More photos of GDC Soon - It’s over!

@Sly: the issue is that guards don't scale to the number of users we're expecting from steam - even with the small number of guards we have, we still have the occasional server admin coming to us saying some guard kicked him off his own server - we get the blame for that, not the guard in particular since the guards are presumed to be in our control.
With the number of guards we'd have to hire to cater to any steam release, the chance of this happening to someone goes up and up.

Re: "what if I want to change my vote" - Ryan/FLAB was concerned about this as well, as far as I know he's building in some way for votes to get reversed if you change your mind within 24h.

Re: "so i have to go to the forums and xyz and oh god this sounds bad because I made assumptions!" no, you just have to type "!up username" or "!down username" into the chat. The same will hopefully get ported for votekicking so you don't have to play "snipe the tiny checkbox" on the score menu anymore, and can just say "!kick username" to start a votekick.

Re: "but the kids will be using it!" sure they will, so this (gasp) fosters everyone being nice instead of just dismissing someone because they're a "wanker". You might actually have to be kind to the majority of players, instead of waving your balls in their face because they can't touch you, this is the internet, get maide scrub i fack yo madder, baide baide.
This literally gives power to the people: to stay safe, all you need to have is the majority of people you interact with to like you. Quite funny how scary that is to some people.

Re: "cost of 0.5 is too high" - lets trial the system before we try to balance the numbers, hey? It's supposed to be expensive so that there's a very distinct weight to spamming voting - if it was 0.25 that would mean you could downvote 4 people in the server every time one person upvoted you, which to me sounds way too lenient.

Bottom line is, if it doesn't work we can revert to conventional moderation and the pitfalls therein. If it does, it'll be a self balancing perfectly scalable system for moderation that lets the community determine the effective terms of service, for the betterment of their enjoyment.

Re: "Where are the photos?!" - I'm bugging MM about that now :)

 

They are volunteers. Geti meant that if we'd like to keep up with the demand of the public we'd have to start hiring people for that.

 

We've been saying for aggggggggggggggggggges that server owners need to take ownership of their own server moderation - they need to make sure they have an ample supply of local admins.
Moderation is still primarily going to be performed by moderators - though it'll specifically be moderators appointed by specific server owners. The reputation system merely allows server owners to place restrictions on who can access their server - e.g. don't allow anyone with less than -10 reputation to join. Ideally this will be tied into seclevs too (so you allow low rep players to join but apply restrictions to them or you could give anyone with more than +50 the power to kick, for example).

 

It's not only about having enough trustable players, is about who is going to watch them. Who will guard the guardians?
The Problem
On the early days of kag there were a controllable amount of players, so the devs took care of the community

But the player-base grew on, and suddenly, devs couldn't handle all the players.

Then the Guard-system were created, so the devs managed the guards, and the guards managed the community. And so far, it worked like a charm.

But as the community continues to grow and with the influx of players from steam, the current number of guards won't be able to keep up with the users.

Adding more Guards isn't an option, because the amount of devs is limited, and while the number of players rise very fast, the number of mature and well-know members is much slower and the number of devs don't rise at all. An overflow of guards wil cause more problems then solutions, since you will have a group of users with special powers that the devs can't afford to moderate.

The proposed solution is a self-moderating community, that will require few or none outside intervention to function well.

Why a player-rating system is a terrible solution
Using up-votes and down-votes as a form of moderation is flawed by design. It assumes that all the players have the same amount of judgment, honesty and maturity. It's easily exploitable, and as a system, it will behoove bad players over good ones.

So you can vote any number of people as long as you have enough people to plus you back, and you can vote them everyday?
This is absurd, similar ideas with much smaller impact went badly. To cite an example, take a look at EpicMafia (it's a website where you can play Mafia, the popular party game, online), you can only upvote or downvote someone once in a lifetime (you can change your vote though) and it gets terrible abuse, with well-know trolls and game abusers holding the highest scores and legit players getting neg bombed. Karma-whoring is also a huge thing, and even legit players have to do it if they want to keep a positive ratio.

