Season 40 - Newsletter

Quarantines Lead to Explosive Growth

May 12, 2020 / by SamE

Hello everyone,

Season 39 started in mid-March, as lockdowns around the world were coming into effect. It was, unsurprisingly, our biggest season ever. But that was apparently just the beginning.

This last Friday, the first day that we advertised in the client for Season 40, we got a total of 181 new signups. For reference, we got only 140 new signups for the entirety of Season 39, with just 30 of those on the day of the client advertisement. We immediately knew we were in for another big ballooning of numbers in the League. Just four days later, we’re up to 360 total signups, an even bigger number than we saw in Season 27, the first season we advertised in the client. (On a log scale, though, we probably won’t exceed that 344% growth.)

It seems clear that there’s actually been an increase in interest in playing Dominion Online, as they’ve also experienced growth in new user accounts and games during this time, in part due to some nice publicity from Vox writer and League player dylanmatthews. Thanks for getting the word out, Dylan!

As such, we will need to grow, starting with the moderator crew. We’re excited to welcome back H.exe (formerly known as Seprix) and welcome new moderators Cave-o-sapien, Earl, eh1414, and truffles. With the 10 of us all remaining on from Season 39, this 50% boost should help everyone keep their division workloads manageable.

We will also need to expand the sizes of some of the tiers of the League. In last season’s newsletter, I had anticipated doubling G to 16 divisions, but it’s now clear that won’t go far enough. If some back-of-the-envelope estimates I made come true, we will also need to double F to 16 divisions, double H to 32 divisions, and then double G again to 32 divisions.

The signups firehose is still running, though, and the moderators are already engaged in a vigorous debate over whether those tier-doubling plans are enough. Given this disagreement, we’ve decided to announce the Season 40 Town Hall, which will take place on the weekend of June 13-14, on the topic of Scaling: How should the League deal with large-scale growth and contraction? Thanks to everyone who suggested other topics; we’ll file those away for discussion at future town halls.

Other Changes

The scaling issues aside, we have also resolved some smaller questions:

Moderator Churn

Remaining from Season 39: alibby1152, Apostolosoruler, crlundy, Gazbag, Lemonspawn, nottoobad, SamE, singletee, volfied, wharf_rat

Leaving: None!

Joining in Season 40: Cave-o-sapien, Earl, eh1414, H.exe, truffles

Suggestions?

As a reminder, you can reply to the forum post with any reasonably-complete suggestions you would like the moderators to discuss at their next rules meeting, tentatively scheduled for the weekend of July 4-5, and we guarantee we will discuss and vote on your suggestion. (You can also still give feedback in the Returning Forms and through Town Halls as usual.)


Minutes from our League Mod meeting rules

As usual, here are the minutes from our League mod rules meeting.

