Guild Wars 2

The change to 2-1-1 made the matches more balanced than any changes made before. It allows worlds to quickly adjust their glicko and find balanced matches.

Changing it to 5-4-3 will prevent the quick adjustment based on glicko and cause bad matchups to continue for much longer.

Please reconsider this change, this will cause problems when the next relinking occurs.

You need to math it, the 5-4-3 spread is better.

Sorry.

I’m very happy that Anet is prepared to change the scoring. To be honest, I quite like 2-1-1, because of the potential for upsets, but it does seem to engender a, “we’ve no chance at first, so why try at all” resigned attitude rather than the more desirable, “go for first place!”.

This is, as ever, the result of terrible dominating stacked servers which nobody can/will do anything about. (Posted Feb 2017).

At this stage scoring changes do seem a lot like a deckchair rearrangement exercise.

How about they change it, we live with it for a couple matches and see how it is BEFORE we start criticizing it and asking for it to be removed. You may be right and it doesn’t work as well but we probably won’t know all the ramifications both good and bad until they try it.

Cerby.1069:

Ummmmm no its not. Each team gets one of the three numbers each round. And the DIFFERENCE between the numbers is 1-2 units…teh exact same as the 3-2-1 system. The final scores under a 5-4-3 system will be the same as the scores under a (3-2-1) multiplied by a coefficent. meaning the scores are simply inflated versions of the originals

It’s worse than that because of the way glicko adjusts itself. With 3-2-1 we required a lot of glicko manipulation to get competitive matches. 2-1-1 made the process faster without the need to adjust, and 5-4-3 will slow the process down.

Straegen.2938:

How about they change it, we live with it for a couple matches and see how it is BEFORE we start criticizing it and asking for it to be removed. You may be right and it doesn’t work as well but we probably won’t know all the ramifications both good and bad until they try it.

We tried it, 5-4-3 is the same as 3-2-1.

I don’t understand,

If it was the same wouldn’t it be 6-4-2?

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

Justine.6351:

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

Think about it carefully. The absolute difference between each place is exactly the same if it is 5-4-3 or 3-2-1. It is exactly as difficult to overcome a deficit in 3-2-1 and 5-4-3.

What’s different is that glicko ratings are changed by the proportional numbers and not the absolute numbers. So in terms of a server moving to the correct tier, 5-4-3 is slower and more unbalanced than 3-2-1, which is in turn slower and more unbalanced than 2-1-1.

The is nothing magical about this math and the results will be predictable.

This linking has more or less balanced matches at the moment, so we won’t notice any problems until we are relinked. Then the blowout matches will occur for several weeks until anet steps in to manually adjust them. The tragedy is that 2-1-1 means they don’t have to and this is a step in the wrong direction.

It’s action being confused with progress.

Something like 100-55-45 would function similarly to 2-1-1 getting us to good matchups while rewarding the second place and still allowing glicko to move. 5-4-3 will just cause problems.

Justine.6351:

I don’t understand,

If it was the same wouldn’t it be 6-4-2?

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

Think about the point spread between the winner and loser after a couple skirmishes.
The winner has 10 the loser has 6, that’s a 4 point difference. With 3-2-1 the winner would have 6 and the loser would have 2, still a 4 point difference. It’s exactly the same just with bigger numbers, that’s why this “change” is completely pointless.

Jim Hunter.6821:

Justine.6351:

I don’t understand,

If it was the same wouldn’t it be 6-4-2?

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

Think about the point spread between the winner and loser after a couple skirmishes.
The winner has 10 the loser has 6, that’s a 4 point difference. With 3-2-1 the winner would have 6 and the loser would have 2, still a 4 point difference. It’s exactly the same just with bigger numbers, that’s why this “change” is completely pointless.

That’s a horrible fail at math.

On 5 matches,

6-4-2, 30-20-10, 20 point difference

5-4-3, 25-20-15, 10 point difference

Justine.6351:

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

No. Why should it? The winner of a matchup is often totally dominating and will win pretty much every skirmish. There is no competition if the second and third place stand zero chance. At least 5/4/3 is better than 2/1/1 which keep them at status quo while the dominating server just pulls ahead. That said it’s no different than 3/2/1 of course.

I fail to see the point of these high level score experiments. It’s not going to fix anything because it cant be fixed that way. Score gains need to be balanced on a low level, ie the worth of objectives over time, comeback mechanics, putting abandoned borders in sleep mode, avoiding runaway points by reducing warscore min/max difference, etc…

IMO the best score system we’ve had so far was the old 3/2/1 just before they added the kittenedly high tier bonuses. Far from perfect, but the shorter timeslices (15>5) and lowered PPT differences overall actually felt pretty good. After the tier bonuses we immiedetly saw 150-200+ PPT for the dominating server and all you could do was sigh…

McKenna Berdrow.2759:

Since this is a similar to the other 5-4-3 thread, here is a link to my response on that thread.

if you continue on reading though everyone debunks and kicks the theory out of the water explaining and showing it is the same as 3,2,1

Hesacon.8735:

Justine.6351:

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

Think about it carefully. The absolute difference between each place is exactly the same if it is 5-4-3 or 3-2-1. It is exactly as difficult to overcome a deficit in 3-2-1 and 5-4-3.