People care less about giving kudos to someone then to hurt someone who made them mad (by, let's say, pwoning them on a RTDM server). Legit players who don't karma-whore will have problems keeping their ratings positive, since most of people wont bother with spending their ratings just to show appreciation, and a young or malicious players wont bother with a legit reason to neg them. Voting on the whole server is long term feasible, as long as you have a small group of friends to back you up everyday. Only players who don't abuse the system will have problems with having enough karma.

People can still get 100 rep, be a dick to everyone, get 100 rep again and repeat. While this isn't a bad idea, it doesn't solve the problem.

No it does not. Here is how a 8yo can do it:

Spoiler: Spoiler Hint

1 - Get your dad's credit card.
2 - Spend 110 bucks and get 11 accounts (most of console games cost more then that).
3 - During one day, be very nice with one of the accounts, get 5 people to like you.
4 - Up all your alts to +1 using this account, then use your now positive ratio to vote them all up.
5 - You now have +10 free ratings everyday. You can now down vote 2+ people up to -10 EVERYDAY.
6 - Use KAG WatchDog to stalk people you dislike. Give them -10 every day, in 3 days you can bring them from +20 to -10
7 - Repeat 6 until everyone you dislike gets banned on major servers.

If a system can be exploited by an angry 8yo on the time period of a day maybe it isn't a good system.
And that's one kid alone. Now imagine that 8yo Billy plays with his 30 classmates. Can you see the problem?

It doesn't compensates for misfire, how are devs keeping an eye on a HUGE group like a community to find zombie accounts if they can't even afford to manage a small <100 group like guards? Plus, zombie alts won't be the major concern IMO, small circlejerks will. So it doesn't matter if you cheat and grief or not, as long as you do it with your buddies you're safe.

Suuure. Let's pick a experimental ban system, and turn it on by default without any mass testing.
What could go wrong?

It will maybe slow it down for some days, but it doesn't change the fact that the power will flow from highly respected members of community to a base of new players, that won't have the same power, knowledge or skill (or even will) to handle problems.

This is, in fact, inferior to the model we have now.
If someone is griefing on a server, you can get a guard on seconds (I'm aware that this could change with the influx of users), but with a player rating system, it can take months for it to be banned on default servers. And this is not only inefficient to catch griefers, but it also makes it easy to get a legit player banned. While for a griefer is a huge profit to grief for 4 months before being banned on most servers, someone that is actually trying to play will get pissed off over a major ban after only 4 months.

Atm, if someone is griefing and my team can't/don't-know-how-to/won't vote kick him, I can record him and know that with proper evidence, he will be banned. With a player based rating system, nothing happens. Someone might alternate between playing and griefing and keep their ratio positive forever.

There is no conclusive definition of nice. Nice doesn't follow strict guidelines. Nice doesn't lose their nice privileges for abusing them. Nice is now determined by everyone that have 10 bucks. Except that whatever they determine will have impact on if you're able or not to enjoy the game. Nice isn't it?

So all you have to do is to handtype the name of the user? Oh, no problem downvoting Mr. 73@2f²h¬-w78³9w8349hg or the fine gentleman called llIlIllIIIlIIllIIIIllII (good luck finding out the L's and i's there) right?

Have they? While I agree with you that recruiting guards to match steam expectations is not possible, why get ride of a useful tool that is working well?
The time that is being spent on this "new solution", could be used to add more tools for owners to moderate their own servers. Like player-run ban lists and admin lists that you could add with one command. An API that tells which servers a player have admin on. Imagine being able to add trustable moderation to your Aussie server, for example, with a simple command like "!addlist lavalord_list". And after you type !guard on that server, a community made IRC bot could PM the admins lurking there to warn them.

Those are just examples, and some of them may be time-expensive and as flawed as the voting system, but it's just to show that there other alternatives. You could increase the amount of moderated servers just by adding an icon for when the server have at least one active admin on, and a checkbox for "Show moderated servers only". In order to get more players, the owners would try to make sure that they always have admins on, so they would hire more local admins, and griefers would avoid these servers while players would frequent them more.
And I highly disagree with these "we have a new moderation tool, time to get ride of all old moderation tools" approach.


