North Star: Improving Reliability and Player Experience

[johnpooch]

Background

We’ve been looking at the numbers, and they paint a clear picture of where the biggest problems are:

• NMR rate: 40.9%
• Active games with at least one civil disorder member: 61.9%
• Players in the “Reliable” tier: 7.89%

NMRs and civil disorder are what ruin Diplomacy games. A 40% NMR rate means nearly half of all order submissions are missed. A game where 62% of active games have an abandoned nation is not a good experience for anyone who is trying to play seriously.

The goal of this initiative is to drive these numbers into a healthier state. We’re not going to eliminate NMRs entirely — life happens — but we believe there is a lot of low-hanging fruit that can meaningfully improve the experience for committed players.

We’ve had some great discussions across Discord about this. This post is an attempt to synthesise those conversations into a concrete plan and get feedback before we start building.

---

The North Star

The core idea is simple: committed players should be able to find and play with other committed players.

To get there, we need to:

1. Give players the information they need to evaluate who they’re playing with
2. Give game creators the tools to manage their staging games
3. Reduce the friction that leads to thoughtless joins and early abandonment
4. Make it harder to ghost a game without consequences

---

Phase 1

Player Profiles and Reliability

Every player will have a public profile page showing their name, avatar, and reliability stats. The stats we’ll show:

• NMR rate — the percentage of movement phases where the player submitted no orders, across all games
• CD rate — the percentage of games where the player entered civil disorder
• Games played

On top of the raw stats, players will earn a Reliable badge once they’ve played enough games to have a meaningful track record and their stats meet the threshold. Players with insufficient history show a New badge. Players with history who don’t meet the threshold show no badge — no negative label, just the absence of a positive one.

Player names throughout the app (in staging game member lists, in the in-game player roster) will be tappable, taking you to that player’s public profile.

Civil Disorder Recovery

Right now, civil disorder is permanent. Once a player enters CD, they stay in CD for the rest of the game. This is wrong and we’re going to fix it.

Players in civil disorder will see an “I’m back” button. Tapping it clears their civil disorder status and sends a notification to all other players in the game so nobody is caught off guard. There’s no automatic deadline extension — other players are notified so they can choose to revise their orders before the existing deadline.

Game Creator Tools

The player who creates a game can now boot any player from a staging game. Booted players can rejoin — this is a soft mechanism, not a permanent ban.

When a player joins a staging game, the game creator receives a notification: “X has joined your game. Check their profile to see if they’re a suitable player.” The expectation is that game creators use the public profile stats to keep their staging game healthy.

When a player enters civil disorder in any active game, they are automatically removed from any staging games they’re in. They receive a notification explaining what happened and can rejoin once they’ve resolved their civil disorder situation.

The Staging Game Experience

Staging games will move into the same layout as active games, with the game map showing the template starting positions. This gives pending games a proper home in the app rather than the current awkward experience of bouncing between screens.

For non-gunboat games, the Public Press channel will be open from the moment a game is created, giving players a place to talk while they wait for the game to fill. This channel carries over seamlessly into the active game.

When a player joins a non-gunboat game, they will be required to post an intro message to the Public Press before the join is confirmed. This is intentional friction — if you’re not willing to write one sentence to introduce yourself, you’re probably not going to be a committed player. There’s no pre-filled default; you have to write something.

Confirmation Before the First Phase

Every player must confirm they’re ready before the first movement phase begins. This eliminates the most common form of ghosting — players who join and never look at the game again. If a player doesn’t confirm in time, they’re removed.

Join Flow Improvements

The join button will open a confirmation dialog showing the phase duration and the commitment being made. If a player doesn’t have push notifications enabled, the dialog will surface a warning and a prompt to enable them. Push notifications aren’t required to join, but players should know they’re increasing their risk of missing deadline reminders.

Notification Improvements

Two improvements to how deadline reminders work:

• A second reminder will be sent closer to each deadline. The first reminder fires based on the existing timing matrix (scaled to phase duration). The second uses a separate matrix tuned to be a “final call” rather than an early warning.
• Email notifications will be enabled by default for high-signal events: deadline warnings, phase resolved, civil disorder, and game start. Players can opt out in their profile settings. Email fires regardless of push notification status.

