Announcement Contact! Season 3 Prep-Meeting 1st of November 2020 @ 1800 UTC.

Contact! Season 3 Planning!
__________________​

​

PRTA Presents: Preparation Meeting for Contact Season 3

​

We are excited to prepare the return of PRTA's Project Reality tournament Contact!. The aim of this tournament is to provide a competitive, friendly, and enjoyable tournament experience for everyone wanting to join.

The meeting is scheduled for the 1st of November 2020 @ 1800 UTC.
​

How will the tournament work?

To start off, there will be a meeting where PRTA management presents its vision for the tournament. This is explained more in detail further below, but here is a short version:

• 16 man platoons. 8 reserves. No limits on the number of participating platoons.
• Skill-based team balancing in each battle for more competitive fights.
• "Trueskill" estimation of platoon skill for a post-tournament scoreboard.
• Incremental increases in battle size. 16v16, 32v32, and 48v48.
• Normal Project Reality Maps, no custom maps.
• No formal battle cycle, participate as much or little as you want to.
• The objective is to get as high as possible on the final scoreboard.

Taking into account the previous tournament problems with internal team drama and balance, we want to seek a more competitive and adaptive form for our tournament. This system should also give more players a chance at joining the tournament, without knowing English.

There will be no "preparation" battles, but teams must fight in enough matches to qualify for larger (32v32, 48v48) battles and to get a place on the final scoreboard. This scoreboard will be ranked by estimated platoon skill derived from the performance of each platoon during the tournament.

What is going to happen during the meeting?

During this meeting, we want to hear a large number of people's interests and concerns with the previous tournament as well as a productive result in the planning of a new one. We want to see if this is a good idea as opposed to the regular system of 2 teams with commanders on each team. During the meeting, we will take notes and gauge the interest of individuals who are willing to sign up, or help plan and execute it.

The meeting will be on TeamSpeak

​

Tournament Concept:

There is a ranking algorithm called TrueSkill which statistically estimates relative player rankings. This uses the team-level win/loss/draw result to update the relative skill estimates for participants in each match. Over a large number of matches, the estimated scores approximate the real relative skill levels of each participant.

We intend to apply a variant of this model to platoon-sized cohorts of players and use the skill estimates for both in-tournament teams balancing and for a public scoreboard at the end.

Platoons:

• The tournament will be played between many platoons of up to 24 players, fielding 16 for each match they participate in and having up to 8 in reserve.
• Players may not be in more than one platoon at a time, because any platoon can fight against any other.
• You are permitted to retire and recruit new players, max 2 in/2 out per match you participate in. This will involve an appropriate relaxation of the platoon's score by assuming that new players use the starting settings for ranking.

Challenges:

• For 1v1s you may only challenge platoons level with or above you on the scoreboard. Challenges must be recorded in the appropriate forum.
• For 2v2/3v3 matches all participants must agree - in the appropriate forum - to play each other.
• No rematches: no platoon may participate in a match with the same set of platoons as the previous match they were in. This is to encourage the mixing of platoons for a more accurate ranking.
  • bad: A,B; A,B (this is a rematch for A & B)
  • good: A,B; A,C; B,C; A,B
  • bad: A,B; A,C; A,B (this is a rematch for B)
  • good: A,B,C,D; A,B; C,D; A,B,C,D
  • bad: A,B,C,D; A,B; A,B,C,D (this is a rematch for C & D)
  • good: A,B,C,D; A,B,C,E; A,B,D,E; A,C,D,E; B,C,D,E; A,B,C,D

• If you do not respond to a challenge - or if you decline without good reason - then it will be recorded as a loss for your platoon but no change for the other. Good reasons include:
• Posting an LOA for your platoon at least 1 week in advance of the match (because sometimes IRL things happen)
• Already having a match arranged for your platoon (at any time, it does not need to collide with a proposed time; to avoid any platoons needing to plan multiple matches in advance)
• Failure to respond to a challenge/failure to turn up for a match in sufficient numbers/rage-quitting will result in a loss being recorded for those platoon(s) but no change for the others.