I’m not so sure that is true.

1-2-3 is,
First place gets 150% value of second
Second gets 100% value of third
Third gets 33% value of first

3-4-5 is,
First place gets 125% value of second
Second gets 133% value of third
Third gets 66% value of first

Dawdler.8521:

Justine.6351:

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

No. Why should it? The winner of a matchup is often totally dominating and will win pretty much every skirmish. There is no competition if the second and third place stand zero chance. At least 5/4/3 is better than 2/1/1 which keep them at status quo while the dominating server just pulls ahead. That said it’s no different than 3/2/1 of course.

I fail to see the point of these high level score experiments. It’s not going to fix anything because it cant be fixed that way. Score gains need to be balanced on a low level, ie the worth of objectives over time, comeback mechanics, putting abandoned borders in sleep mode, avoiding runaway points by reducing warscore min/max difference, etc…

IMO the best score system we’ve had so far was the old 3/2/1 just before they added the kittenedly high tier bonuses. Far from perfect, but the shorter timeslices (15>5) and lowered PPT differences overall actually felt pretty good. After the tier bonuses we immiedetly saw 150-200+ PPT for the dominating server and all you could do was sigh…

That’s quitter talk. Go EotM or something.

Jim Hunter.6821:

Justine.6351:

I don’t understand,

If it was the same wouldn’t it be 6-4-2?

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

Think about the point spread between the winner and loser after a couple skirmishes.
The winner has 10 the loser has 6, that’s a 4 point difference. With 3-2-1 the winner would have 6 and the loser would have 2, still a 4 point difference. It’s exactly the same just with bigger numbers, that’s why this “change” is completely pointless.

Indeed, the end result for scoring is the same no matter if it’s 3-2-1, 5-4-3 or 1000-999-998. The difference is in how glicko calculate the ratings based on scores… It use the % of the score between the the highest ranked and the rest to make the changes. With higher numbers you make the % much closer which will lessen the rating gain/loss.

For exemple: If a match had 100 skirmishes and ended at 300-200-100, the score % difference between the worlds is higher than if it was 500-400-300 (100 is 33% of 300 while 300 is 60% of 500). The higher the difference, more rating you gain/loss based on the rating difference between worlds.

All the yadda-yadda about how glicko works aside, I honestly rather take 3-2-1 over a similar system with higher numbers because the huge glicko swings actually work for varieirty. Staleness is terrible, and we already had years of stale matchups in the past.

Jeknar.6184:

Jim Hunter.6821:

Justine.6351:

I don’t understand,

If it was the same wouldn’t it be 6-4-2?

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

Think about the point spread between the winner and loser after a couple skirmishes.
The winner has 10 the loser has 6, that’s a 4 point difference. With 3-2-1 the winner would have 6 and the loser would have 2, still a 4 point difference. It’s exactly the same just with bigger numbers, that’s why this “change” is completely pointless.

Indeed, the end result for scoring is the same no matter if it’s 3-2-1, 5-4-3 or 1000-999-998. The difference is in how glicko calculate the ratings based on scores… It use the % of the score between the the highest ranked and the rest to make the changes. With higher numbers you make the % much closer which will lessen the rating gain/loss.

For exemple: If a match had 100 skirmishes and ended at 300-200-100, the score % difference between the worlds is higher than if it was 500-400-300 (100 is 33% of 300 while 300 is 60% of 500). The higher the difference, more rating you gain/loss based on the rating difference between worlds.

All the yadda-yadda about how glicko works aside, I honestly rather take 3-2-1 over a similar system with higher numbers because the huge glicko swings actually work for varieirty. Staleness is terrible, and we already had years of stale matchups in the past.

Oh so it will be varied matches easier?

Justine.6351:

Oh so it will be varied matches easier?

No, it’s the oposite. 5-4-3 is a 3-2-1 where it is harder for the servers move up/down. The larger the end score gaps, the easier it becomes to move tiers. So if you are lowering these gaps (by using higher numbers) you are essentially making it harder.

Justine.6351:

Hesacon.8735:

Justine.6351:

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

Think about it carefully. The absolute difference between each place is exactly the same if it is 5-4-3 or 3-2-1. It is exactly as difficult to overcome a deficit in 3-2-1 and 5-4-3.

I’m not so sure that is true.

1-2-3 is,
First place gets 150% value of second
Second gets 100% value of third
Third gets 33% value of first

3-4-5 is,
First place gets 125% value of second
Second gets 133% value of third
Third gets 66% value of first

This is accurate. All the geniouses in this thread going on about the absolute difference between the scores have clearly forgotten that there’s more than one way to compare numbers. I suppose they think there’s no difference between a 50% pay raise and a $1 pay raise either?

McKenna Berdrow.2759:

Since this is a similar to the other 5-4-3 thread, here is a link to my response on that thread.