Small Changes

The default number of NMR extensions when creating a game changes from 0 to 1.

---

Phase 2

Player Replacement

Players in civil disorder can be replaced by other players who sub in. The replacement system will include a way for available nations to be listed as open, a way for players to claim them, and a reliability boost for players who step up to replace an abandoned nation.

Medals

Beyond the reliability tier, players will be able to earn medals for specific behaviours we want to encourage. The first candidates are:

• Chat medal — awarded to players who communicate actively
• Reliability medal — awarded for sustained reliability over a long period
• Experience medal — awarded for having played many games

---

What We’re Not Doing (Yet)

• Reliability-gated games (where only Reliable players can join). The manual boot gives game creators this control without adding a hard filter.
• Automatic deadline extensions when a player returns from civil disorder. Too many edge cases.

---

Questions for the Community

• Are there behaviours that contribute to NMRs that we haven’t addressed here?
• Is the required intro message on join too much friction, or about right?
• Are there other medals that feel like obvious candidates for Phase 2?
• Anything in Phase 2 that you think should be pulled into Phase 1?

[KittyofEvil]

Enthusiastic thumbs up to this; these changes sound great and will be a big improvement to the site. I particularly like the changes to the staging experience: having players go right into the map and having to confirm participation will engage players from the start.

I hadn't considered forcing a message on-join, but I think I like the idea; it sets the tone that this is a game about communication. As for your other questions, there's nothing else I can think of right now.

On having a player's NMR rate displayed on their profile: will it be permanent? I'm thinking of the way that Chess.com does it, where the score is something like 'timeouts in the last 6 months'; that way, you can always get back to a good score by playing consistently for a while. The all-time NMR rate could still be displayed as a secondary score somewhere on the profile, but maybe 'recent NMRs' should be the headline? Just an idea, maybe overcomplicating it.

The only thing I'm not sold on is putting the responsibility of policing players onto the gamemaster instead of having hard filters: I think being able to set up entry requirements is the way to go.

What if the last player who joins is the unreliable one, or indeed if a few people join before I've had the chance to vet them? Then, presumably, the game would start with no ability to remove them.

It also means that I'm not sure what I'm getting into if I join someone else's game: will this GM curate a game of reliable players, or will they let anyone in? It means that if I want to ensure a game of reliable players, I need to set up my own game instead of joining one, which splits up players.

I'm also thinking about this from the point of view of the new or unreliable player who will be the target of removals: I would rather not be able to join a game at all than potentially join several games in a row and not be able to start them, presumably without explanation; I would find that very frustrating. It's a much better experience when everyone knows what game they are joining before they press 'Join'.

Another advantage of explicit filters: games with entry requirements can be visible at the bottom of the Find Games section so that new or unreliable players understand that there is something to be gained by playing and sticking with games.

To put it in one sentence: if the stats are going to be presented to us to let us filter out players, why not let the computer make that decision for us?

I hope that didn't come across as too negative; even without reliability-gated games, these would all still be great changes that would make me a lot more enthusiastic to join games on the site.

[JorenC]

As always, good thinking, and love your calmness, John.

Phase 1: Foundational Reliability & UX

Player Profiles and Reliability

• Public Profiles: NMR, CD rate, Total games played.
  This is also low hanging fruit to add win/loss/draw rate, and other stuff. Makes people engaged!

• Reliability Badges: Reliable/New: No negative labels
  Agreed.
  We need to have some sort of pop up when you tap the reliabile or new badge, to explain what it means. & I'd hide this label during the game. No need to show that Italy is a newbie; only on player info (if someone goes looking) and pre-game, related to the person, not the nation (e.g. Joren vs Italy)'.

Civil Disorder Recovery

• "I'm back" button & notification All other players in the game will be alerted when someone returns.
  I like this. We could do this automatically (when they send orders), but for someone to acknowledge they left the game - it works better. Makes it more salient

Game Creator Tools

• Staging Controls: Game creators can boot players from staging games.