Matches:

• It is the responsibility of the platoons to arrange their own matches, there will be no set schedule, arranged breaks due to any holidays, or limits on the rate or number of matches any platoon may participate in.
• Matches will be between an even number of platoons on appropriately sized normal maps (no downloads required):
  • 16v16 (2 platoons): skirmish/1k map, platoons will swap factions half-way through.
  • 32v32 (4 platoons): 1k/2k map, medium-sized single-shot match.
  • 48v48 (6 platoons): 2k/4k map, full-scale single-shot match.
• Teams for each match will be decided using the tracked skill estimates to give the highest quality match (not chosen by the platoons), in order to prevent excessively unbalanced matches.
• If there are reserves active, both teams can agree to increase the number of players for the match, so long as there is an equal number of additional players on each team.
• Any player outside your team (not on a participating platoon roster) is not permitted to play in a match.
• A faction-swapping match is won by winning both rounds, drawn by winning one & losing one, and lost by losing both rounds.
• A single-shot match is won by having at least 10% of your starting tickets remaining. If no team reaches this threshold the match is a draw.

Qualification:

• A staged approach to platoon qualification will be used with 12 points available per completed match to be distributed evenly between all participating platoons. This is to give the algorithm some time to gather data before trying to balance larger matches.
  • A 1v1 gives 6 qualification points to each platoon.
  • A 2v2 gives 3 qualification points to each platoon.
  • A 3v3 gives 2 qualification points to each platoon.
• Platoons will initially only be allowed to play 1v1 matches.
• When a platoon reaches 18 qualification points they will be allowed to participate in 2v2 matches.
• When a platoon reaches 36 qualifying points they will be allowed to participate in 3v3 matches.

Scoring:

• Estimates of platoon skill will be maintained throughout the tournament.
• Platoons must reach 36 qualification points by the end of the tournament season to appear on the final scoreboard, otherwise, they will be dropped due to insufficient data.
• The scoreboard will show an estimate of the relative skill of each participating platoon, the higher you are the stronger your platoon.
• A prize/penalty for the most/least active platoons may be included at the end of the tournament season.

See you at the meeting!

​

 

Plz dont make shit faction matchups like last tournament. Whoever was in charge last time did not do a good job with that.

 

Read bruh

 

Boring

 

You're a little bitch

 

to the guy who was making rating concept https://arxiv.org/abs/1606.05289

 

Sloan

But afaik Sloan has an education from a better school than Cornell

 

concept leaves many questions, i think it is unstable and the system will break because

* 16 + 8 is a lot of people. Clans cant field so many, so they need to meet and handshake to just get numbers; all clans and solo gamers who don't have friends are left unattended
*Estimating rating from the groups where up to 50% can be substituted for other players might be not useful
* trueskill assigns pairs/teams with maximum draw probability, while in concept all teams have veto power. People are not stupid and see for themselves who is on what skill level, in the result all sophistication will boil down to matches by teams agreeing between themselves.
* somebody somewhere said that for 4v4 matches you need 48 games to approximate each participant's skill. This includes rematches that are forbidden in the concept.
* winner of the tournament is determined by qualification points which are not the mu-3sigma aka lowest estimate of the skill
* it will take a while to get out of 16vs16, and I'm not sure everyone will bc you need 3 wins to get out. And then 6 wins in 2vs2 which is at least 6 games, it looks just unrealistic with how many teams will be leaving 1v1
* 2teams vs 2teams with the pool of f.e. 6 teams is not a row of combinations, it is basically 1-2 combinations, and the bigger the player amount, the less games actually will be played before you face an unresolvable veto. IT is basically a skirmish tournament with bigger games as reward. Not everyone like skirmish but everyone likes big games which btw are freely accessible on public
* getting prior probabilities from playing 16vs16 looks like a waste of time before real games of balanced 50vs50. Can just get data from prstats.tk and transform it to prior somehow and balance according to it

well it looked very promising at first, but the concept with 2 teams picking other teams one by one like in children courtyard football matches looks simpler, fewer resources are wasted on sorting bc its outsourced to people and in general, it will arrive to the final goal - balanced 50vs50 - faster than what is written in the first post

I'v read about trueskill only yesterday, so if you have papers or something interesting to read about irl implementations please send links

 

Not sure if people would be happy being told who else will be in the match as well as what the team composition will be, but that can certainly be discussed in the meeting .