  1. Returning form questions
    1. Division balance
      • No change, because 78% of responders found their division to be somewhat or very balanced. We will continue to poll this.
    2. 7-Player divisions
      • No change, because 86% of responders in 7-player divisions had somewhat or very positive experiences.
    3. Flexible divisions
      • Motion to generally reduce the number of temporary 7-player divisions (i.e. those in E/F/G) in Season 40.
        • YES (unanimous)
      • This is a new issue that has arisen with the explosive growth of the League in Season 39 and especially Season 40. Previous town halls only imagined the sort of small variations in interest we had seen between Seasons 27 and 38, but clearly new decisions will need to be made for these new scales.
  2. Suggestions from Discord
    1. Matches remaining messages
      • Motion to add a breakdown of how many matches people have remaining to the late season bot reminder messages.
        • YES (unanimous)
      • This was a suggestion meant to highlight exactly how many matches each player has left. It will make it easier for moderators to see who has the most to do to catch up.
  3. Suggestions from moderators
    1. Playing under multiple accounts
      • Motion to add something to the rules forbidding playing as multiple League players simultaneously.
        • YES (unanimous)
      • This should be obvious, but it wasn’t explicitly in the rules yet.
    2. Requests to avoid playing someone you know
      • Motion to keep a record of pairs of players (e.g. family members) who do not wish to be placed in the same division, with no promises but the desire to do the best we can to avoid placing them together within the purview of moderator discretion (e.g. in creating mixed time zone divisions).
        • YES (unanimous)
        • We’ve heard a few requests for this sort of thing over the seasons. There isn’t a good formulaic way to guarantee it, but if we have leeway in e.g. creating hybrid divisions, we can try to make it happen.
    3. Accommodating the large size of Season 40
      • Motion to, for the numbers we’re seeing in Season 40, increase the sizes of the F, G, and/or H tiers rather than adding an I tier.
        • WITHDRAWN
        • Arguments for (SamE): It’s very important to me to never unexpectedly demote anyone, so I wouldn’t want to move any existing players down into a new tier. Filling tier I with all new players doesn’t make as much sense, either; then the new players in H will be significantly better than the returners. The addition of new players hasn’t necessarily changed the range of ability levels. I think we’ll have enough space if we just expand F to 16 and G and H to roughly 32 divisions.
        • Arguments against:
        • pt 1 (volfied): We’ll need to figure out the fairest way to draw a line between the surprise demoters and the new players, but I’m confident there’s a way to do it. Adding all new players to I would create too many problems down the road. The upshot of the vote on expanding the league this season was that it should do so vertically, rather than horizontally. Subdividing skill slices is an acceptable way to do this. Skill variation between tiers need not be equal. More tiers produces more balanced divisions and people were quite happy with the balance this season.
        • pt 2 (nottoobad): Given that it’s likely many tiers will double, I personally would be ok with skipping I for now (but I’d advocate for it in the town hall if size stays this way). I agree that there’s no good way to add an I tier this seasons without surprise demotions. What would concern me though, is if the numbers indicate we should expand H to 32 and not expand other tiers– to me, that would lead to too much of a skill difference, as volfied says. Curious to see what the signups look like before deciding.
        • Motion to delay the vote for the tier structure of Seasons 40-41 to Friday (later seasons will follow scaling rules determined in the town hall).
          • YES (unanimous)
        • Since this is going to just be a short-term decision, we’d like to get a better feel for the numbers before deciding.

Minutes from our League Mod meeting rules follow-up

The moderators met again to finish up the topic of tier sizes and promotions for Season 40. We have decided to keep the current 8-tier structure and just balloon the size of the lowest tiers for Season 40. This was a difficult decision, but neither of the alternatives (unexpected demotions or a tier composed solely of new players) was attractive.

We have also realized a need for a more general “scaling” solution – under what conditions do we add or remove a tier from the League? In other words, our rules for tier sizes need to be much more general than fitting the relatively narrow range of player counts we had from Season 27 to Season 38.

The Town Hall will address exactly this question and hopefully help us come to a fully-general answer. For Season 41, we will be releasing our first attempt at a solution which can then be used as a starting point for the Town Hall discussion, along with the other options we’re considering.

Here are the minutes from that followup meeting:

