This is the talk page for discussing the FortForecast.MagicFuckingPointSystem article.
<namespace> Obormot: So recapping what we've been discussing so far.
<namespace> quanticle: I'm not entirely sure how to introduce this, but part of the mission of FortForecast is to figure out how to actually get people to do things in some kind of coordinated way.
<namespace> And Obormot happens to have had experience in a high effectiveness group that operates entirely online.
<namespace> A WoW guild.
<quanticle> Well, I guess WoW guilds are pretty effective? I don't know, I've never played World of Warcraft.
<Obormot> I mean, some are.
<namespace> One of the best WoW guilds, to my understanding.
<Obormot> Well no
<namespace> Oh?
<Obormot> I mean
<Obormot> There were a LOT of people playing WoW.
<namespace> Oh true.
<Obormot> Why don't we say "99th percentile of players, on our server"
<namespace> That definitely works, yeah.
<Obormot> (about 25k population at its peak, as I recall)
<namespace> And so a problem you run into with WoW guild raiding is that it's not easy to divide the rewards, so you have to decide who
gets them. For which people have developed elaborate systems that have empirical battle testing.
<namespace> Obormot's guild used: http://wowwiki.wikia.com/wiki/EPGP which can be discussed later.
<feepbot> EPGP | WoWWiki | FANDOM powered by Wikia (EPGP is based on the concept of Effort Points and Gear Points. Effort Points quantify the effort each member put towards the (common) guild goals and Gear Po [snip]
<namespace> But how this relates to FortForecast.
<namespace> Is that the problem of controlling how effort should be directed is actually fairly similar to the loot distribution problem.
<namespace> You can model it something like:
<namespace> Inputs (time, work, money, etc) -> Goals -> Outputs (Information, code, etc)
<namespace> Most of the outputs of a group like this are probably easily shared and copied, so they're not really subject to this.
<namespace> But if you go up a level to the goal choosing, that's a scarce resource that can't really be infinitely divided among people.
<namespace> And this isn't a high level thing.
<namespace> This is a thing that occurs even at the low levels. Example:
<namespace> In #lesswrong, you occasionally get someone who comes in and asks something like "Will the singularity happen?"
<namespace> There's only so much group effort that can go towards question answering and discussion, and that's a really hard question to
answer. It'd take a lot of time to seriously discuss it.
<namespace> So what happens in practice is we ignore the newb and they usually get dejected and fuck off. :P
<namespace> Which is probably fine in this case.
<namespace> But it illustrates the concept, there are certain things which tax the group but benefit individual members, and other things that help the group at the cost of individual members.
<namespace> Which is actually really similar to the loot distribution problem.
<namespace> quanticle: Following so far?
<quanticle> Yeah.
<namespace> So part of what makes a group in that vein possible is the expectation of reciprocal action. If everybody keeps cooperating then you should eventually expect to be rewarded for your efforts in some way, even if it's just with some say in what projects the site should be tackling.
<namespace> Without a way to track it, then it kind of gets tracked mentally by prestige. Since of course humans have this same problem in real life and have to deal with it somehow.
<namespace> The problem is that our way of dealing with it is lossy and ambiguous and basically hackable through social adroitness.
<namespace> Since it's literally the same mechanism we use for like, everything else in society and look how that turns out. :P
<namespace> So if you can formalize a bit of it in a way that's trustworthy, you build trust and help eliminate pain/drama.
<namespace> All that said.
<namespace> The good news is that this is a problem people have already worked on in an environment where it has to actually work or they fail.
<namespace> Obormot: You want to talk about EPGP briefly or?
<Obormot> Sure
<Obormot> Basically the idea is this...
<Obormot> There are some actions/behaviors that you don't want your members to engage in at all; those you ban straight up, punish,
whatever
<Obormot> Those aside, though
<Obormot> There are two categories of things that aren't discouraged
<Obormot> First there are things that you want everyone to be doing as much as possible
<Obormot> That are unboundedly good
<Obormot> Second, there are things which are good for people to be doing, healthy, expected, certainly not discouraged, BUT
<Obormot> you don't want anyone doing them TOO much, and you don't want there to be a serious imbalance in who's doing those things
<Obormot> The second category are going to be things which people just want to do, of their own accord, and don't really need to be incentivized to do; they're their own incentives
<Obormot> The first category are things that you do generally need to incentivize people to do, even if people "want" (or want to want) to do them
<Obormot> (in WoW, the first category is "help the raid kill bosses" and the second category is "get gear, thus making your character more badass", basically)
<Obormot> The idea of the EPGP distribution system is that you use the first category to rate-limit the second.