You only need about 16 to join in, you can have up to 8 additional "backup" players, but this could certainly be reduced and having a smaller upper limit would make it a bit more fair and give better skill estimates. Removing the ability to retire/recruit would also help convergence, but may have some practical issues.

We could ensure that you only get the team composition after committing to a match? (then backing out counts as a forfeit)
We could hide the skill estimates until the end of the tournament? (though this is unlikely to be a proper fix)

The estimate here will be made at the platoon-level, not the individual player level, so we have a mix of 1v1, 2v2 and 3v3 matches. For qualification there is much more weight on the matches with fewer platoons.

The winner will be based solely on μ-3σ. Qualification points will not appear on the final scoreboard, you just need to get a pre-set number of them in order to appear on the scoreboard at all.

You need to participate in 3 matches, but if you lose them all you will still qualify for 2v2 battles from that point. The qualification points are split evenly between all participants as a rough estimate of the amount of entropy gathered, not just to the winner(s).

We could arrange things so that no team-composition-related veto is possible (i.e. play or forfeit). The first 3 matches for every team will be 16v16 skirmish rounds, but the intent of that is to get some data to start balancing the larger rounds to try to avoid a repeat of some of the issues from the last tournament. Platoon size can certainly be discussed, though, we went for 16 players in order to avoid having too many platoons but still allow full-scale battles .

I've been playing around with data from our public server related to this, but it's heavily biased by people who will just leave if they are not winning vs people who will stick around on a losing team for longer than average, and people who will only play with their friends vs people who don't have any regular group they play with. I don't believe it forms a useful prior.

I'm using https://pypi.org/project/trueskill/ with some of my own extensions to handle some data issues we get in real games that the original package didn't account for.

 

that's what people definitely will want to know. The method should be transparent

only 16 lol, tracking 10 people is complicated in PR, and you want to track 24 to know who are this 16?

50% of reserves work for both 24-man group and less. The skill will have less noise from per-person skill differences, but whats the point if you have this huge uncertainty from reserves

Lets assume you've got 6 platoons to split; to make a balanced match platoons skill should be close to their overall mean, and one outlier who is much better will not allow to calculate it to be as balanced as it there were twice as many teams. Having outliers in the form of super pro gamers and super weak gamers do not allow making it balanced. In my experience of team games in wargame if you have the "anchor" who can't keep up the pace you lose as if there are nobody at all

then if one veto, all forfeit? maybe if one veto, they get substituted by 2nd best candidate group. But what if there won't be such a group bc you only have 6 platoons to make it balanced

the point is not that people should work without information, the point is people already calculate the balance for themselves. If their feelings of balance will be hurt you get a veto

in the ref i linked 4v4 means 4 entities - players, groups - versus 4 other entities. It is too many games for getting information to make it balanced. Admins on public servers with crude squad switching manage to make it more balanced at the next round, so playing skirmishes just for the sake of better balance seems very eeeeh

and here is the problem. Toxic people like fecht_niko would disagree, but I think game is meant to be played for fun and not to determine who is the winner. Elo/Trueskill aim not to determine the global winner, but to make a balanced game with equal opponents. The community is small and old and already knows who is capable of what. The goal must be to make a balanced game, not to determine who is the best

like you gonna forbid people to refuse to play the game? that sounds naive

There was 1 epic gamer move to fix the previous tournament: scrap the idea of two fixed teams and start switching squads for balance like on pub games. But it was many times refused to not admit that previous organizational activity was already wasted and to keep straight face when everything burns around you

In the light of that your skill estimates will be vague estimates (I'm sure it will be like that bc beautiful theories often appear to be too clean to work irl with veto, 50% reserves, dislike of skirmishes, big platoons), why would you try to waste a lot of time and gamer potential on skirmish battles. It is not as attractive as 100man battles



how to solve all the problems? assign two master groups, they one by one pick available groups to their team, each new map you choose new masters

 

It's more difficult to organize 90 people, than 10 people. There are too few active people to manage 90, that a smaller number for the team sizes is better.

