RE: The CW is implemented now causes players to plan their day around CW participation

Indeed, that is the general complain I hear as well.
And I totally agree with your opinion about it.

Edit: the points for playing would, indeed, only work in a scenario with a cap (both on number of players and on number of times 1 player can play).

Personally, I do think those have their merits. Be it, to avoid too much stress on everyone involved.

Edited 8/26/2022 16:15:23

Definitely CANNOT provide points for playing without a cap. I was thinking you either get a win (full points) or partial points for playing, not just unlimited attempts at getting a win and partial points for every attempt.

I don't see the logic in letting everyone play as many times as they like. Just end up with a small # of people playing furiously. In which case, might as well just keep it as is and let players use alts in the same clan - same result. And still further fortifies participation over skill - whoever is willing to play in 6 slots per day has a distinct advantage over those who don't.

At least 40 CW cap and 1 game per day creates an upper limit boundary on how much a single player can sacrifice himself to the cause. I would argue b/c of the mechanics though, even this is too high a limit, it's already too much effort.

Minimally, publishing the templates for timeslots for the next 7 days would allow players to plan the slots they want to be around for. Nothing is worse than staying up late/waking up early for a slot just to find that it's not something you want to play.

Nothing is worse than staying up late/waking up early for a slot just to find that it's not something you want to play.

"Daddy, why did you not spend more time with me when I was growing up?"
"Sweetie, I was staying up late for a Clan Wars slot."

Okay... another option for y'all to troll me.

On the topic of having to organize your daily schedule around CW

Last night, I had the following snap thought: what if, each day, for a clan, in stead of counting the number of wins, we count the number of wins, divided by the number of games played, with a (beneficial) adjustment for the number of games played.

Concrete, we could, for instance, use
Clan's Daily Score = ( # Wins * m / # Games + # Games * Participation Bonus ) / (m + Participation Bonus ).

Where m is a multiplier, to be calibrated to balance the impact of Wins vs Participations.

Remarks:

1. Each Game played will add a constant Participation Bonus / (m + Participation Bonus ) to the Daily Score.

>>> Note: it does not make sense for the Participation Bonus (PB) to be decreasing as a function of the number of Games played. That would destroy the very purpose of the PB, i.e. to convince people to play (to help their Clan, regardless of the game's result).

2. For a given number of Games, the increase of the Daily Score as a function of the number of Wins is m / ( # Games * (m + Participation Bonus ) ).

>>> Note: the more games a Clan plays, the less the marginal impact of a win on the Daily Score. This feature, together with the PB, should help reduce the stress of having to win.

3. The division 1/(m + Participation Bonus) is a rescaling, to set the Daily Score to 1, in the (reference) case where a clan plays and wins only 1 game.

>>> As a result, visually, the Daily Score seems to decrease, when you increase m and/or the PB, but this is just a visual effect. What matters is that the order is preserved and that is clearly the case.

4. The higher you set the Participation Bonus, the more weight is given to playing (rather than winning).

5. By contrast, the higher you set the multiplier m, the more weight is given to winning.

>>> Hence, these 2 parameter have to be carefully calibrated, to obtain a nice balance between Playing and Winning.

6. I would abandon the entire Map + Captured Territories concept. An go for a pure Score-based ranking.

>>> Doing so will also allow for a fixed seasonal term.

---

For instance, set m = 1.25 and set the Participation Bonus to 1/3.
Assume a clan won the first game. So, their Daily Score = 1.

Now, they would be evaluating whether or not it is worth playing another game.
* If they stop, their Daily Score will be 1.
* If they lose the second game, their Daily Score drops to 0.82.
* If they win the second game, their Daily Score increases to 1.21

Depending on the skill of the second player, I would say that it is worth playing the second game.

This is definitely true the higher the number of Games played becomes.

Assume, in the example above, that the second player lost. The Clan's Daily Score has dropped to 0.81.
Now, still using m = 1.25 and a PB = 1/3, when evaluating whether or not to play a third game, the Clan will conclude that
* If they stop (1W, 1L), their Daily Score will be 0.81.
* If they lose the third game (1W, 2L), their Daily Score increases to 0.89 (because of the PB).
* If they win the third game (2W, 1L), their Daily Score increases to 1.16.