And you say:

McKenna Berdrow.2759:

Why 5,4,3?
How is 5,4,3 different than 3,2,1?
Since there has been a lot of debate on this, with most people being right but debating different points, there are three things that are true:

1. The scores will be closer.
2. The placements won’t change.
3. This could impact world rating used for glicko matchmaking.

And I argued:

Hesacon.8735:

Changing it to 5-4-3 will prevent the quick adjustment based on glicko and cause bad matchups to continue for much longer.

Because:

Hesacon.8735:

Cerby.1069:

Ummmmm no its not. Each team gets one of the three numbers each round. And the DIFFERENCE between the numbers is 1-2 units…teh exact same as the 3-2-1 system. The final scores under a 5-4-3 system will be the same as the scores under a (3-2-1) multiplied by a coefficent. meaning the scores are simply inflated versions of the originals

It’s worse than that because of the way glicko adjusts itself. With 3-2-1 we required a lot of glicko manipulation to get competitive matches. 2-1-1 made the process faster without the need to adjust, and 5-4-3 will slow the process down.

And you also said:

McKenna Berdrow.2759:

Do we expect 5,4,3 to be a cure-all for scoring problems?
No. 5,4,3 will not make match scores dramatically better than 3,2,1. The scores might look closer, but if a world is dominating 5,4,3 isn’t going to stop them from winning. Population imbalance is the biggest factor in scoring problems and changing the skirmish score values isn’t going to change that. Population imbalance is a different problem that we are working on solving outside of changing the skirmish score values.

To which I argued:

Hesacon.8735:

Think about it carefully. The absolute difference between each place is exactly the same if it is 5-4-3 or 3-2-1. It is exactly as difficult to overcome a deficit in 3-2-1 and 5-4-3.

What’s different is that glicko ratings are changed by the proportional numbers and not the absolute numbers. So in terms of a server moving to the correct tier, 5-4-3 is slower and more unbalanced than 3-2-1, which is in turn slower and more unbalanced than 2-1-1.

And can also predict now:

Hesacon.8735:

The is nothing magical about this math and the results will be predictable.

This linking has more or less balanced matches at the moment, so we won’t notice any problems until we are relinked. Then the blowout matches will occur for several weeks until anet steps in to manually adjust them. The tragedy is that 2-1-1 means they don’t have to and this is a step in the wrong direction.

So again I beseech you: please don’t make this change. We will be in a matchmaking limbo that will require Anet’s intervention to resolve and frustrate a lot of players during the repeat weeks of bad matchups.

Fay.2357:

Justine.6351:

Hesacon.8735:

Justine.6351:

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

Think about it carefully. The absolute difference between each place is exactly the same if it is 5-4-3 or 3-2-1. It is exactly as difficult to overcome a deficit in 3-2-1 and 5-4-3.

I’m not so sure that is true.

1-2-3 is,
First place gets 150% value of second
Second gets 100% value of third
Third gets 33% value of first

3-4-5 is,
First place gets 125% value of second
Second gets 133% value of third
Third gets 66% value of first

This is accurate. All the geniouses in this thread going on about the absolute difference between the scores have clearly forgotten that there’s more than one way to compare numbers. I suppose they think there’s no difference between a 50% pay raise and a $1 pay raise either?

You misspelled geniuses, so there’s that.

The problem is not the absolute difference in numbers, that is totally irrelevant since the 1st, 2nd, and 3rd place servers will be exactly the same.

No, the problem is that the proportional differences that are listed in the nested quote are going to prevent glicko from adjusting the matchups and we will have worlds locked into tiers they don’t belong in, with trouble moving up or down to get a more satisfying matchup.

The math within the nested quote is the exact point when you take the next logical step in match making!

Hesacon.8735:

Fay.2357:

Justine.6351:

Hesacon.8735:

Justine.6351:

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

Think about it carefully. The absolute difference between each place is exactly the same if it is 5-4-3 or 3-2-1. It is exactly as difficult to overcome a deficit in 3-2-1 and 5-4-3.

I’m not so sure that is true.

1-2-3 is,
First place gets 150% value of second
Second gets 100% value of third
Third gets 33% value of first

3-4-5 is,
First place gets 125% value of second
Second gets 133% value of third
Third gets 66% value of first

This is accurate. All the geniouses in this thread going on about the absolute difference between the scores have clearly forgotten that there’s more than one way to compare numbers. I suppose they think there’s no difference between a 50% pay raise and a $1 pay raise either?

You misspelled geniuses, so there’s that.

The problem is not the absolute difference in numbers, that is totally irrelevant since the 1st, 2nd, and 3rd place servers will be exactly the same.

No, the problem is that the proportional differences that are listed in the nested quote are going to prevent glicko from adjusting the matchups and we will have worlds locked into tiers they don’t belong in, with trouble moving up or down to get a more satisfying matchup.

The math within the nested quote is the exact point when you take the next logical step in match making!

If you read the actual thread, you’ll see plenty of idiots making stupid conparisons to arbitrarily large scores with an absolute difference between them of one. You’re basically the only person making a coherent argument.

