requires the top players to play enough games to get a feel for the suspect meta.
Yeah, that's entirely fair. I asked some Smogoners whether or not a minimum game-count would be necessary with a GXE floor of 82 earlier, and someone said no.

In which case, minimum GXE + minimum game count is probably the best solution here.

And doing away with COIL may not be necessary. Raising awareness of the fact that COIL has a built-in adjustable minimum GXE requirement may be enough.
That doesn't solve the problem of "people near the limit will need to play very many battles before they qualify", which tier leaders don't like. The best workaround for that is to have both a minimum COIL and a minimum GXE, but honestly, a minimum GXE and a minimum game count is much simpler, and doesn't require roping in programmers to adjust the COIL B factor for every ladder.
 

Kink

it's a thug life ¨̮
is a Forum Moderatoris a Community Contributoris a Contributor to Smogonis a Tutor Alumnusis a Tiering Contributor Alumnus
Moderator
I have a few things to say about this suspect.

Firstly, regardless of what the original intent of COIL, regardless of the difficulty of there being a 65 game limit to achieve 2650 with 82% GXE, and all of that nonsense, I think a lot of you are totally looking at this shit the wrong way.
I believe the suspect system is inherantly bad, but when you raise the bar like this you're just magnifying the issues with the old way in an attempt to fix a smaller issue that was previously there.
I'm sorry but I disagree with this entirely. What this does is ween the above average from the good or even great, which I find nothing wrong with. There are too many players out there that fail to see the nuanced threads of competitive mons, especially a tier like UU where we need to test everything that falls from OU and be the first to determine the relative strength of a mon for all subsequent lower tiers. These examples are elements that'll affect tournament play and more importantly the health and competitiveness of the tier.

What do you think the point of 82% GXE is anyway? For Charizardlover9000 to easily achieve a voting position in a tier? If Charizardlover9000 wants to achieve reqs, I'd hope the experience is difficult for them. Most certainly, they'll learn a thing or two about UU and hopefully achieve a higher grasp of why the voting process is so important, and why it's important to try your best to learn about the tier you're voting on.

https://puu.sh/ze9QQ/e074c2159c.png
https://puu.sh/ze9C2/9c9cea5a25.png

I went 46-11 to get reqs in 57 games (and I'm definitely not in top form), then other people should get it in 65, and they have. The fact that a user like cookees managed to achieve reqs speaks not only to his dedication to be a good player, but that newcomers that care enough to become a good player will also be able to achieve reqs. Y'all need to believe in yourself and achieve the achievable, again, regardless of game limits, or restrictions. I promise, it's possible, and you'll all be better players as a result.

Lastly, breloom is fine, versatile, priority, good variety of sets, nothing too crazy. Houndoom is fine. Buzz is pretty insane and I'm 50/50 in terms of banning right now, I'll post a more decisive thought on it shortly.
 

pokeisfun

UU player and community member
is a Community Contributoris a Tiering Contributor
I have a few things to say about this suspect.

Firstly, regardless of what the original intent of COIL, regardless of the difficulty of there being a 65 game limit to achieve 2650 with 82% GXE, and all of that nonsense, I think a lot of you are totally looking at this shit the wrong way.

I'm sorry but I disagree with this entirely. What this does is ween the above average from the good or even great, which I find nothing wrong with. There are too many players out there that fail to see the nuanced threads of competitive mons, especially a tier like UU where we need to test everything that falls from OU and be the first to determine the relative strength of a mon for all subsequent lower tiers. These examples are elements that'll affect tournament play and more importantly the health and competitiveness of the tier.

What do you think the point of 82% GXE is anyway? For Charizardlover9000 to easily achieve a voting position in a tier? If Charizardlover9000 wants to achieve reqs, I'd hope the experience is difficult for them. Most certainly, they'll learn a thing or two about UU and hopefully achieve a higher grasp of why the voting process is so important, and why it's important to try your best to learn about the tier you're voting on.

https://puu.sh/ze9QQ/e074c2159c.png
https://puu.sh/ze9C2/9c9cea5a25.png

I went 46-11 to get reqs in 57 games (and I'm definitely not in top form), then other people should get it in 65, and they have. The fact that a user like cookees managed to achieve reqs speaks not only to his dedication to be a good player, but that newcomers that care enough to become a good player will also be able to achieve reqs. Y'all need to believe in yourself and achieve the achievable, again, regardless of game limits, or restrictions. I promise, it's possible, and you'll all be better players as a result.

Lastly, breloom is fine, versatile, priority, good variety of sets, nothing too crazy. Houndoom is fine. Buzz is pretty insane and I'm 50/50 in terms of banning right now, I'll post a more decisive thought on it shortly.

You talk about how players can't see nuance but then dismiss Breloom as not broken in a single sentence.

And by the way it's not like you got reqs with the first requirement of under 65 games on your first try, so to say it's easy for good players like you - and we all know you're a very very good player - is puzzling to say the least.
 

Kink

it's a thug life ¨̮
is a Forum Moderatoris a Community Contributoris a Contributor to Smogonis a Tutor Alumnusis a Tiering Contributor Alumnus
Moderator
You talk about how players can't see nuance but then dismiss Breloom as not broken in a single sentence.

And by the way it's not like you got reqs with the first requirement of under 65 games on your first try, so to say it's easy for good players like you - and we all know you're a very very good player - is puzzling to say the least.
there's nothing that I can say about breloom that hasn't already been said. I'm on the unban side of things. nor am I a council member, so I don't necessarily commit to explicating my entire reasoning. In this case, I don't want to affect the mindset of too many users, since the requirements are difficult enough for people to make their own decisions without listening to my nonsense.

You're right, I did it on my second try after the ladder stabilized. It took all of about 3-4 hours total. Other people have also achieved it. It's an achievable feat. I think more good comes from trying to challenge yourself versus making excuses. I don't think I ever said it was "easy", I said it's achievable. With the added value that because it's a more difficult achievement, the rewards that come out of it are also greater. Some rewards aren't immediately tangible, they only start to take form after months or years. I see this as part of that facilitative process.

Y'all need to believe in yourself and achieve the achievable regardless of game limits, or restrictions. I promise, it's possible, and you'll all be better players as a result.
 
Status
Not open for further replies.

Users Who Are Viewing This Thread (Users: 1, Guests: 0)

Top