PS:

Spoiler

Just pointing out that I have a huge respect for MM and Geti as game designers. I think, from brief talks, texts on the devlog and MM video presentations that they have a really good grip on what make multiplayer games fun. Now that I'm re-reading it, I know that some of this text may sound a bit harsh or rude, but it's not my intention to offend anyone.
I'm just explaining what I believe it's a good intentioned however poor decision, for a game we all love.

Also, if some moderator wants to put this thread back on rails, feel free to move it to a new topic.

 

Okay, some good points there. Some not-so-good ones that are based on false assumptions, but that's okay :p

First off, guards aren't going to disappear the second we implement this system. The plan has always been for a gradual phase out. Personally I don't even see the point in specifically removing them seeing as we don't recruit any more and people leave or become inactive - the phasing out is happening naturally.
Also, collecting up evidence and reporting it for a proper ban won't stop being an option. We'll still ban people when you have video evidence of them breaking the rules.

In addition, we won't be adding the default rep thresholds for joining to begin with (the default threshold will be there but it'll be disabled). They would only be defaulted to 'on' after we were happy with the system - at first only those servers that specifically opt to use it will have those limits.

We'll be tweaking and improving the system for a long time after its initial implementation. We'll likely be playing with things such as having the time before you can upvote someone double each time you do it (i.e. 1 day, then 2, then 4, then 8...) to reduce circlejerks (and possibly increase the cost too). Plus if circlejerks really do remain an insurmountable problem it would be relatively straightforward to write a program to track down cases of it.
Also, the intention is to add tab completion of names to make up/down voting easier (and I'd ideally like to see some sort of proper GUI for it).

I doubt the system will be viable straight out of the box, but that's okay - its not meant to be. Its the sort of system that will take a lot of fine tuning to get right, which we intend to do. If we can never find a sane balance that isn't horribly abusable then we'll just scrap it, but we're all hopeful that it won't come to that.

And finally... Yes I would love to add more moderation tools such as those you described. I particularly like the idea of filtering moderated servers - I'll see about adding that once I'm done with this semester of uni if no-one else has gotten around to it by then.

@ilaks: If I knew the answer I'd include a response to your post. But I'm not currently aware of what the plan is with regards to rep visibility - I've only ever discussed the system in the context of moderation.

 

@Unnamed player:
How would you go about implementing community moderation?

-"Little Johnnie can downvote me with his 110 dollars of alts!"
Sure, but you can also downvote his alts and his main, and get your friends to do the same if he's sapping your rep too much. If they're below 0 rep, they cant vote. He's going to have to expend a lot of time and effort to do any real harm to more than one person. Same goes for a circlejerk of hackers/griefers - if the community (or even just a vigilante section of it) work together, they can get them to sub-zero levels where they can no longer "jerk eachother" positive.

Our plan is for Player ratings (flat heirarchy) and Per-Server Administration (not just one or the other).
Larger servers already need to recruit administrators, and that's how it's handled in most other games we've played or observed. The Player rating system is intended to create global consequences of being a dick or a nice guy, not replace moderation on the whole.

As your diagrams show though, we can't really keep the guard system going, so we're going to need a bit more self-moderation from the community. The bigger servers generally have reliable moderators and I'm hoping/assuming that'll continue to be the trend as the number of large servers increases.

@Ilaks:
Considering you could infer your reputation by trial joining a bunch of different-threshold servers, I think it's simpler to display your reputation and "transaction log" publicly. Prevents dishonest voting and allows the public to track abuse more easily. Might cause some elitism

I'd really like a way to view the web of up and downvotes graphically and if I get time it's certainly something I'll be looking into. It'd be easy to identify circlejerks and karma farms (especially if they had a related-IP connection shown as well).

 

Just as an addendum as I seriously have to get some study, then optimisation done:

A tab complete for usernames like on irc (planned) would solve that pretty nicely, no? Yet another section in the esc menu is possible, but we all know how prone to misclicks and someone leaving the game that is.

Sure, but likes are free and anyone can assign them. That's why we went further than a simple +1 -1 free upvote downvote reddit type system.

Not sure how it affects gameplay at all?