<Obormot> Each member has two quantities associated with them: EP (effort points) and GP (gear points)
<Obormot> Both start at 0
<Obormot> Each goes up as a consequence of actions/behaviors in that category
<Obormot> Do unboundedly-good prosocial thing? EP goes up
<Obormot> Do self-incentivizing indirectly-prosocial self-benefiting thing? GP goes up
<Obormot> Whenever there is any scarce resource that people want
<Obormot> It is allocated according to EP/GP ratio
<Obormot> Whoever has the highest such ratio gets the resource (and their GP goes up)
<Obormot> So the more EP-generating things you do, the higher your priority in such allocation; the more GP-generating things you are
allocated, the lower your priority subsequently
<Obormot> So as long as you can define those categories, and place relevant behaviors/actions into them, EPGP works to allocate your scarce resources and incentivize members' contributions.
<Obormot> Ancillary benefits include:
<Obormot> New members immediately get the instant gratification of being top priority for resources, as soon as they contribute anything whatsoever (as EP and GP start at 0, any contribution makes EP positive, and positive/zero = infinite priority!)
<Obormot> Having now given and gotten something of value, they are drawn in, at which point their priority goes down to below that of regulars/veterans; it fluctuates greatly at first, then stabilizes
<Obormot> This incentivizes early contribution, but doesn't make people "pay dues" excessively to get anything at all
<Obormot> Because EP can be assigned for anything, and relative values set to whatever the administration wishes, the system makes it easy to design incentive structures that encourage whatever you like
<Obormot> There is tangible benefit to sustained contribution, without locking out newcomers
<Obormot> I think that's sufficient summary for now
<namespace> Yeah.
* namespace claps
<namespace> quanticle: But so, I think this sort of thing is the vein we should start digging for gold at.
<Obormot> (maybe that was a little less "briefly" than intended, sorry, lol)
<namespace> Yeah it went a bit longer than I was expecting but that's fine.
<namespace> Important to describe the concept.
<namespace> More than it is to be brief.
<namespace> As far as I know nobody has tried it in this context, it sounds like it will plausibly work (which is way way more than I can say about...almost any other proposal I've heard in this vein), it avoids a lot of issues like the elite insular community that can't get new blood because new blood has no investment, etc.
<namespace> It's flexible.
<namespace> And it's empirically validated in a *similar* context.
<namespace> quanticle: Thoughts, concerns, questions?
<quanticle> What are some examples of things that generate EP vs. things that generate GP. I've not played World of Warcraft, but I have numerous friends who have, so I'm well aware of the classes and terminology.
<quanticle> So, on a raid, if I'm a cleric and I heal the tank...
<quanticle> Is that EP
<quanticle> GP?
<Obormot> That would be slicing things entirely too thinly
<Obormot> No, usually, it's much simpler than that
<Obormot> Show up to the raid? Have a bit of EP
<Obormot> The raid downs a boss? Everyone gets a chunk of EP
<Obormot> You personally farm/craft/etc. and contribute some consumables for raid use? Some EP for that
<quanticle> Okay. And what would generate GP?
<Obormot> It stands for "Gear Points", so, getting shit
<Obormot> You got a piece of loot? GP goes up
<Obormot> That's basically it.
<Obormot> Like, the domain here is
<Obormot> "limited resource, that is 'property' of the raid as a whole, and must be allocated to individual members of it"
<quanticle> Oh, so if I get loot as a result of the guild defeating the raid boss
<quanticle> I get GP
<Obormot> Exactly.
<Obormot> And how do we know who gets loot?
<Obormot> Whoever a) wants it, b) has the highest EP/GP ratio.
<quanticle> Actually, even this has been enlightening. The Onyxia raid video makes a lot more sense now.
<Obormot> lol
* quanticle now understands the point of DKP
<Obormot> Well they used DKP!
<Obormot> DKP is a different system
<quanticle> Oh
<Obormot> DKP is the *first* loot system.
<quanticle> Where there wasn't an EP/GP distinction?
<quanticle> You just got points?
<Obormot> You got points and you spent points
<Obormot> (usually by bidding)
<quanticle> Ah, okay.