• Automated Safety: If a player enters civil disorder in an active game, they are automatically removed from all staging games they are currently joined in.
  HELL YEAH

• Vetting: Creators receive notifications when a player joins, prompting them to review the player's profile stats.
  The 'someone joined' notification is good, in general - also if we have the pre-game chat.
  If we do this, I'd strongly suggest we also add Game Masters (a non-player person who can control the game). This allows moving discord games with variants here, and all features we make are making more sense for a GM than for one player with GM powers (enter homelander meme here). There is a risk with one player having 'super powers', as long as it's for the pre-game only.

I agree with AJ that this can better be pre-gated on join. If you as GM KNOW you will only accept reliable players, why have to manually vet it? It seems like we're adding steps for no reason? What is the benefit over allowing users to presetting this?

_Maybe it's more work (hide not eligible games, add messaging if a non-eligible game is opened via link etc, but I'd say YES AND - allow vetting AND allow gating. _

Staging Game Experience

• Layout Update: Staging games will utilize the active game map layout instead of separate screens.
  Yes, but we CAN'T show nations before the game starts - we DO NOT want secret diplomacy to start before the game starts. Everything needs to be public and all info needs to be available to everyone - even before they join.

Also, the 'public press' needs to have a first message from the system saying that all messages in public press will be visible to all visiting players - especially if we pull (which should be ready: #551)

• Pre-game Communication: Public Press channels open immediately upon game creation.
  HELL YEAH

• Introductory Friction: Joining a non-gunboat game now requires writing a mandatory introductory message, serving as a filter for committed players.
  HELL YEAH. Also, new player join notifications should go out to everyone in the staging game, so it doesn't become a 'dead wait' for 3 days, but "oh we have someone new! Welcome' every couple of hours experience - and not just to the GM.
  Basically, if we add this introduction message directly into the Public Press channel, that would notify all users in the game as a message - might need some specific, additional custom notification text, but it's the same mechanism. "John joined the game: "Hi all, I'm john!".

Join & Engagement Mechanics

• Readiness Confirmation: Mandatory confirmation from every player before the first movement phase begins; failure to confirm results in removal.
  This needs to be optional. There can be reasons to not do this - we tried to create one 34 player chaos game, organised with only active players and everything super organised, but every turn at least one player would drop. And because this actually takes a turn time (you need 24 hours to give everyone the time to confirm, one player being busy with life - which always happened - meant all 33 other players had to wait another day. It's really good, but really needs to be optional
• Join Flow: Updated dialogs surface information about phase duration and push notification status, highlighting the risks of missing deadlines.
  This is combined with the 'first message for public press'
• First turn start time: _This is for duration-time games only, but indeed, if we want a 1-hour game, with everyone active, we don't want to have fixed times, but we also don't want to have the game start at a random time when random person appears in the middle of the night. Allowing (at least for duration-time games) a 'game start at' means when all players join the game does not start yet (or starts with 'extended deadline' as in fixed-time now) and at least everyone knows when the game starts. _

Notification Improvements

• Deadline Reminders: A second "final call" reminder will be implemented closer to the deadline.
  Not sure - agressive and I'm not sure it'll help. I wake up in the morning and see 10 missed notifications of a game (currently - you saw the screenshot!), I don't think another one helps. If I'm playing, I'll respond to the first one.

• Email Defaults: High-signal events (deadline warnings, phase resolution, etc.) will have email notifications enabled by default.
  _Very strongly disagree with this one - we need to be very careful. This can become spam very fast, then our registration/password reminders become spam. It should be an option for the user to turn on themselves, but not by default. MAYBE only a deadline warning with no orders (and basically that email is not 'you have orders to give' but 'you should turn on notifications!!!'. _

• Game Settings: Default NMR extensions for new games increased from 0 to 1.
  Not sure. This just means that if one newbie CDs at the beginning, everyone needs to wait 2 turns instead of 1 before they can play. Not the best experience. Combined with the CD 'I'm back button', I don't think it's needed.