That being said, you’re operating under the assumption that large and rapid swings in glicko rating is a desired outcome of the scoring system. This is not an appropriate assumption to make. In fact, I would argue that you’re going in the entirely wrong direction with that assumption. The glicko system itself contains a variety of levers that can be adjusted to change how it reacts to inputs. Your complaint, that the glicko rating won’t see large enough changes with 543, is potentially valid. However, going back to 321 isn’t the answer to that complaint. There’s another lever that should be used to change that: volatility.

The thing about glicko is that it’s based on the assumption that the scores are accurately representative of the strength of what they measure. If one score is twice another score, glicko assumes that whomever the first score belongs to is twice as strong as whomever the second score belongs to. Therefore, the scoring system should strive to most accurately represent that relative power levels of the servers; it shouldn’t be the tool you use to try and increase the volatility of the matches. 211 was obviously flawed in this regard, and potentially so were 543 and 321. However, argumemts for one or another scoring system should be made on the proper basis of whether they accurately represent server strength.

+1 Hesacon.8735, your argument is great and really well laid out for all to understand. My argument is gonna be more of the same but its less civil, as is my nature.

McKenna Berdrow.2759:

Since this is a similar to the other 5-4-3 thread, here is a link to my response on that thread.

Read ur response. The first response to it on that other thread is pretty much my sentiments exactly, I’ll add some more detail though:

Tl:DR: anet thinks by inflating the scores and tossing bigger numbers around it will make the system confusing and complicated, to the extent that the average player will THINK the scores are closer and that their skirmishes matter more than they actually do.
For example: you will have occasions where a score ‘like’ 44-46-40 where the numbers are close may exist (idk if those specific values are possible but they just have to be close). Players see the winner as getting 5 points so they say “hmmm if our server wins we get to 45, we go from 3rd to 2nd place!” It is of course after the fact that at most they will only move 2 points closer towards bridging the gap to the 1st place server. kitten de kitten

In other words they think you are stupid. And by capitalizing on ur stupidity they can make the game seem more fair. Given fake news is all the rage these days, this is basically a low effort form of fake news where you present an old concept as something ‘new and brilliant’, when infact it is the same old concept with inflated numbers.

I’m even less impressed by this move now. Link a response that makes me MORE impressed.

Then there’s the glicko which I didn’t consider when I first posted. And those arguments are correct, it makes it harder to break the glicko and get new pairings compared to 3-2-1. So they are slowying down progress, calling you stupid, and telling you this method is new and exciting combining the best of the last two attempts and deserves its own trial period before we comment on it. So yet another excuse to slow down progress on development of the mode, and pretend like they did work the previous 2 months.
Its brilliant. They get to say they are working on the mode, while not doing any work on it for another 2 months.

Srsly, we should boycott at this point.
But we know we can’t hold a dead kitten hostage…..so there’s no point

Cerby.1069:

McKenna Berdrow.2759:

Since this is a similar to the other 5-4-3 thread, here is a link to my response on that thread.

Read ur response. The first response to it is pretty much my sentiments exactly, I’ll add some more detail though:

Tl:DR: anet thinks by inflating the scores and tossing bigger numbers around it will make the system confusing and complicated, to the extent that the average player will THINK the scores are closer and that their skirmishes matter more than they actually do.
*For example: you will have occasions where a score ‘like’ 44-46-40 where the numbers are close may exist (idk if those specific values are possible but they just have to be close). Players see the winner as getting 5 points so they say “hmmm if our server wins we get to 45, we go from 3rd to 2nd place!” It is of course after the fact that at most they will only move 2 points closer towards bridging the gap to the 1st place server. kitten de kitten *

In other words they think you are stupid. And by capitalizing on ur stupidity they can make the game seem more fair. Given fake news is all the rage these days, this is basically a low effort form of fake news where you present an old concept as something ‘new and brilliant’, when infact it is the same old concept with inflated numbers.

I’m even less impressed by this move now. Link a response that makes me MORE impressed.

Just because you’re unable to compare numbers properly doesn’t mean Anet also fails to do so…

Justine.6351:

Jim Hunter.6821:

Justine.6351:

I don’t understand,

If it was the same wouldn’t it be 6-4-2?

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

Think about the point spread between the winner and loser after a couple skirmishes.
The winner has 10 the loser has 6, that’s a 4 point difference. With 3-2-1 the winner would have 6 and the loser would have 2, still a 4 point difference. It’s exactly the same just with bigger numbers, that’s why this “change” is completely pointless.

That’s a horrible fail at math.

On 5 matches,

6-4-2, 30-20-10, 20 point difference

5-4-3, 25-20-15, 10 point difference

I was comparing 5-4-3 to 3-2-1

Straegen.2938:

How about they change it, we live with it for a couple matches and see how it is BEFORE we start criticizing it and asking for it to be removed. You may be right and it doesn’t work as well but we probably won’t know all the ramifications both good and bad until they try it.

Hah! If only.

