Ranked Draws & Rank-Based Scoring: Fibonacci-Diplo

This is the home for suggestions for site improvements, changes to house rules, and new variants.
Forum rules
It's okay to suggest new rules variants in this forum, but proposing new *maps* should be done in the linked "New Map Variant Proposals & Voting" subforum.

Re: Ranked Draws & Rank-Based Scoring: Fibonacci-Diplo

Postby nanooktheeskimo » 06 Jun 2018, 08:45

Eliminated players could in theory get points—it depends on the Elo adjustment after the initial rating. You’ll see a lot less players eliminated though, because there really isn’t as much incentive to go for a two way in this system. You don’t get more points for tying for top at 17 centers apiece than you do at 12 centers apiece...and without the incentive to kill smaller powers for more points, you’re more likely to see big powers trying to work with smaller powers to push themselves closer to soloing rather than worrying about a 2 way turning into a 5 way...since a 5 way scores the same for the board topper as a 2 way does.

If this were implemented, it would probably work best as its own separate rating system rather than folded into our current standard one—so it would be its own drop down option, same as “1900” or “AMed” is now.


NoPun can give his own answer of course, but that’s my understanding, and that’s the effect that carnage coding has in FtF games—usually there’s only one to two players eliminated, a small power or two, a midsize power or two, and two or three larger powers. There’s just not as much incentive to eliminate people when the size of the draw is irrelevant, only your placement in it.
Platinum Classicist
(h/t lordelindel)

I am your (co-) Leader.

GM of WitA 7, WitA 8.

Come play face to face!

Need a forum game GM'ed? PM me!

Mod (but I'm normally not talking as one)
User avatar
nanooktheeskimo
Site Admin
Site Admin
 
Posts: 9728
Joined: 19 Jun 2014, 19:52
Location: East TN
Class: Star Ambassador
Standard rating: 1209
All-game rating: 1389
Timezone: GMT-5

Re: Ranked Draws & Rank-Based Scoring: Fibonacci-Diplo

Postby NoPunIn10Did » 06 Jun 2018, 13:08

nanooktheeskimo wrote:If this were implemented, it would probably work best as its own separate rating system rather than folded into our current standard one—so it would be its own drop down option, same as “1900” or “AMed” is now.

I'm not sure it actually needs a separate system unless the players who invest such emotion into their "standard" rating put up a massive fuss.
Lead Volunteer Developer & Forum Admin

Variant GM & Designer
User avatar
NoPunIn10Did
Premium Member
 
Posts: 2445
Joined: 17 Aug 2011, 00:17
Location: North Carolina
Class: Ambassador
Standard rating: 1000
All-game rating: 1471
Timezone: GMT-5

Re: Ranked Draws & Rank-Based Scoring: Fibonacci-Diplo

Postby V » 06 Jun 2018, 13:30

NoPunIn10Did wrote:
nanooktheeskimo wrote:If this were implemented, it would probably work best as its own separate rating system rather than folded into our current standard one—so it would be its own drop down option, same as “1900” or “AMed” is now.

I'm not sure it actually needs a separate system unless the players who invest such emotion into their "standard" rating put up a massive fuss.


I don’t think it should require a massive fuss from anyone.
The more I read of the likely outcomes coming from this change, it doesn’t sound like “standard” as currently played on PlayDip. Many would salute that change & merely point out it’s “the new standard” but for many others this tinkering with scoring systems to influence game strategy is not a fascination. I think Nanook is on the right track with how this change should be handled.
Platinum Classicist
Voilà! In view, a humble vaudevillian veteran cast vicariously as both victim and villain by the vicissitudes of Fate. This visage, no mere veneer of vanity, is a vestige of the vox populi, now vacant, vanished.
V
 
Posts: 622
Joined: 04 May 2014, 21:28
Class: Star Ambassador
Standard rating: 1718
All-game rating: 1754
Timezone: GMT-7

Re: Ranked Draws & Rank-Based Scoring: Fibonacci-Diplo

Postby boldblade » 06 Jun 2018, 15:23

I love this idea.
boldblade
 
Posts: 338
Joined: 05 Feb 2014, 17:33
Class: Star Ambassador
Standard rating: (1474)
All-game rating: (1488)
Timezone: GMT

Re: Ranked Draws & Rank-Based Scoring: Fibonacci-Diplo

Postby asudevil » 06 Jun 2018, 15:36

DQ wrote:
I'm trying to understand your point here - you're arguing against an alternative scoring structure option because you don't agree with the incentives that system provides, yes?