I think a group of 24, with 16 actives should be perfectly doable. That's the size most clans are.

You are expressing the difficulty you have with organizing 24 people, but there are plenty of people who are capable of that.

I don't think anyone knows perfectly how to balance in PR, it's a very very broken system to use people's emotions and feelings to determine this. We can use numbers and data to get an even more accurate balance. And you are criticizing it. Which is better, a number-based decision making, or decisions made on feelings? <- Rhetorical question.

This whole argumentative list is kind of exhaustive, you want to replace the score-board with feelings?

We envision a system that caters for both competitive players, and those that want to have big games that are balanced. This will give each group the ability to get both.

Your comments seem to be making a lot of effort to shift the tournament to something you want to see, using made-up facts, based on your opinions. It doesn't seem like you are referencing any particular past experiences or problems that you are trying to alleviate.
Your suggestion is making different teams every battle, based on multiple squads. That requires a very active and engaged command team and removes any team identity. We did this in TART 1, and it was decided it was a bad way to do the tournament. You even have the balls to trust the "fictional & active" command team, to base their balance on their emotion... I respect you Chuva, but in this, I am more experienced than you.

I'm trying to learn from the past here, but you might not know history too well, and we will repeat it with this approach. Our suggestion is different and new. Let's try it out!

 

As I said above, smaller platoon sizes do make the team balancing much more effective overall, IMO the number of reserves should be cut down as small as possible. It's already set at half the level of a normal tournament, but if most people agree that it could be reduced further then we should do that.

Nope, just the platoon(s) that forfeit would be punished for it. Those that are ready to turn up and fight should not be penalised because they did nothing wrong.

That's what this approach does.

The intent is to quickly gather enough data to allow some approximate balancing of the larger battles. If you really don't want to spend a lot of time in the qualification phase it's entirely within the rules to use the first morning of the tournament to have each platoon play 6 1v1 rounds and then play only full-scale 3v3 rounds from the first afternoon of the tournament right through to the end. With the exception of the qualification period and some rules to enforce at least some mixing it's left entirely up to the platoons to decide when they play and who participates in each battle.

 

tl;dr the system is redundant, this is small community and we all know each other skills. Also, you can't make a system to measure skill... This is impossible. Not even gonna argue about that.

16vs16 is low, you'll only be able to play skirmishes, 24vs24 would fit right in most 2k maps and all 1k, no assets. If you can't field 24 you can play with less, simple. Large 100p battles done by merging teams can be scrapped imo, too much planning, and commanding that is a mess, a headache.

 

You can obviously choose to not participate in the bigger fights So, you can still participate in smaller skirmishes, with 1 platoon or 2 platoons. And since we have all these teams set up here, you could also more easily arrange friendly games for training or just fun without it affecting your score.

It will definitely be a headache if we have an attitude like that. Let's try to approach it with a positive attitude, I've been around long enough that I know being negative from the onset is really not a good start to anything. I expect anyone joining to be willing to help out or be supportive. And this isn't really aimed at you Raptor, but there are a lot of negative nancies in PR, people love to criticize and complain without putting any work into anything. I think a requirement should be from the onset that a good chunk of goodwill and effort will be made on the side of the participants. <3

Well feel free to stop arguing , I've invited clans from the Russian Community, South American community, North American Community, European Community, the Chinese community, and the Japanese community.

Hopefully, someone will be able to challenge the mighty JWK!

And maybe some of the younger clans will be able to improve their skill level.

 

Well then you'll need to go tell the people who did it that they were wrong.
Good luck with that.

 

very text

 

Can i bring U-777 with me?

 

what is that?

 

wut, why on earth haven't u guys moved over to Squad yet?

 

I am here to balance the assets.