So, it would always make sense to play.

For those interested, you can play with the numbers at https://docs.google.com/spreadsheets/d/124B2b6Q_zzlSFH9k5pCQWNIHszw2qUoqplGYLmBMkQA/edit?usp=sharing

Edited 8/27/2022 13:00:36

Minimally, publishing the templates for timeslots for the next 7 days would allow players to plan the slots they want to be around for.

Yes, that would be very handy. Even doing that in 24 hours in advance would be beneficial and not breaking current template selection algorithms.

Last night, I had the following snap thought: what if, each day, for a clan, in stead of counting the number of wins, we count the number of wins, divided by the number of games played, with a (beneficial) adjustment for the number of games played.

What issue is this supposed to solve?

wins/games is just going for average WR, so that just makes small clans with good players the optimal CW clan

Clearly neither of you 2 read my post (correctly), nor, and definitely, did you look at and/or understand how the proposed formula works.

The issue(s) that I am trying to solve are clearly outlined, as is how the formula would succeed in solving these issues.

I surely appreciate feedback, but please, do the little effort of reading what you comment.

Thanks.

Edited 8/27/2022 21:01:35

If the goal of the suggestion was:

So, it would always make sense to play.

Then the suggestion is 99% redundant, because currently for all clans, except MH and MB, it already always makes sense to play another CW game.

Thanks.

Edited 8/27/2022 21:05:11

Do yourself and everyone else a favour and read the entire post.

Educate yourself, instead of shooting comments based on a few lines.

For instance: no, in the current CW mechanism it does NOT, per se, make sense to play. Because, (a) there is a max 40 players cap and (b) losses don't count at all for anything. So, a clan could have 40 active players on a day and still get a 0 result.

The fact that participation, in itself, is not rewarded, is something that my formula aims to solve.

When you know that you help your clan by playing, even when you lose, you should be more incentivised to play

Appreciated.

Edited 8/27/2022 21:49:57

I am very sorry if you do not understand how current CW mechanism treats CW losses and how the participation, in itself, is rewarded. (Just as previously with underestimating the SEAD popularity or the premises of introducing the 20-cap)

Appreciated.

---

To all the guys reading this, if your clanmates do not appreciate your losses in CW and therefore you are not comfortable with losing in CW, then apply to Prime (former NEETs).

In Prime if you bring us 29 losses out of 30 games in a CW season, you will be just as MVP as someone who scores 29 wins out of 30 games. We have full understanding of how these losses help the clan in CW, and would be thankful to you for your efforts.

(BTW, we would be happy even with your 30/30 losses, but you still would probably want that one win for yourself to qualify for the rewards :) )

PS. This is not an ad, this is really to help folks, who are not supported by their clans to just play CW regardless of winrate.

Losses aren't *exactly* the same as wins. Norman pointed out the other day that high-participation clans get a "squad bonus" on win-rate by overwhelming the top-down matchmaker and getting weaker opponents. The more of your clan that joins each slot, the weaker your opponents get because you start at the top and work down.

This means that if you lose, you forfeit some of that squad bonus, because the squad bonus itself depends on over-performing the 50/50 win rate without having to actually over-perform your rating. If you lose, it means you didn't get the squad bonus for that game so even though the loss will generate a win later that's a 50/50 tradeoff when you could've gotten a slightly better-than-even deal.

Little tricky to wrap one's head around but the idea is simple to understand when you think about it for a clan like M'Hunters: for M'Hunters, a loss is strictly worse than a win. The higher your sustained win rate (above 50%), the more you should prefer wins to losses. Conversely, the lower your sustained win rate (below 50%), the more you should prefer losses to wins.

Edited 8/27/2022 23:09:23

The "squad bonus" is an interesting insight.

Just ran the numbers on previous CW season and looks like on average MH get more lower CW rated opponents, than MB. Probably that is also the outcome of SEComBomb template focus. Dont have the exact number yet of "average lower CW rating"(maybe will calculate that later).