<quanticle> So whoever had the most points would win
<Obormot> Yes
<Obormot> Well, wanted to spend the most, really
<Obormot> So DKP had many faults
<Obormot> After a certain point, no one serious used it
<Obormot> Because better systems had come along (of which EPGP was the best - imo, anyway - at least for persistent raid groups, as
opposed to one-off raids or w/e)
<Obormot> With DKP, if you were a newcomer, joining a raid full of veterans
<Obormot> The only way you were going to get anything was if no one else wanted it
<Obormot> Otherwise you had to sit there, raid after raid, contributing effort but knowing in advance that you ain't getting shit (except scraps from the veterans' table, as it were)
<Obormot> On the one hand, this is a reasonably accurate model of how resource allocation works in other contexts
<Obormot> On the other hand, it's obviously not very conductive to recruitment
<Obormot> There's the twin needs, in any group that depends, for success, on a bunch of people all contributing as much effort as possible:
<Obormot> 1. You gotta pull in good people
<Obormot> 2. You gotta get your people to stay, and keep contributing
<Obormot> DKP was bad at #1 because the prospect of slaving away for weeks or months before you had accumulated enough to have a shot at the good stuff was daunting (and then you could be outbid by a veteran who'd been hoarding his points for longer; and even if you won, you might've just spent all your points on one thing; etc.)
<Obormot> DKP was also bad at #2, because after you had accumulated a certain large pool of points, the incentive to keep contributing dropped off
<quanticle> Right. Why bother doing things when you can just show up, spend your DKP, and get choice items from the bank?
<Obormot> EPGP, on the other hand, is good at #1, because your first reward is basically guaranteed, as soon as you contribute something of value
<Obormot> And, EPGP is good at #2, because going up in priority is easy at the start, and bouncing back from getting some gear is easy, and as it gets harder, well, ratios equalize; and as long as you *keep contributing*, you stay at a good ratio, and meanwhile, the higher your EP gets (if you're a veteran), the faster it drops when you get something
<Obormot> (also there's the decay feature, which helps even more)
* kuudes has quit (Ping timeout: 246 seconds)
<Obormot> So, insofar as it can be applied to another context, we may with *some* justification expect these desirable qualities to transfer
<namespace> Sure. And I wouldn't want to do a straight transplant.
<namespace> It can be *inspiration*, but you're so not going to just copy this straight up and get good results.
<Obormot> (The analogue of DKP, in the "project-oriented community" scenario, of course, is the natural reputation system)
<namespace> It's not the same thing, it's just similar enough that a solution that *looks kind of like EPGP* might work.
<Obormot> (Where it's hard to build up social standing, but eventually you have so much that you barely have to contribute to get your way, etc.)
<Obormot> (Of course the analogy doesn't stretch *too* far, but it's an interesting parallel imo)
<namespace> I mean, DKP does in fact have fairly basic incentives similar to the natural reputation system yeah.
<namespace> In terms of the outcomes at least.
<namespace> quanticle: So here's the really hard part for a system like this.
<namespace> In the context of FortForecast, you have to award the EP.
<namespace> And that requires human judgment to really be sane.
<namespace> Which means you need to have clear guidelines about what is a fair amount of EP for a given amount of work.
<namespace> Otherwise it just becomes de facto reputation system.
<namespace> Or *worse*.
<Obormot> I would say that the biggest wins would be regularizing/systematizing EP awards
<Obormot> wins in terms of system predictability and functioning-as-intended
<quanticle> So, to step back a level, I think the main problem is that in the context of FortForecast, EP and GP don't *do* anything. The
whole reason EP/GP works is that you can use it to get loot.
<namespace> No.
<namespace> The core idea, is that you give people limited community attention/resources in a way that isn't authoritarian but does help
people gain legitimacy and assistance in their endeavors.
<Obormot> They do the things we make them do
<namespace> ^
<namespace> That's the other hard part, IMO.
<Obormot> Like I said: whatever member behavior or actions we decide have the property of "it is good, expected, and healthy for our members to be doing this; but we don't want anyone doing it too much, and we don't want there to be an imbalance in who does it" - those we associate with GP
<namespace> Is balancing that in a way which is fair without being heavy handed or bossy.
<Obormot> ("associate with" meaning, here, allocate by EP/GP ratio priority)
<quanticle> Limited community attention/resources. Okay, that's fine and all, but what attention or resources do we have to give?
<namespace> The million dollar question.
<namespace> Right there.
<Obormot> Such actions/behaviors can be anything from "a downvote" to "post a question" to "post a project proposal"
<namespace> IMO that and how to award EP are the things that make this succeed or fail.
<namespace> But yes in the abstract what Obormot is saying.
<Obormot> There are several potential pitfalls which I think now's a good time to comment on (also because I might forget later)
<Obormot> One way for the system to fail is to forget to close feedback loops
<Obormot> For instance, if proposing a project increases GP, why would people use the system to propose projects instead of just
communicating with potential collaborators out of band?