I never thought to take note of how long the last change got before people were up in arms about it being the new ruiner of everything. This probably counts as some kind of depressing karmic balance.

(I’m agreeing with you by the way, all for just seeing how it goes, in case that isn’t clear.)

Jeknar.6184:

Jim Hunter.6821:

Justine.6351:

I don’t understand,

If it was the same wouldn’t it be 6-4-2?

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

Think about the point spread between the winner and loser after a couple skirmishes.
The winner has 10 the loser has 6, that’s a 4 point difference. With 3-2-1 the winner would have 6 and the loser would have 2, still a 4 point difference. It’s exactly the same just with bigger numbers, that’s why this “change” is completely pointless.

Indeed, the end result for scoring is the same no matter if it’s 3-2-1, 5-4-3 or 1000-999-998. The difference is in how glicko calculate the ratings based on scores… It use the % of the score between the the highest ranked and the rest to make the changes. With higher numbers you make the % much closer which will lessen the rating gain/loss.

For exemple: If a match had 100 skirmishes and ended at 300-200-100, the score % difference between the worlds is higher than if it was 500-400-300 (100 is 33% of 300 while 300 is 60% of 500). The higher the difference, more rating you gain/loss based on the rating difference between worlds.

All the yadda-yadda about how glicko works aside, I honestly rather take 3-2-1 over a similar system with higher numbers because the huge glicko swings actually work for varieirty. Staleness is terrible, and we already had years of stale matchups in the past.

I honestly don’t even attribute scores to how glicko affects matches anymore. Between The linkings preventing the smaller paired servers from achieving anything at all and anet manually adjusting glicko scores, it’s pretty much a non issue for me.

I like the 2-1-1 system because I have actually seen it improve peoples play style. It might not be the case on every server but in a number of matches across several tiers in NA that I have played on I saw the PPT players focusing the stronger server instead of just k-training the weaker server for easy points. When I was on the dominant server in a match up we were constantly being hit. There were several matches where the winner was determined on Friday, which kept people playing all week.

I personally never even look at the score while I’m playing but apparently there are still a lot of players that care about it. The 2-1-1 system kept them playing longer and made the matches more interesting, it deserves a decent chance before it is tossed out for the same old system.

Hesacon.8735:

You misspelled geniuses, so there’s that.

Made me literally lol.

Funny – most of the matchups of the 2-1-1 weeks the only thing I saw that had yet to be determined by Friday was who came last.

Fay.2357:

That being said, you’re operating under the assumption that large and rapid swings in glicko rating is a desired outcome of the scoring system. This is not an appropriate assumption to make. In fact, I would argue that you’re going in the entirely wrong direction with that assumption. The glicko system itself contains a variety of levers that can be adjusted to change how it reacts to inputs. Your complaint, that the glicko rating won’t see large enough changes with 543, is potentially valid. However, going back to 321 isn’t the answer to that complaint. There’s another lever that should be used to change that: volatility.

The thing about glicko is that it’s based on the assumption that the scores are accurately representative of the strength of what they measure. If one score is twice another score, glicko assumes that whomever the first score belongs to is twice as strong as whomever the second score belongs to. Therefore, the scoring system should strive to most accurately represent that relative power levels of the servers; it shouldn’t be the tool you use to try and increase the volatility of the matches. 211 was obviously flawed in this regard, and potentially so were 543 and 321. However, argumemts for one or another scoring system should be made on the proper basis of whether they accurately represent server strength.

I am coming at it from the angle of: is this matchup fun? As someone on SoS who had the exact same matchup of maguuma and SBI for months on end, got steamrolled by a glicko locked yak’s bend when they had a link and a clip on their shoulder, and we won’t talk about a JQ matchup; I want fun matchups.

You have to remember: matchups last an entire week. It’s not like spvp where you get a team that doesn’t know what home is and can’t rotate, but you know 10 minutes later is a new match. If you’re going to be steamrolled, you’re going to be steamrolled an entire week.

I don’t want the same matchup every single week. You learn a server’s tendencies and want variety. What’s worse than the same matchup every week is the same bad matchup every week.

While there are other levers to pull with glicko or ditching glicko altogether, the 2-1-1 scoring works today for resolving matchup problems. No fancy new system, no glicko replacement, no Anet babysitting.

With matchmaking working, Anet would then be free to pursue other changes to improve wvw. Instead some resources are going into an experiment with a predictable result.

Ben K.6238:

Funny – most of the matchups of the 2-1-1 weeks the only thing I saw that had yet to be determined by Friday was who came last.

Yeah, but we have yet to have a system that does that. At least with 2-1-1 the fight is fun until the end.

Fay.2357:

Cerby.1069:

McKenna Berdrow.2759:

Since this is a similar to the other 5-4-3 thread, here is a link to my response on that thread.