My point is that if you are trying to encourage solos...there are easier ways to do it...that doesn't require a totally different system.

IF that is the case, then ... could you explain what you think the harm of having an alternative scoring system would be, for the site and upon the ranking system?
more options =/= better. We are one of the biggest sites and still have only a few thousand active players. Continuing to fracture the scoring makes it less useful. Or changing HOW games are scored should then not affect the main ELO and rating since its a different game when you score rank based draws. People play differently so its not the same game.
Captain FANG, forum team championships WINNER
Part of the surviving nations of WW4/Haven

Unless I am in the cheater's subforum. 99% of what I say is NOT as a mod.
User avatar
asudevil
Premium Member
 
Posts: 16578
Joined: 18 Jul 2011, 02:20
Class: Star Ambassador
Standard rating: 1351
All-game rating: 1447
Timezone: GMT-7

Re: Ranked Draws & Rank-Based Scoring: Fibonacci-Diplo

Postby asudevil » 06 Jun 2018, 15:42

nanooktheeskimo wrote:
NoPun can give his own answer of course, but that’s my understanding, and that’s the effect that carnage coding has in FtF games—usually there’s only one to two players eliminated, a small power or two, a midsize power or two, and two or three larger powers. There’s just not as much incentive to eliminate people when the size of the draw is irrelevant, only your placement in it.


That's also because FtF has limited time so there by default are larger draws because they only play 7 years
Captain FANG, forum team championships WINNER
Part of the surviving nations of WW4/Haven

Unless I am in the cheater's subforum. 99% of what I say is NOT as a mod.
User avatar
asudevil
Premium Member
 
Posts: 16578
Joined: 18 Jul 2011, 02:20
Class: Star Ambassador
Standard rating: 1351
All-game rating: 1447
Timezone: GMT-7

Re: Ranked Draws & Rank-Based Scoring: Fibonacci-Diplo

Postby NoPunIn10Did » 06 Jun 2018, 15:48

asudevil wrote:
nanooktheeskimo wrote:
NoPun can give his own answer of course, but that’s my understanding, and that’s the effect that carnage coding has in FtF games—usually there’s only one to two players eliminated, a small power or two, a midsize power or two, and two or three larger powers. There’s just not as much incentive to eliminate people when the size of the draw is irrelevant, only your placement in it.


That's also because FtF has limited time so there by default are larger draws because they only play 7 years


The seven year thing is popular in European tournaments, but rank-based scoring isn’t solely used in finite-length games.
Lead Volunteer Developer & Forum Admin

Variant GM & Designer
User avatar
NoPunIn10Did
Premium Member
 
Posts: 2445
Joined: 17 Aug 2011, 00:17
Location: North Carolina
Class: Ambassador
Standard rating: 1000
All-game rating: 1471
Timezone: GMT-5

Re: Ranked Draws & Rank-Based Scoring: Fibonacci-Diplo

Postby NoPunIn10Did » 06 Jun 2018, 16:01

super_dipsy wrote:Unless I have misunderstood the suggestion, this system would work in conjunction with the current one. In other words, playing games with this option will affect your site rating too, but just differently.

This is correct.
super_dipsy wrote:The only way to avoid this would be to say that games played with this option are unranked in terms of our current structure, but I don't think that is what is being proposed.

Correct; it's not what is being proposed. I would like to see rank still displayed for unrated games that have this particular draw option turned on (as that would be a good teaching tool when playing mentor games on this system).

super_dipsy wrote:- We still use our elo-like current system

Correct.
super_dipsy wrote:- For games with this option, it only takes effect if there is a draw (is this right? or is the solo result modified too?)