<Obormot> That would be a failure of the system
<Obormot> One solution to which is "because only officially defined projects reward EP for participation and for results" - closing the loop
<Obormot> (There are a number of such cases, not going to try to enumerate them all)
<Obormot> Another pitfall is perverse incentives... this is a broad category, but what I have in mind specifically is the notion of having the numbers, and also user behavior, go out of whack because we're incentivizing something that isn't really an unbounded good
<Obormot> Like what if there's a board where people can ask questions, and posting a comment marked as an "answer" *automatically* awards EP? That would be bad
<Obormot> Similarly, another one that namespace pointed out the other day: perhaps starting a thread adds GP [side note: all examples speculative for now, not indicative of final design at all]. Certainly we don't want anyone spamming threads, etc. But does this discourage asking good questions or posting threads that are useful to everyone?
<Obormot> Yes it does
<Obormot> Solution? Starting a thread increases GP, but starting a *good* thread then also awards EP
<namespace> But how? ^_^
<Obormot> (remember, EP can be awarded for anything)
<namespace> (This is still the open question for me on that issue, IMO.)
<Obormot> how what
<namespace> How do you evaluate every thread to determine EP award?
<namespace> That could get to be a lot of work really quickly.
<namespace> We want to do things that scale relatively well.
<Obormot> Crowdsource it and award EP for contributing to that
<Obormot> And/or
<Obormot> Set up a crowdsourced moderator prefiltering system
<namespace> That was the line I was thinking along yeah, just my problem with *that* solution is now you're starting to get into some dangerous legibility issues.
<Obormot> Award EP for that
<namespace> But I guess those can be dealt with at the crossing.
<Obormot> Have a "report dangerous pattern" button
<namespace> Oooh.
<namespace> What if we had like.
<namespace> A public EP/GP viewer? Hm.
<namespace> Are all EP generating actions intrinsically public knowledge?
<quanticle> namespace, Obormot: Have you looked at StackOverflow's system? Could we do something like that with EP/GP?
<namespace> Not that much.
<namespace> By which I mean.
<Obormot> I don't have a great opinion of SO's sytem
<namespace> I know it's up/downvote-y.
<Obormot> *system
<namespace> But nothing beyond that.
<Obormot> namespace: To answer your question, yes, canonically everything ever is public knowledge, and I think it really has to be that
way
<quanticle> namespace: Yes, but downvotes cost points, and as you gain more points, you unlock additional powers (like the ability to moderate, or edit other people's stuff)
<Obormot> I was just about to say
<Obormot> That that's bad
<namespace> Yeah...
<namespace> IMO.
<Obormot> And exactly the pitfall I was going to mention
<Obormot> Like
<namespace> There's like.
<Obormot> You don't ever want more power to be unlocked by EP or by EP/GP ratio,
* namespace thinks of how to put this
<namespace> +1
<namespace> That, but also a more general principle related to it.
<Obormot> You don't ever want the numbers to usurp, even a little, human control of the system
<namespace> ^
<Obormot> That's got to be firewalled
<namespace> Well I also think there's a type issue there.
<namespace> Mod powers are *not* awarded based on effort.
<namespace> They're awarded based on judgment/social status.
<Obormot> Indeed
<Obormot> You can have whatever system you like for allocating *control*, for allocating moderation or administrative power, etc., but that system can't be the same as the one that allocates resources
<namespace> :3
<namespace> Because as we all know.
<Obormot> Otherwise you've enslaved yourself to the system and it no longer serves you
<namespace> Otherwise the mods allocate themselves the resources.
<Obormot> Yep, that too
<Obormot> You instantly create a whole host of perverse incentives
<Obormot> Which defeats the whole point of the thing
<Obormot> That, btw, is the other reason why 100% transparency is, imo, a requirement in EPGP or similar
<Obormot> You just can't leave any room a) for anyone to behave unethically w.r.t. the system, b) for anyone to suspect (without it being
easily publicly verifiable/disprovable) unethical behavior
<namespace> Right so that does create one issue. How do you handle a project not everyone can know about? Do you just have it internally in that case to the project members?
<Obormot> So yes, public logs of all point movement
<Obormot> I mean
<Obormot> The first answer to that is "maybe reconsider the notion of having projects that not everyone can know about".
<namespace> :3
<Obormot> No, seriously
<Obormot> I'm not trying to be a dogmatic radical-transparentist here, but
<Obormot> It really does merit serious consideration
<Obormot> Whether that's a thing we want to have
<namespace> Oh sure.