Read ur response. The first response to it is pretty much my sentiments exactly, I’ll add some more detail though:

Tl:DR: anet thinks by inflating the scores and tossing bigger numbers around it will make the system confusing and complicated, to the extent that the average player will THINK the scores are closer and that their skirmishes matter more than they actually do.
*For example: you will have occasions where a score ‘like’ 44-46-40 where the numbers are close may exist (idk if those specific values are possible but they just have to be close). Players see the winner as getting 5 points so they say “hmmm if our server wins we get to 45, we go from 3rd to 2nd place!” It is of course after the fact that at most they will only move 2 points closer towards bridging the gap to the 1st place server. kitten de kitten *

In other words they think you are stupid. And by capitalizing on ur stupidity they can make the game seem more fair. Given fake news is all the rage these days, this is basically a low effort form of fake news where you present an old concept as something ‘new and brilliant’, when infact it is the same old concept with inflated numbers.

I’m even less impressed by this move now. Link a response that makes me MORE impressed.

Just because you’re unable to compare numbers properly doesn’t mean Anet also fails to do so…

Sowwy I’m but a mere muggle. I can’t see the invisible numbers, since I’m not a wizard like you!

THere is no debate on what I posted in that section you quoted. Its basic math, look at my previous post before the one you quoted for a simplistic indepth analysis on why, although you should already have read that and understood the basic maths of it.
If you have questions I’ll help answer, but ur talking nonsense at this point. Whats ur argument? Did saying ‘fake news’ upset you?

You couldn’t disprove Hesacon’s post and mine is the exact same.

Ben K.6238:

Funny – most of the matchups of the 2-1-1 weeks the only thing I saw that had yet to be determined by Friday was who came last.

Funny – that was usually determined by the end of the weekend in the old system (which has returned).

Anyways blah blah blah argue over numbers you all want, it’s returning to the old system. 321 543 975 it doesn’t matter, players will play the same under those conditions go for the easiest points, so I hope the third place teams have stocked up on tampons for their behinds.

Fay.2357:

Justine.6351:

Hesacon.8735:

Justine.6351:

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

Think about it carefully. The absolute difference between each place is exactly the same if it is 5-4-3 or 3-2-1. It is exactly as difficult to overcome a deficit in 3-2-1 and 5-4-3.

I’m not so sure that is true.

1-2-3 is,
First place gets 150% value of second
Second gets 100% value of third
Third gets 33% value of first

3-4-5 is,
First place gets 125% value of second
Second gets 133% value of third
Third gets 66% value of first

This is accurate. All the geniouses in this thread going on about the absolute difference between the scores have clearly forgotten that there’s more than one way to compare numbers. I suppose they think there’s no difference between a 50% pay raise and a $1 pay raise either?

Yea, posters saying that it’s the same are the ones that need to l2math – sorry, but ANet has its numbers correct here. A win which determines a server’s tier is based solely on the fact the server won at this point given manual adjustments, and is often the basis for transfers which are what really define how volatile a server is; not by how much or how little a server necessarily wins by, unless inspected by ANet, not to mention Glicko rates quality by difference, anyways, which is tightened up with the new numbers. To enable better matchmaking and closer PPT games, the 5/4/3 relationship does a strictly better job at tightening scores. To break even on a 3-2-1 distribution, the difference between the efficacy of skirmish wins breaks down much more in favor of the server that wins the skirmish since the relationship of a win from first to third is 300% whereas the 5/4/3 breakdown is only 167%, cutting the value per win down by almost half.

It helps prevent runaway matches much better while also strictly enforcing the same exact scoring differences if every server performs exactly the same as before and always performs at a first/second/third place allocation for every single skirmish.

The only valid criticisms are that it may make matches too stale and that fights still may not be even, although manual re-adjustments enforce this doesn’t happen for the former, and the latter will never change so long as people are so mobile in their transferring habits, anyways.

Absolute and relative.
Yes. 3-2-1 scoring is different from 5-4-3 even though its still 1 point between each.

Justine.6351:

I don’t understand,

If it was the same wouldn’t it be 6-4-2?

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

No, 5/4/3 is like 3/2/1 with a free +2 each round. in the end all these +2 only inflate the scores making the difference seem smaller because the numbers are larger. Just take 168 from the end score of each week and you will have the 3/2/1 score.

nsleep.7839:

Justine.6351:

I don’t understand,

If it was the same wouldn’t it be 6-4-2?

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

No, 5/4/3 is like 3/2/1 with a free +2 each round. in the end all these +2 only inflate the scores making the difference seem smaller because the numbers are larger.

Lets do some math!

3 is 3x larger than 1.

This means that the winner gains a score 300% of the loser in 3-2-1.

5 is 1.66x larger than 3.

This means that the winner gains a score 166% of the loser in 5-4-3.

This matters. The glicko algorithm works based off of relative comparisons, not absolute value comparisons. The 5-4-3 system has the same absolute value differences as the 3-2-1 system, but the relative differences are massively different. This has the effect of making the final evaluation by glicko much tighter than in the 3-2-1 system.