We should set this for 'reliable' games - even set to 2 instead of 1: reliable players miss a deadline because of shizzle, and we don't want to break their reliability/NMR track. You know it's going to be accidental if a reliable player became reliable.

Lasting effect of 'unreliable'

_Agree with AJ, we need to have some sort of 'moving average' or weighting algorithm that counts recent games more strongly than older games, to allow players to 'recover' easier.

Phase 2: Advanced Replacement & Rewards

All good. The above already seems like 5 phases to me :D.

[Counterink]

In response to AJ (and a quick one hopefully) I feel that tracking NMRs based on a number of games last played, rather than an amount of time, would be preferrable, considering that people would then have to actually prove themselves to be reliable instead of waiting for their penalties/low performance to go away.

Secondly, I may be misunderstanding this, but will players be penalised for going into CD or missing orders if a significant number of other players in their games have done so already? I'm all for giving extra credit to people who actively attempt to keep games alive, such as by giving substitute players a reliability boost or a particular medal, but at a certain point, playing in a game where most other people are absent (or have been removed, if some of these ideas are put into place) is not an enjoyable experience, and therefore I feel that these people should not be penalised with a loss in reliability. Saying this though, if only 1 player abandons the game, then I feel that this is not a sufficient reason for other players to do so as well, though even then, this depends on the variant. I suggest that either A.) reliability loss should be calculated in proportion to the amount of players remaining in a game (so that if many people go into CD then missing turns offers negligible loss), or B.) that there be a line drawn on a per variant basis which would negate any reliability loss once a certain number of people leave.

I think pairing this with a reliability gain for still making moves despite absences (similar to the bonus for substitute players) would help prevent this from just killing games where multiple people leave, but my main point is that I don't think people should be expected to continue play in games where, say, 1/3 or 1/2 of the players are missing, as it is just not fun.

Do let me know what you think about this (or if anything I've said is redundant).

[johnpooch]

Thanks for this, Counterink — really good points.

On the question of reliability and broken games: I'm not sure I want to go down the route of a proportional calculation, mainly because it adds a lot of complexity and would be hard to explain to players.

I think the better approach is to give players explicit mechanisms to exit a game gracefully:

1. Easy draw — if the remaining active players agree the game is effectively over, they should be able to end it cleanly with a draw.
2. Concede — a player can explicitly concede defeat rather than just going silent. This is different from civil disorder in that it requires a deliberate decision, which I think is worth rewarding.

The idea would be that conceding carries no reliability penalty, but only under certain conditions — for example, after a minimum number of phases have been played, or once a certain number of other players have already gone into civil disorder. This prevents abuse (you can't just join and immediately concede penalty-free) while still giving players a graceful exit from genuinely broken games.

We haven't figured out the right thresholds for when penalty-free concession kicks in, and the UX for surfacing this clearly is also an open question. Curious what others think — does this approach make sense, or is there something about the proportional calculation that you feel solves a problem this doesn't?

[JorenC]

I think there are all very good points here.

Advanced players:

• Know the draw mechanism (how it exists and what it does)
• Know what reliability means, how it's impacted and how important it is.
  We can't expect new players to understand. Even if we explain it, they don't read it - that's not how you learn an app - you learn by trying.

Basic players:

• Play a game until it's not fun anymore.
• Might try again after a false start.
• Only later learn these 'advanced' concepts of reliability and draws.

We always told them 'It's very important, just add some random orders' but that is not the way - and didn't work.

---

Maybe we SHOULD rethink CD not as a 'abandoned' but 'conceded' option. Like Starcrafts 'GG'. Like Chess dropping the king.
Especially for newbies. We might even have different rules for newbies and non-newbies.