<Obormot> I think that in those cases where there is a need for secret *projects*, it should - if it's a non-pathological scenario -
involve secret *groups*, for which then the obvious remedy is to have an entirely separate membership/participation/resource/incentive structure, perhaps with an actual separate deployment of the software
<namespace> Right, fair enough that's what I had in mind.
<namespace> I just wanted to be sure we could assume all EP/GP is public and auditable by any member.
<Obormot> Yeah.
<namespace> And thus there can be complex views to help you examine them and report the dangerous patterns.
<Obormot> Yep, definitely
* kuudes (~kuudes@unaffiliated/kuudes) has joined
<Obormot> (secrecy per se aside, there was an analogue to this in WoW as well, where, as I think I've mentioned, there often was such a thing as "within this raid guild, there are multiple raid groups, including some that are more 'elite'/exclusive than the main one"; such groups usually did *not* use the EPGP or other allocation system of the main group, but had their own thing)
<Obormot> (could there be an analogue to this in our thing? perhaps)
<Obormot> (I should note that such smaller, more elite/exclusive groups, typically skewed closer to "managed communism" than to "regulated capitalism" on the spectrum of loot systems, which I do not think is a coincidence)
<namespace> Fewer people, higher internal trust presumably.
<Obormot> "higher internal trust" is true, but not where I'd locate the cause
<Obormot> I'd say "higher degree of sublimation of personal interest to group interest"
<namespace> Ah.
<namespace> More dedicated?
<Obormot> Yes, and more willing to sacrifice for the good of the raid
<Obormot> Like, if you're trying to maintain a raiding guild of 100 people, keep it functioning and healthy over the course of months or years, new content, people joining and leaving, schedules and life circumstances changing, different personalities and background, etc.
<Obormot> Then it's important to maintain member satisfaction; it's important to ensure that people feel in control and rewarded and appreciated; that they don't burn out or develop resentments; that no one feels slighted, and no one feels that anyone is favored; you have to recruit, also...
<Obormot> All of these things are more important than *being maximally effective at downing this boss right now and then the next five bosses this week*
<Obormot> If you focus on the latter and ignore the former, your guild will break and explode, and people on WoW-related news websites will place stories about your public meltdowns in the Drama section, and laugh at you
<Obormot> On the other hand
<Obormot> If you get 10 guys together and you go "ok dudes, we, these particular 10 people, are going to show up every single Sunday for several months, play for 6 hours straight each time, and we will push through absolutely the most challenging content in the game, which only a small handful [or sometimes: none at all] of people in the world have done"
<Obormot> That is a different scenario
<Obormot> There's no room for "I'm not the tank but I want that piece of tank gear", because if you do that you will fail
<Obormot> What a group like that *promises* - which a larger, more skill-diverse, less elite/exclusive, group cannot promise - is the incredible rush of pushing yourself - your concentration, your skill, your endurance, your coordination, your ingenuity - to the maximum, and *succeeding at something really really hard* as a result
<Obormot> That is the intrinsic motivation which takes the place of the extrinsic motivation of getting loot
<Obormot> As a result, the extrinsic motivation is no longer a resource which it is vitally important to allocate
<Obormot> In that scenario, your needs are the group's needs; the group's successes are your successes; there is no separation between you and the group, and consequently the need for equity in loot allocation falls away, and everything is allocated strictly by group-level optimization
* namespace nodnods
<Obormot> (And this shows in the reactions people have to other people getting loot. In a larger, somewhat-more-casual raid group, it's like: Alice gets <awesome piece of gear>. Alice: yay! :D Bob: grats [Bob is happy for Alice but also jealous, Bob wanted that thingie too]. In tighter-knit, more-hardcore groups, it's like: Alice gets <awesome piece of gear>. Who raid: FUCK YEAH!! :D [because every thing that anyone in the raid
<Obormot> gets, is increased performance *for the group as a whole*, which is all that matters])
<Obormot> *Whole raid
<Obormot> Of course, that sort of thing doesn't scale, and neither can it last, just as you cannot build a whole country like a kibbutz
<namespace> Yup.
<Obormot> But it may be entirely possible, and perfectly healthy, to occasionally cleave off subgroups who follow that model, then to meld back into the overgroup at the completion of a project (and never having really separated from it, their members continuing to participate in the overgroup even as they throw themselves into the subproject)
<Obormot> (so to refer back to a previous comment: the raid *guild* I was in was ~99th percentile, but the smaller raid *group* I was part of was much, much further toward the right tail of the raid content achievement distribution)