@Fivesmith, thank you again for your incredibly valuable insights.

Do me a favour: provide fact-based and constructive comments. Please.

I am floating ideas, because I am perfectly aware of the current CW mechanism (incl. all of its features), observed the effects and ran the numbers.

The Clan-rating feature was mentioned already earlier in this topic. While its a priori intention is good, actual CW practice has shown that it is flawed, as also highlighted by l4v. If it can be fixed, great, but currently the CR is distorting rather than aiding the match-making.

In addition, other than being a parameter in the m-m process, to me, the CR is pointless. Yes, it appears as a figure on a clan's page, but who actually cares about that figure? What people look at are the clan's results in the past seasons.

RE SEAD: please get your facts straight. When referring to the latter template, I was not judging its popularity, but arguing that it would be interesting to have template rotation, based on a voting system. In this context I was wondering if having 3 SE templates isn't a bit too much. However, l4v brought up some good arguments (see, it is possible...) as to why SE make sense in CW, to which I then agreed.

Finally, I am baffled as to how you conclude that my (initial) proposal of reducing the player cap to 20 implies that I don't understand the CW mechanisms.

In fact, this is a complete non-argument. In se, the cap has no direct relationship to, for instance, the match-making, the clan rating, e.a. Increasing or decreasing the cap would only influence the latter, through the composition of the player-set and/or a rescaling of the score, rating, ... (which would not alter any order).

Mark, when I made that proposal, the idea was to counter the CW-fatigue among players and not the engine of CW itselft. I never even hinted the latter.

Edited 8/28/2022 06:22:45

Yes, it appears as a figure on a clan's page, but who actually cares about that figure? What people look at are the clan's results in the past seasons.

Muli's Userscript uses it for sorting and in a world where the ratings were less noisy they'd be pretty neat because they're basically clan-level Elo ratings. It's the closest we have to a clan skill rating that's not just for the Clan League clans (Clan League results are noisy in their own ways and become stale very fast).

Mark, when I made that proposal, the idea was to counter the CW-fatigue among players and not the engine of CW itselft. I never even hinted the latter.

+1. I hope we can step back from fighting over specific changes and mechanics and remember that we all have the same intent: to make CW an enjoyable, rewarding experience for as many players & communities as possible. Whatever our disagreements on approach may be, we're only disputing different ways to achieve shared goals. We should assume good faith and reason from one another if we want to actually achieve those goals someday because we're much more powerful if we can discuss and then cooperate to push a shared solution.

I get the point on the Clan-level ELO and concur it's a nice feat.

However, I don't think the CR is a correct representation of the overall skill in a clan. In most clans, only a handful of players participate in CW. If we want a true measure of a clan's overall skill, I think some sort of (weighted) average of the (global) QM-rating would give a more realistic picture.

On CW, as mentioned, I am definitely a fan and all my respect to the team behind it. It's just unfortunate that the community isn't really heard, while they have valid remarks.

Edited 8/28/2022 06:38:50

I think some sort of (weighted) average of the (global) QM-rating would give a more realistic picture.

Speaking of which, I like your community-run "CW 2.0" idea. The issue of course is that (imo) the community simply can't create a casual-inclusive or continuous/repeated real-time event. Only Fizzer has the reach to do that.

But what we can do is create a CW 2.0 as a clan rating/ranking system on top of Quickmatch. That way we won't have to attract players or create any games. We'd just need to use the query game API continuously to get data from Quickmatch games and create a ranking for clans on top of that, which then we could ourselves try to improve our clans' rankings on as its own metagame.

CW 2.0 is an interesting idea, and leveraging for a pilot of it makes it easy. Certainly no one is going to sign up for both CW & then also have to arrange timeslots for CW 2.0 RT games, so either QM whenever-someone-wants-to-do-it or MD games is likely the only way to test it out.

Hang on a second.

I think krinid is mixing 2 things.

1. QM: I mentioned that in the context of calculating a Clan Rating-like figure that would be a more realistic representation of the average/overall skill of a clan.