It’s worth noting that the OP of this thread has stated the argument that they feel the larger swings of the 3-2-1 system are healthier for WvW than the tighter swings of the 5-4-3 system. I personally feel that this is faulty due to trying to use the scoring system to control that instead of the volatility, but the OP is at least arguing from a sound logical basis. All of this garbage about ‘oh 5-4-3 is the same as 3-2-1’ is just a good case of people completely failing to understand relative comparisons.

Fay.2357:

nsleep.7839:

Justine.6351:

I don’t understand,

If it was the same wouldn’t it be 6-4-2?

5-4-3 seems to keep the winner and the loser of a skirmish in competition?

No, 5/4/3 is like 3/2/1 with a free +2 each round. in the end all these +2 only inflate the scores making the difference seem smaller because the numbers are larger.

Lets do some math!

3 is 3x larger than 1.

This means that the winner gains a score 300% of the loser in 3-2-1.

5 is 1.66x larger than 3.

This means that the winner gains a score 166% of the loser in 5-4-3.

This matters. The glicko algorithm works based off of relative comparisons, not absolute value comparisons. The 5-4-3 system has the same absolute value differences as the 3-2-1 system, but the relative differences are massively different. This has the effect of making the final evaluation by glicko much tighter than in the 3-2-1 system.

It’s worth noting that the OP of this thread has stated the argument that they feel the larger swings of the 3-2-1 system are healthier for WvW than the tighter swings of the 5-4-3 system. I personally feel that this is faulty due to trying to use the scoring system to control that instead of the volatility, but the OP is at least arguing from a sound logical basis. All of this garbage about ‘oh 5-4-3 is the same as 3-2-1’ is just a good case of people completely failing to understand relative comparisons.

Oh, taking glicko in consideration? Let’s see…

1 – Because the slower moving glicko making matchups more stale is exactly what we want.
2 – I don’t even know why they use a rating created to rate 1v1 matches where there won’t be two stronger sides feeding off of the weaker one.
3 – In the short terms matches, not considering the glicko, it’s basically the same as 3/2/1 to determine the winner.

Some questions the scoring can answer:

• Which of 2-1-1, 3-2-1 or 5-4-3 represents the teams’ relative strengths the best?
  • This is what Glicko wants out of the scores.

5-4-3.

Some matchups are subjectively very unbalanced when you are on the battlefield. You can look at the war scores for a more objective view.

After the population rebalancing even closer victory points distribution could be justified, like 9-8-7.

On the other hand the “how much a team won” is not that important. Players have even asked for 1-up-1-down outcomes for matchups.

• Which of the 2-1-1, 3-2-1 or 5-4-3 encourages 2v1 against the strongest team?
  • This creates more interesting matchups.

2-1-1.

A skirmish victory reward would be a nice extra incentive – a bragging right or cosmetic reward would probably do. But no rewards before population rebalancing!

• Which of the 2-1-1, 3-2-1 or 5-4-3 prolong the winner decision the longest into the matchup?
  • Decision point is preferably towards the end of the matchup.

3-2-1 and 5-4-3. Those can decide the matchup winner by 2/3 into the matchup the earliest.

The upcoming Last Stand feature can prolong the decision even more than this 1/6th difference.

Currently there is no matchup victory reward for winning a matchup. But that is coming after the population rebalancing.

If there is lot of 2v1 against the strongest team the matchups are not so predictable.

• Which of the 2-1-1, 3-2-1 or 5-4-3 is the simplest to understand?
  • Players want to know the matchup status at glance.

None. I would argue 1-0-0 would be the simplest to understand. Then calculate Glicko rating adjustment whichever way you want (multiply or add victory points or derive additional points from war scores before adjusting the ratings).

Most players (including me) have no clue how the current score affects Glicko ratings. Unless they visit an external web site. Or unless they are on a guest server, then there is no change.

The WvW panel UI needs projected effect on the servers’ Glicko ratings. Maybe with a +/- comparison to the break even point. Throw in % chance to move up or down a tier in the next matchup.

• Which of the 2-1-1, 3-2-1 or 5-4-3 allow the weakest team to recover to win a matchup the easiest?
  • If there was a matchup victory reward no team should just give up trying.

Until there is a matchup victory reward it doesn’t matter which team has the most victory points – Glicko ratings aside.

2-1-1 keeps the weakest team in the race the longest. Firstly by keeping the distance to the strongest team at 1 point per skirmish. And secondly by encouraging 2v1 against the strongest team.

2-(-1)-(-1) where the losing teams have a victory point removed would keep the weakest team in the competition even longer. But then again the meta would become: ignore the first days of the matchup and catch up later.

PS. Did I mention population rebalancing too many times? :P

Korgov.7645:

And secondly by encouraging 2v1 against the strongest team.

You say this like its some sort of pro.

It only work if a server truly hate another server so much they are fighting to loose the matchup. No I mean literally, for 2v1 to function in a 2/1/1 score world, the weakest server has to fight tooth and nail to come last while making sure the dominating server comes 2nd, that aint easy.

Especially not since your 2v1 “ally” will at the same time pound your weak kitten mercilessly after they got tired wiping against the strongest server.