  1. Season 40 Tier Sizes
    1. Number of Tiers
      • Suppose we have at most 786 total players in the League for Season 40. Motion to, in that case, maintain the current 8-tier structure, expanding to at most 1, 2, 4, 4, 8, 16, 32, and 64 divisions per tier (that’s 131 total divisions, so 786 spots).
        • YES (2 dissenting)
      • We had a long and difficult debate on this one. Given the scale Season 40 will be at, it would be ideal if we could add a ninth (or maybe even tenth) tier. However, doing so at this stage would involve either demoting players who weren’t expecting it (e.g. 4th-6th place finishers in H) or forming a new tier entirely out of new players. We have never demoted players unexpectedly in the past, always notifying players of the possibilities of where they might end up prior to them playing any games. Moreover, a tier entirely out of new players would be problematic because some of those players would clearly be better than the worst returning players in H, and it would be ideal to have experienced players in as many divisions as we can.
      • Dissent (truffles): The presence of 64 Divisions in H tier means that, at an upperbound, almost half of all league players will be in one tier. I think categorizing/generalizing such a large number of league players as one single tier will increase player variance and skill imbalance within H divisions. Given that there’s only roughly 12 full H divisions of Season 39 Returning Players, I would like to bound H by 32 divisions (rather than 64), and create a division I, while avoiding any surprise non-demotion of H returners. This leaves 20 divisions to fill for new players in H, and space to place the rest in I. Having even an extra tier of skill separation, especially given the influx of new-to-dominion players appearing in the client, will help reduce player variance and division-internal skill balance.
      • Suppose we get more than 786 confirmed on-time players. Motion to add extra divisions beyond the maxima indicated, and at most five 7-player divisions to fit everyone.
        • YES (1 dissenting)
      • We do not think this is particularly likely at this point, but if it does happen, we will not worry about the consequences of an extra promotion or two in F or G. The flux of players in or out of the League should dwarf any such differences.
      • Dissent (volfied): The majority are probably right about the fluctuations making this work. I just don’t like giving ground on 7-player divisions. They should only be for their original purpose of players coming back from breaks.
      • Motion to add any extra divisions (see previous motion) to H.
        • YES (4 dissenting)
      • This seems like the simplest rule, and will also make it easier to form more tiers at the bottom in the future.
      • Dissent (nottoobad): if we need extra divisions, it might be worth considering where in the scheme they best fit. Almost half of the league will be in H this season; perhaps a small move in the other direction would be best. Specifically, we are over capacity in E; if more players continue to sign up in that level or above it might make more sense to add a division there instead.
      • Dissent (truffles): It seems that we are only expanding at the bottom and not at the middle, especially if we are placing Level 57 players in F. Given that the cut-offs for Season 40 E Division (57+) is higher than what the cut-off was for Season 38 D Division (55+), it may suggest that increasing back D to 6 or 8 divisions would help the overcrowding in the middle tiers. The initial concern for D division was imbalance in skill, but now that the number of Dominion players and League players have grown drastically, there may be enough space to increase D back to 8 Divisions while maintaining the competitiveness of the tier.
    2. Late Signups Protocol
      • Motion to, for Season 40 only, invite late signups to play on a rolling basis, i.e. exactly enough that we know we will fit within 786 total players if all of them and everyone else who is pending, confirms. This will occur on May 19, 20, and 21 and be the responsibility of the mod who closes out the batch whose deadline is that day (but they can delegate as usual).
        • YES (unanimous)
      • This should allow us to add as many Waiting List players as we can while minimizing the number of divisions that might remain stuck at 5 players.
      • Motion to not form any 7-player divisions in Season 40 with Late Signups, but only invite them to fill divisions to 6 players (unlike in Season 39).
        • YES (unanimous)
      • This isn’t something we should be doing in general. In Season 39, we chose to do it in order to make (nearly) every division at each of the E, F and G tiers 7-player. This season, and in general going forward, we plan to simply keep every division to 6 tiers, only forming a very small number of 7-player divisions.
    3. S39 -> S40 Outstanding Promotion Rules
      • Motion to not change the demotion of 5th place finishers in 7-player tier E divisions.
        • YES (unanimous)
      • Since E won’t be growing for the next season, it is already maxed out, and we need to demote these 5th place finishers.
      • Motion to change the 5th place demotions in 7-player F divisions to instead remain in F.
        • YES (unanimous)
      • Conversely, we are doubling the size of F, so there is room to keep these players in F.
      • Motion to promote 2nd place finishers in Season 39 H divisions.
        • YES (unanimous)
      • The players who finished in 2nd place in Season 39 H divisions seem to compare favorably to the new players currently on the cusp of G. It makes more sense to promote them.

What Season 40 Will Look Like

Hello everyone!

We are now officially at by far the largest season of the League ever. And it’s not even close. After breaking the record with a total of 336 players last season, we’ve blown that out of the water with 787 this season, once we fill out divisions to 6 players with signups off the waiting list. Here’s what that looks like:

img

Needless to say, this was unprecedented. We’d gotten used to the steady Season 27-38 numbers. How do we handle something like this? What tiers do we put all of those new players in?

At the Season 37 Town Hall, we all collectively developed plans to expand the number of tiers to eight, which came into effect in Season 39. The motivation behind this change was to keep the tier sizes more comparable, reducing the variance of skills within a given tier.

The problem is that we didn’t have a natural way to increase the number of tiers should the player base balloon like this. We weren’t the only ones not anticipating a pandemic! But given this rapid change, it’s clear that we need to develop general rules for how to add or subtract tiers, and how to move returning players if we need to do so.