Correct. A solo is unaffected. The winner wins, and everyone else loses.
super_dipsy wrote:- In a draw, the way points are awarded will be altered; instead of the sharing of pointss done today, you have a different scale based on rankings which would be applied to the rating algorithm to calculate the deltas

Correct.
super_dipsy wrote:- An important point which I am not clear on - am I right in thinking that with this system you could get points for being eliminated? Eg if I am the last to go in a 2-way draw, I get points because I was ranked third?

Yes, you can still get points for being eliminated. The best way to think of this is that, in a draw, rank is a total replacement for the concepts of winner (in the draw) and loser. If the game ends in a draw, every power is ranked, and players can draw their own conclusions about whether that rank was a win or loss.
super_dipsy wrote:- Can I also get points if I surrender (ie I was last to surrender in a 2-way draw)? If not, then do you have to adjust the scoring allocation based on the fact tht only 2 players are going to participate in the rank-based scoring?

To be clear, you mean "surrendered" as in, either voluntarily kicking yourself out of the game or NMR-ing, right? Are you saying that all five of the other players surrendered, or only just the one?
Surrendered players don't participate in the rank-based scoring, but their countries, should they not find replacements, are used as placeholders. I'll give an example further below to illustrate.
super_dipsy wrote:- If you surrender, you simply get what our current system would normally give you for a surrender

Correct. A surrendered player gets what our system would normally give you: a total loss.
super_dipsy wrote:- If you pick up a country, you get a pro rata of your eventual result based on turns played as we do today

Correct. The scoring system treats you as a normal player, whether you were eliminated or survived until the end, and after all the calculations are done, your Elo delta gets prorated (or waived, in the case of a negative delta with Rating Shield).

Practical Examples
These examples will involve the following players:
  • Adam
  • Betty
  • Charlie
  • Dmitri
  • Elsbeth
  • Froderick
  • Gregory
  • Hannah
  • Isolde
  • Winnie the Pooh
Assumptions
  1. Assume for simplicity's sake that each of them starts each example with the exact same Elo rating.
  2. Assume also that these are longtime players, so there are no K-factor bonuses that multiply their Elo deltas.
  3. Assume that within a single game (of either rating system), there are 420 points available. Each player's "expected" value is thus 60 in-game points. This is approx. 14.29% of the total score available, or 1/7.
  4. Assume that a loss in a standard game, with the usual draw-sized scoring, a loss would represent 0 points within the game and an Elo delta of -14.29 points, following the formula here:

    100*(Actual - Expected)/Total = 100*(0-60)/420 = -14.29

    (Again, to be clear, I don't know the actual numbers of the Elo formula, but when I've tried to reverse-engineer it statistically, 100/7 seems to be approximately the average points lost.)
Example A: No ties, no surrenders
  • Adam: France
  • Betty: Russia
  • Charlie: Italy
  • Dmitri: Germany
  • Elsbeth: Austria
  • Froderick: Turkey
  • Gregory: England
    Fall 1902: Austria / Elsbeth is eliminated.
    Fall 1904: Turkey / Froderick is eliminated.
    Fall 1906: Gregory / England is eliminated.
    Fall 1908: Betty / Russia is eliminated.
    Winter 1908: Adam, Dmitri, and Charlie terminate the game in a draw.
    Adam / France has 10 centers.
    Dmitri / Germany has 7 centers.
    Charlie / Italy has 17 centers.
Under the Current Draw-Sized System
In-game points (420 basis)
  • Adam: 140
  • Betty: 0
  • Charlie: 140
  • Dmitri: 140
  • Elsbeth: 0
  • Froderick: 0
  • Gregory: 0
Elo deltas
  • Adam: 100*(140-60)/420 = +19.05
  • Betty: 100*(0-60)/420 = -14.29
  • Charlie: 100*(140-60)/420 = +19.05
  • Dmitri: 100*(140-60)/420 = +19.05
  • Elsbeth: 100*(0-60)/420 = -14.29
  • Froderick: 100*(0-60)/420 = -14.29
  • Gregory: 100*(0-60)/420 = -14.29
Under Fibonacci-Diplo
Rank
  • Adam: 2nd
  • Betty: 4th
  • Charlie: 1st
  • Dmitri: 3rd
  • Elsbeth: 7th
  • Froderick: 6th
  • Gregory: 5th
In-game points (420 basis)
  • Adam: 105
  • Betty: 42
  • Charlie: 168
  • Dmitri: 63
  • Elsbeth: 0
  • Froderick: 21
  • Gregory: 21
Elo deltas
  • Adam: 100*(105-60)/420 = +10.71
  • Betty: 100*(42-60)/420 = -4.29
  • Charlie: 100*(168-60)/420 = +25.71
  • Dmitri: 100*(63-60)/420 = +0.71
  • Elsbeth: 100*(0-60)/420 = -14.29
  • Froderick: 100*(21-60)/420 = -9.29
  • Gregory: 100*(21-60)/420 = -9.29