Dawdler.8521:

Korgov.7645:

And secondly by encouraging 2v1 against the strongest team.

You say this like its some sort of pro.

It only work if a server truly hate another server so much they are fighting to loose the matchup. No I mean literally, for 2v1 to function in a 2/1/1 score world, the weakest server has to fight tooth and nail to come last while making sure the dominating server comes 2nd, that aint easy.

Especially not since your 2v1 “ally” will at the same time pound your weak kitten mercilessly after they got tired wiping against the strongest server.

I think it is for the better to give the weakest team a kingmakers role instead of silver medalist makers role. The weakest team also has an option to build up momentum for the next skirmish by leaving the current skirmish for the other teams to fight over. Or fight to lose as you put it.

In 3-2-1 and 5-4-3 systems the weakest team only has one role: PPT piñata.

Recovering to win is not easy under any system. As it should not be. But it is easier to win X skirmishes than to win X skirmishes and ensuring the dominating team also comes 3rd X times.

Typically speaking, especially with links, most PPT gain is based on coverage on off-hours. This is also why EU has closer matchups than NA.

5/4/3 ends up allowing for a server to have a much better chance of staying in the fight versus 3/2/1 and 2/1/1 because of its proportions favoring a lesser-magnitude loss when losses occur.

Again, there are two totally different arguments that are being made here, and both are valid to some extent:

Assuming no manual glicko re-adjustments, the 5/4/3 spread does stagnate matches more because the magnitude of losses is greater. This is still a problem with Glicko and why ANet needs faster re-adjustment windows, because things like stacking and transfers mess the algorithm up to near-uselessness anyways.

However, the advantage here is that the matchups will be closer in terms of PPT because the figures are relatively closer, thus from a scoring and PPT perspective, 5/4/3 is an objective improvement.

5-4-3 will result in servers fighting for second place again and boosting night capping.

The irony being that 2-1-1 actually makes nightcapping contribute more to PPT than 5-4-3 or 3-2-1.

I’m really glad I’ll be dead before you guys get a chance to run things

Same dumb people making the same dumb arguments. Scores changes peoples behavior. People don’t play exactly the same under different scoring systems, so taking the scores on a match and converting them to whatever scoring system you want to compare is dumb. The 543 closer reflects peoples actual efforts when comparing war score to skirmish score. People will also play harder as the matches will by slightly closer than 321 and wont be a blow out by Monday as in the 211.

But this has all been disused in the previous thread. Pretty much everyone arguing here are the same people arguing there. Logic prevailed. If you’re against it, why don’t you give it a chance, as you clearly can’t see what the change will mean. We’ve already lost a lot of participation under the 211 system. Hopefully, we’ll get them back now.

Isn’t it the case that the team which would be first with 3-2-1 will still be first with 5-4-3, the team that would be second will still be second, and the team that would be third will still be third? Which means that the only difference is the glicko calculations…..

Adamarc.7463:

Isn’t it the case that the team which would be first with 3-2-1 will still be first with 5-4-3, the team that would be second will still be second, and the team that would be third will still be third? Which means that the only difference is the glicko calculations…..

Yes.

But don’t tell BG players that, they’ll label you as an alien scientist cause you don’t know enough maths to be a real scientist.

Xenesis.6389:

Adamarc.7463:

Isn’t it the case that the team which would be first with 3-2-1 will still be first with 5-4-3, the team that would be second will still be second, and the team that would be third will still be third? Which means that the only difference is the glicko calculations…..

Yes.

But don’t tell BG players that, they’ll label you as an alien scientist cause you don’t know enough maths to be a real scientist.

Well, I’m graduated in Math and I can shove my kittening Degree unto them if they say somenthing like that to me.

DeceiverX.8361:

The irony being that 2-1-1 actually makes nightcapping contribute more to PPT than 5-4-3 or 3-2-1.

You mean contribute more to the Glicko rating adjustments, not PPT.

If you compare the nightcapper team and the second team, yes, 2:1 ratio is greater than 5:4 or 3:2. If you compare the 1st team vs the 3rd team, then the ratios are 5:3, 2:1 and 3:1. In that sense the night skirmishes (and any other skirmishes) are more valuable to win under 2-1-1 than under 5-4-3.

The problem with 2-1-1 was that it didn’t incentivize the 2nd and 3rd place server to focus the 1st place server. For one, you’d need both servers focusing the 1st place server during the same scrimmage and how often is that realistically gonna happen? Secondly, whenever that does happen, only one of those two servers will be the beneficiary of it. The other gets nothing unless they are somehow able to keep it up for another 2 hour scrim and trade the win. But you’d need some serious coordination/collusion across both servers and we know most players don’t care enough to go down that road.

Additionally, most players don’t decide which server to attack based on the score. They make their decision based on which server they perceive to be stronger/weaker at the time or which one would provide better fights.

So what we had with 2-1-1 was a pretty ridiculous system where it was pointless for the two weaker servers to even play. Now with 5-4-3, if you can’t get 1st place during a given scrim, there’s at least a reason to still play and get 2nd.