This will be the topic of the Season 40 Town Hall, tentatively scheduled for the weekend of June 13-14. For those who are new, a Town Hall is just a designated discussion taking place in the #town-hall Discord channel around a particular topic, closed out by some emoji reaction polls to gauge the feelings of the whole League community. To seed that discussion, and also to make the plans now for Season 41, the moderators have selected the following “scaling” rules:

  • The maximum number of divisions at a given tier is double the maximum for the tier above (starting with 1 in A), except for every third division, in which case it is the same number as the one above. That is, these maxima are: 1 A, 2 B, 4 C, 4 D, 8 E, 16 F, 16 G, 32 H, 64 I, 64 J, 128 K, …
  • The lowest tier has (approximately) the same number of divisions as the second lowest tier.

These two rules imply the following maximal numbers of players as a function of the number of tiers:

  • 8 Tiers: 1 + 2 + 4 + 4 + 8 + 16 + 16 + 16 = 67 divisions = 402 players.
  • 9 Tiers: 1 + 2 + 4 + 4 + 8 + 16 + 16 + 32 + 32 = 115 divisions = 690 players.
  • 10 Tiers: 1 + 2 + 4 + 4 + 8 + 16 + 16 + 32 + 64 + 64 = 211 divisions = 1266 players.
  • 11 Tiers: 1 + 2 + 4 + 4 + 8 + 16 + 16 + 32 + 64 + 64 + 64 = 275 divisions = 1650 players.

According to this schema, with our 787 players in Season 40, we would have expanded to 10 tiers, adding Tiers I and J. However, we couldn’t do this in Season 40, because we hadn’t announced transition rules that would make it possible, and we didn’t want anyone to demote unexpectedly. So how do we get from Season 40’s eight tiers to these plans for Season 41? Well, here are each of those transition schemes:

img

This is a lot! Let’s first focus on the first table – that’ll cover the most likely scenario, where we’re about at the same size. Each row explains what happens to players in the tier indicated on the left – for instance, the top finisher in an F division promotes to E, while 5th and 6th place demote to G. You’ll notice that we include two new letters there – I and J – because we’re expanding to 10 tiers, but it isn’t quite as simple as sticking them at the bottom. You can think of it as adding one tier between the current F and G, and one tier below the current H, then relabeling the Season 40 G and H accordingly. That’s why the winner of a Season 40 H division will be back in H – it is a promotion, we’re just labeling the tiers differently.

But getting first isn’t the only goal of H tier players; getting in the top three could also matter for your tier next season. And if you’re currently in G, well, I’m sorry to say that those bold/italics letters indicate that there’s quite a bit of uncertainty as to where you’ll end up next season. For instance, first place finishers by default will go to the new G (remember, that’s the intermediate tier we’re creating between Season 40’s F and G), but could make it to F depending on numbers. The same is true for the other positions; the tier listed there is above is the more pessimistic scenario. Just do your best, and we’ll sort it out when we see how numbers look next season.

(The next little table can be ignored for 99% of the League; it just covers what happens to that one 7-player division in B. Because it’s a standard flexible division, caused by more returners than leavers at that tier, only the bottom two players demote.)

And that’s just what happens if we stay at more or less the same size as Season 40. We just experienced a 134% increase in the size of the League, and we have no idea if that’s temporary or just the beginning. We originally thought last season’s increase was our quarantine bump! We need to be prepared for future scenarios, and we want to commit that we will not surprise you with an unexpected demotion. In that spirit of transparency, we’re laying out the scenarios for what we will do given what seem to be the four most likely ranges of player counts for Season 41.

It’s possible this is just the beginning, and we’ll see another doubling in Season 42. In that 11-tier case, the third additional tier will be between the current G and H, following the transition scheme shown in the “11 Tiers” table. It’s also possible that we shrink somewhat, losing 100 players or so. If that happens, then we’ll follow the 9-tier system, which essentially adds just the one tier between G and H. And finally, if we shrink all the way back down to Season 39’s numbers, we’ll keep the eight tiers after all with fairly predictable promotions shown on the far right.

Once we have another client announcement and start to project out the size of the League, we’ll have further clarity on which of these scales Season 41 will operate at. Unfortunately, this likely won’t be until the very end of the season. For now, just do your best, and come to the Town Hall if you’d like to weigh in on these matters for Season 42 and beyond.

Best of luck with the new season, everyone! Please be friendly to all of the new players in the League, and if you have any questions, don’t be afraid to reach out to your moderator!