Example B: One surrender and one tie
  • Adam: France
  • Betty: Russia
  • Charlie: Italy
  • Dmitri: Germany
  • Elsbeth: Austria
  • Froderick: Turkey
  • Gregory: England
    Fall 1902: Austria / Elsbeth is eliminated.
    Spring 1904: Turkey / Froderick surrenders from NMR-ing too much. Turkey goes into CD.
    Fall 1904: Turkey is eliminated before a replacement can be found.
    Fall 1906: Gregory / England is eliminated.
    Fall 1908: Betty / Russia is eliminated.
    Winter 1908: Adam, Dmitri, and Charlie terminate the game in a draw.
    Adam / France has 9 centers.
    Dmitri / Germany has 9 centers.
    Charlie / Italy has 16 centers.
Under the Current Draw-Sized System
Should be exactly the same as Example A.
Under Fibonacci-Diplo
Rank
  • Adam: Tied 2nd/3rd
  • Betty: 4th
  • Charlie: 1st
  • Dmitri: Tied 2nd/3rd
  • Elsbeth: 6th
  • Gregory: 5th
  • Froderick: Surrendered
  • Turkey (CD): 7th
Froderick, the surrendered player, gets a total loss. Note that because Turkey never got a replacement, it is automatically shifted to the back of the pack at end-game. This means that Elsbeth, who was actually the first eliminated, gets bumped up to 6th place. No one will receive Turkey's Elo delta at game-end.

In-game points (420 basis)
  • Adam: 84
  • Betty: 42
  • Charlie: 168
  • Dmitri: 84
  • Elsbeth: 21
  • Froderick: 0 (surrendered)
  • Gregory: 21
  • Turkey (CD): 0
Elo deltas
  • Adam: 100*(84-60)/420 = +5.71
  • Betty: 100*(42-60)/420 = -4.29
  • Charlie: 100*(168-60)/420 = +25.71
  • Dmitri: 100*(84-60)/420 = +5.71
  • Elsbeth: 100*(21-60)/420 = -9.29
  • Froderick: 100*(0-60)/420 = -14.29 (surrendered)
  • Gregory: 100*(21-60)/420 = -9.29