Imagine: You're penalized when you abandon a game, but this is based on
(a) amount of SCs left compared to average owned by all players (basically, your winning chance)

• We use average owned SCs, because there are less total SCs owned at start - leaving with 3 at turn 1 is bad, 'GG-ing' with 3 while Russia has 17 SCs is less bad. This basically calculates your 'chance of winning'.
• This looks gracefully to people leaving when the game is almost over for them. If they're eliminated, (0 scs) they get no penalty for abandoning.
• We should not necessarily super actively mention this - just calculate it in the background.

(b) percentage of other players in CD.

• If you abandon as the first player, you get a higher penalty. If 5/7 players are already CD, it's "logical" you leave, and you get a discounted penalty - but there should be one.
• This should be calculated at how many people abandoned BEFORE you abandoned. E.g. if 3 players CD in turn one, they ALL get a 7/7 penalty, but if one player abandones after that, they get a 4/7 penalty (examples).

(c) We reward new player for playing while others are in CD

• For newbies only, if we see they keep active while other players CD, we actually boost their reliability - these are players willing to invest and play, even in the (likely - in the inevitable snakepit of newbie games) adverse situations. This means they get to 'reliable' faster.

(d) We actively explain this to them

• At the end of a game win/draw/loss screen, we also show how the game affected their reliability and why - especially for new players - so they get a boost message 'You kept playing! Great - you're close to becoming reliable, and good games will open up!' or 'You abandoned early. This game is as good as the people playing - if you want the best games, keep playing'.

[KittyofEvil]

I agree with John that having a system that is easy to understand and explain to players is preferable: I was imaging the NMR score as more of a percentage, an easily-understood 'This player has abandoned 25% of their game' kind of note. While the complexity of implementing a proportional system can be overcome, it's the nebulousness to the userbase that I think is more of an issue. For similar reasons, I don't like the idea of having one rule for new players and one rule for advanced: the system should be consistent and treat all players equally.

I also don't like the idea of having users weigh how much abandoning a game will affect their rating: I would rather that users are discouraged from abandoning a game at all and encouraged to aim for a 0% NMR rate.

I would prefer that users are directed towards the Draw Proposal mechanic somehow. Joren is absolutely right that casual players can easily miss the option; perhaps this could be helped with a redesign of the UI to make the button more prominent, or even a popup message that displays under certain conditions ("4 of 7 players have abandoned this game - you can end it with this button, you know.").

I also think a Concede button would make sense once the Player Replacement system is up and running, but I think it should still be acknowledged that concessions do affect games negatively. Perhaps concessions should be tracked as a separate score to NMRs or something, but I would still rather play with someone who sticks with the game to the end.

On a related note, the ability to set up standing orders when you concede would be nice as a way to less the impact of your departure.

[mretzel86]

Hej,

First of all thank you John (and Joren of course) so much for your enthusiasm and passion with this project.

First I will try to describe the players coming her with the Skill-Will-Matrix. Maybe by doing so you will get some other ideas what to improve:

High Will / Low Skill Players:
New player. They heard of the game, registered, see the home screen and want to start a game. They read the game rules, but have no clue, what they are doing and just add random games and create a bunch too which sounds awsome. Than start playing. Might get in the second category if overburdend.

Low Will / Low Skill:
New player as well. But are inpatient. The community is not big enough yet, so games take long to fill (couple of days), they get frustrated and log off without leaving the games they joined (ruining the game for the experienced ones).

High Will / High Skill:
These are the comitted players. They keep playing to the bitter end and are happy if they get a rank, new variants or improve their stats.

Low Will / High Skill:
These are players who have played some time but got annoyed or dissatisfied with something.

So what to do to get everyone to the high skill / high will level or in your language what to do to reduce NMR rates (CD)? Focus on the high will / low skill players and the low will player / high skill players. Personally I would drop the low will / low skill player. Ignore the High Will / High Skill players (to some extent).They will stick around as they see you are putting so much heart in this project and keep adressing issues.
I know some of the discussion is going on on Discord but I will pick up some ideas I liked.