>>> This, imo, is something we can suggest to fizzer (by e-mail, or during one of his Q&A sessions).

Calculating and publishing, for each clan, (i) the arithmetic average of all players' Global QM rating, or (ii) some sort of weighted average of the players QM rating for the different QM templates should not require that much work. The input figures already exist. It's just a matter of correctly processing them.

While (ii) would obviously be the more correct representation, I assume we should go for (i), as it is easier and more transparent.

2. CW 2.0 (community-driven): when I floated that idea (and in my head, still so) it was not related to the QM idea in 1.
The idea was to set up a (small-scale) experiment, with interested parties, to test some of the improvements we have discussed here.

>>> If I were fizzer and someone comes to me with some perceived improvements, I would ask (or at least wonder) are you sure this will work the way you expect and intend it?

The answer is: no, we are (or at the very least, I am) not. Forgive me, if and when you think this is overkill, but I'm a scientist, so I like to test theories empirically, before making conclusions.

>>> Yes, it will require some work to set this up, but (maybe naively) I am rather optimistic about that.

* I agree that RT will be tough to handle, both for the team managing it and for the players.
> At some point, I had the idea of sticking to 1 timeslot, but then you run into issue that no 1 timeslot is convenient for everyone.

* So: MD it is.
> However, to avoid the MD-driven issues you observe in CL, I would set the boot time to 24h max.

* Then there is the topic of templates, participation, registration, match-making and game-creation.
> While an automated procedure is always preferred over manual processing, I think the latter is manageable, assuming that
>> ideally, we work with a closed set of clans, to avoid having to generate traffic;
~~~ Games can then be announced in Clan chat, by an intra-clan liaison;
>> games are launched x-daily >>> this means you can have several games running in parallel, but I don't see an issue here;
>> registration (i.e. template selection) can be done via a Google Sheet
~~~ I concur that this is far from ideal. If you have a better solution, please share.
>> Match-making can be done on the basis of the QM-template ratings. Assuming the player pool is not too large (which seems reasonable, on the basis of CW), this shouldn't take too much time.
~~~ I would omit the age of the account. Imo, account-age is not a measure of player-skill.
>> Game creation is of course the easiest task.

---

There is, of course, the open question: which features do we want to test?

>>> If I were fizzer and someone comes to me with some perceived improvements, I would ask (or at least wonder) are you sure this will work the way you expect and intend it?

I think it's a little presumptuous to expect him to ask that, because he doesn't ask the same of his own sweeping changes. Not being snarky or anything, but I would expect something other than a demand for certain effectiveness to be the reason for how he basically ignores community suggestions even when they're well-reasoned & widely-supported. Maybe he's not so much interested in effectiveness but also in terms of what he himself would most enjoy working on?

Either way we shouldn't try to tailor our own suggestions based on guesses about his decision-making process. That turns this whole process into a noisy game of political coalition-forming, and coalition-forming has undue impact on prioritization and such. Rather we should treat him as a black box and just tell him:

1. What's working well and not working well for us
2. What we think would improve ours and other players' experience
3. Some ideas we have about improvements, with our estimates of how impactful they will be and how much we'd like him to prioritize them

Then we let Jesus take the wheel. If instead we have to figure out Jesus' decision-calculus, then we will get nowhere and just waste a lot of our own time filtering our own ideas based on a probably-wrong model of how Jesus will parse and prioritize our suggestions.

---

> However, to avoid the MD-driven issues you observe in CL, I would set the boot time to 24h max.

If we're just building another MD clan-level competition, then why not just Clan League? I don't understand how a MD clan-level competition would have anything to do with Clan Wars rather than just being its own thing.

Re: If we're just building another MD clan-level competition, then why not just Clan League? I don't understand how a MD clan-level competition would have anything to do with Clan Wars rather than just being its own thing.

Fundamentally, it would not be exclusively CW, nor would we be building something (new). In essence, we'd set up a real-life experiment, to test some development ideas.
But, yes, this can also be done in the context of CL or whatever competition we got going on.

Edited 8/28/2022 20:43:59