Example C: Replacements, Surrenders, and Bears, Oh My!
  • Adam: France
  • Betty: Russia
  • Charlie: Italy
  • Dmitri: Germany
  • Elsbeth: Austria
  • Froderick: Turkey
  • Gregory: England
  • Hannah (joins as a replacement)
  • Isolde (joins as a replacement)
  • Winnie the Pooh (joins as a replacement)
    Fall 1902: Austria / Elsbeth is eliminated.
    Spring 1903: Froderick surrenders. Turkey goes into CD.
    Fall 1903: Turkey is eliminated before a replacement can be found.
    Spring 1904: Dmitri surrenders. Germany goes into CD.
    Fall 1904: Hannah takes over Germany.
    Spring 1905: Gregory surrenders. England goes into CD.
    Fall 1905: Winnie the Pooh takes over England.
    Fall 1906: Winnie the Pooh / England is eliminated.
    Spring 1907: Charlie / Italy surrenders.
    Fall 1907[b]: Isolde takes over Italy.
    [b]Spring 1908
    : Betty surrenders. Russia goes into CD.
    Winter 1908: Adam, Hannah, and Isolde terminate the game in a draw.
    Adam / France has 10 centers.
    Hannah / Germany has 3 centers.
    Isolde / Italy has 11 centers.
    Russia (CD) has 10 centers.
    Winnie the Pooh played 15% of England's turns (prior to being eliminated).
    Hannah played 50% of Germany's turns.
    Isolde played 10% of Italy's turns.
Under the Current Draw-Sized System
In-game points (420 basis)
  • Adam: 140
  • Betty: 0 (surrendered)
  • Charlie: 0 (surrendered)
  • Dmitri: 0 (surrendered)
  • Elsbeth: 0
  • Froderick: 0 (surrendered)
  • Gregory: 0 (surrendered)
  • Hannah: 140
  • Isolde: 140
  • Winnie the Pooh: 0
Elo deltas
  • Adam: 100*(140-60)/420 = +19.05
  • Betty: 100*(0-60)/420 = -14.29 (surrendered)
  • Charlie: 100*(0-60)/420 = -14.29 (surrendered)
  • Dmitri: 100*(0-60)/420 = -14.29 (surrendered)
  • Elsbeth: 100*(0-60)/420 = -14.29
  • Froderick: 100*(0-60)/420 = -14.29 (surrendered)
  • Gregory: 100*(0-60)/420 = -14.29 (surrendered)
  • Hannah: 100*(140-60)/420 = +19.05
      Because Hannah was a replacement, Hannah's Elo delta is prorated.
      Her final Elo delta is 19.05 * 50% = +9.52
  • Isolde: 100*(140-60)/420 = +19.05
      Because Isolde was a replacement, Isolde's Elo delta is prorated.
      Her final Elo delta is 19.05 * 10% = +1.90
  • Winnie the Pooh: 100*(0-60)/420 = -14.29
      Because Winnie the Pooh was a replacement, Winnie's Elo delta is prorated.
      His final Elo delta is -14.29 * 15% = -2.14
      If he had a Rating Shield, his final Elo delta would be 0.
Under Fibonacci-Diplo
Rank
  • Adam / France: 2nd
  • Betty: Surrendered
  • Charlie: Surrendered
  • Dmitri: Surrendered
  • Elsbeth / Austria: 5th
  • Froderick: Surrendered
  • Gregory: Surrendered
  • Hannah / Germany: 3rd
  • Isolde / Italy: 1st
  • Winnie the Pooh / England: 4th
  • Turkey (CD): Tied 6th/7th
  • Russia (CD): Tied 6th/7th
In-game points (420 basis)
  • Adam: 105
  • Betty: 0 (surrendered)
  • Charlie: 0 (surrendered)
  • Dmitri: 0 (surrendered)
  • Elsbeth: 21
  • Froderick: 0 (surrendered)
  • Gregory: 0 (surrendered)
  • Hannah: 63
  • Isolde: 168
  • Winnie the Pooh: 42
  • Turkey (CD): 10.5 (awarded to nobody)
  • Russia (CD): 10.5 (awarded to nobody)
Elo deltas
  • Adam: 100*(105-60)/420 = +10.71
  • Betty: 100*(0-60)/420 = -14.29 (surrendered)
  • Charlie: 100*(0-60)/420 = -14.29 (surrendered)
  • Dmitri: 100*(0-60)/420 = -14.29 (surrendered)
  • Elsbeth: 100*(21-60)/420 = -9.29
  • Froderick: 100*(0-60)/420 = -14.29 (surrendered)
  • Gregory: 100*(0-60)/420 = -14.29 (surrendered)
  • Hannah: 100*(63-60)/420 = +0.71
      Because Hannah was a replacement, Hannah's Elo delta is prorated.
      Her final Elo delta is 0.71 * 50% = +0.35
  • Isolde: 100*(168-60)/420 = +25.71
      Because Isolde was a replacement, Isolde's Elo delta is prorated.
      Her final Elo delta is 25.71 * 10% = +2.57
  • Winnie the Pooh: 100*(42-60)/420 = -4.29
      Because Winnie the Pooh was a replacement, Winnie's Elo delta is prorated.
      His final Elo delta is -4.29 * 15% = -0.64
      If he had a Rating Shield, his final Elo delta would be 0.
Lead Volunteer Developer & Forum Admin