Thoughts to improve the High Will / Low Skill Players:

• new players get a "Beginners badge"
• they can only join/create 2 or 4 players variants. For the first 2-3 games or such.
• they cant join more than x games simultaneously until they got 10 games or such. Lets them focus.
• they can only play 24 h games. With good reliability you can play shorter duration times.
• Maybe introduce "tutor games" which are somehow highlighted as such so experienced player can teach new players. This could be a teaching 1vs1 game or maybe the classic variant with a neutral "war reporter" who can only message (probably to difficult to implement).
• The restrictions will be explained in a "Welcome Message" as they log in the first time and how to remove them
• Concede option as Joren described it

Thoughts to get rid of the Low Will / Low Skill Players:

• remove players from staging games who go in CD state in any game. If they come back they can rejoin. -> If possible with a message why they have been kicked
• remove staging games dynamically after a couple of days, e.g. 2-player variant after 4 days, 7 player variant after 14 days, 12 player after 24 days etc.

Thoughts to motivate the Low Will / High Skill Players again:

• Improve notifications system
• improve overview which game needs ordes! (Those are my pain points, and probably the most boring to implement)
• civil disorder recovery

Right now you are spamed (!) with messages, especially in games with fixed durations where you cant submit orders. Thats why I think as Joren its critical too spam more or even spam with emails. Pls dont do it. No messages if no actions are required (maybe a costumable setting for players preferences?)
Whats good about the app, if you click on the message you will be fowarded to the game. But it doesnt say if actions are needed, only that the phase resolved. So I always keep checking the games despite no orders to give.
Nice have in the notification: 1) Game Name. 2) Game Variant, 3) Country you are playing, 4) How many orders to give. 5) Time remaining. I saw you started on this with the "deadline approaching" notification.
In the "My games" home tab it would be great to have the same overview (extend it): 1) which nation am I playing, 2) Do I have orders to give or are they confirmed. 3) owning SCs (or eliminated) 4) unread messages -> with one look at the tab I should be able to see if I have to give orders or reply to a message in one specific game.
The option to "hide" games maybe with a new tab. Once I am eliminated from a game I dont want the game in my "active" home screen

Thoughts to keep the High Will / High Skill Players:

• More badges to "collect / recive"
• Set a continent as "time zone" in your profile. Easier to know when to expect an answer.
• Game variants, but you are already doing a gfantastic job with the variant creator!
• Copying orders is a hassle in the app. You have to do it manually instead of with a button (old app). In comination with the break line it will be much easier to coordinate with allies
• Statistics like Number of Games, NMR = (NMR turn/ number of all turn), CD = CD games / number of all games, etc. Maybe two number: overall and in the last 10 games / 50 turns or so? Same as AJ suggested. Make the formula public so everyone understands the numbers. Maybe combine the statistic with a lable/Badge later on.
• Ranking. Here I would propose the dynamic ELO System. Means if you dont play you loose point over time and new players gain more points / loose less point and experienced players vice versa.
• Tutor a game will get some kind of reward (Badge / Stats)
• If the community gets bigger maybe start a Token system. Earn tokens for 5 non-NMR games or so and play one game with that token. If you "spend" something, you dont want to loose it.
• As long we are small I wouldnt exclude anyone and give him/her a chance! I agree where with AJ aswell. Thats why I wouldnt give the gamemaster any rights to exclude players but rather do it automatically. If the algorithem says I`m out it a neutral judgment, and not because the GM "thinks" I am unreliable (or discriminates me in any other way).

Further comments / my opinion on "planed" changes:

• I agree with your phase I and II plan
• NMR extensions prolong the game / phase and most of the time the player still doesnt submit his orders (maybe with more players this will change in the future). If I miss a turn, I missed it. Thats life.
• Pre-chat: I will hate it but will live with it. Im not the chaty player...
  What to do about players who join, chat (spam) and than leave?
• Ready up: This will delay the staging game quite a lot. Dont know yet if I will like it. Maybe Optional? Or maybe only for variants with a lot of players or short / fixed duration phases?
• Join flow improvemen / message on join: Great thing!

I hope I could give you some feedback with what I think will make the game / app more enjoyable. Maybe it will inspire to better ideas. However, most important: Keep enjoying!