Variant GM & Designer
User avatar
NoPunIn10Did
Premium Member
 
Posts: 2445
Joined: 17 Aug 2011, 00:17
Location: North Carolina
Class: Ambassador
Standard rating: 1000
All-game rating: 1471
Timezone: GMT-5

Re: Ranked Draws & Rank-Based Scoring: Fibonacci-Diplo

Postby NoPunIn10Did » 06 Jun 2018, 16:09

asudevil wrote:My point is that if you are trying to encourage solos...there are easier ways to do it...that doesn't require a totally different system.

It's not solely about encouraging solos. It's primarily about encouraging less draw-whittling. Draw-sized scoring incentivizes attacking and eliminating the weakest opponents, while rank-based scoring encourages attacking opponents stronger than yourself.

more options =/= better. We are one of the biggest sites and still have only a few thousand active players. Continuing to fracture the scoring makes it less useful.

I think the hand-wringing over fracturing the community is frankly overblown. We certainly do have a lot of options to filter games by, but I wouldn't assume that every new option will split the community into players that will only play that sort of game.

Or changing HOW games are scored should then not affect the main ELO and rating since its a different game when you score rank based draws. People play differently so its not the same game.

I'm not totally opposed to rank-based scored games being left out of the "standard" rating, though I do think that exclusion is philosophically problematic (since rank-based or lead-based scoring is fairly "standard" in the Diplomacy community as a whole).
But it wouldn't need to be cordoned off from all of the various Elo ratings on the site. It would certainly be part of the "All Games" rating, for example.

The math on this was designed specifically so that it wouldn't have to be a parallel system.
Lead Volunteer Developer & Forum Admin

Variant GM & Designer
User avatar
NoPunIn10Did
Premium Member
 
Posts: 2445
Joined: 17 Aug 2011, 00:17
Location: North Carolina
Class: Ambassador
Standard rating: 1000
All-game rating: 1471
Timezone: GMT-5

Re: Ranked Draws & Rank-Based Scoring: Fibonacci-Diplo

Postby asudevil » 06 Jun 2018, 16:44

NoPunIn10Did wrote:
asudevil wrote:My point is that if you are trying to encourage solos...there are easier ways to do it...that doesn't require a totally different system.

It's not solely about encouraging solos. It's primarily about encouraging less draw-whittling. Draw-sized scoring incentivizes attacking and eliminating the weakest opponents, while rank-based scoring encourages attacking opponents stronger than yourself.


Not really...it encourages you to gain SCs...wherever that is. So taking out the last 3 SC of a low power if that puts you over the guys above you...is as good as taking 2 off of one of them (cause they lose while you gain)...but while the swing could be better by attacking higher people...its still likely easier to take out the lower guy.

I'm not totally opposed to rank-based scored games being left out of the "standard" rating, though I do think that exclusion is philosophically problematic (since rank-based or lead-based scoring is fairly "standard" in the Diplomacy community as a whole).
But it wouldn't need to be cordoned off from all of the various Elo ratings on the site. It would certainly be part of the "All Games" rating, for example.

The math on this was designed specifically so that it wouldn't have to be a parallel system.


I see it as a totally different style of game. But that may just be me.
Captain FANG, forum team championships WINNER
Part of the surviving nations of WW4/Haven

Unless I am in the cheater's subforum. 99% of what I say is NOT as a mod.
User avatar
asudevil
Premium Member
 
Posts: 16578
Joined: 18 Jul 2011, 02:20
Class: Star Ambassador
Standard rating: 1351
All-game rating: 1447
Timezone: GMT-7

PreviousNext

Return to Suggestions

Who is online

Users browsing this forum: No registered users and 2 guests