Consider two people living in an apartment; they are our old friends person A and person B. Person A is bothered when the trash gets to be 5 days old, and Person B is bothered when the trash gets to be 7 days old.
Something that sometimes happens in this situation is that person A gets bothered by the trash every 5 days and takes it out. Person B never feels that the trash needs to be taken out, and so never does. This is unfair to A, but B reasonably opines that she would be happy to take out the trash, but only on her own timeframe.
What to do? One solution is for person A to make person B take out the trash every other 5 days. The most common solution is for both to come to some sort of consensus (every 6 days?).
A PassiveAggressive solution is for person A occasionally to wait until person B voluntarily takes a turn. Thus, person A can "control" the situation without any discussion, consensus, or "making" B do anything. If person A takes the trash out every odd week, and then lets it sit on the even weeks until B takes it out, then the average time the trash is sitting is 6 days, and person A is only doing half the work.
Application to wikis
There are tasks to be done on wikis which are analogous. The situation is more complicated because:
- there are 30 people instead of 2
- most of the tasks aren't suited to being scheduled or clearly assigned (imagine: "ok, you welcome the first newcomer, then he'll welcome the next one, then I'll welcome the third one, then she'll...")
- forcing the community to come to a clear, detailed consensus about these issues would likely be a never-ending source of argument ("how about we refactor every 3 weeks instead of 2?" "since i'm not a member, i don't want to welcome this person", etc)
Discussion
So, I recommend that people adopt the last option in the first section. When you feel like you are doing more than your fair share of the work, cut back unilaterally. Although I do think you should mention it so as to clue everyone else in.
For example, I think SunirShah feels that he does more than his fair share of chores around here. The solution is for him to just unilaterally not do as much, and to mention what it is that he'd like others to do more of. He is worried that if he doesn't do these maintanence items, MeatballWiki will collapse around him like an abandoned building. I think a lot of the time, either:
- Sunir is doing something that others would do, but maybe not as quickly (i.e. delete off-topic pages posted by some newcomer)
- Sunir is doing something that others aren't sure needs to be done
-- BayleShanks
If Sunir feels that we aren't being diligent enough, this must be addressed directly rather than addressed by him picking up the slack. While it's true that by showing leadership (by doing more himself) Sunir might convince us to do more (ModelDesiredBehavior), ultimately housekeeping is more than a one-man job, and he will run out of time if he sets higher housekeeping standards than the rest of the community, and stays true to his high standards. Doing more himself is not an option even if Sunir thinks the rest of us are too lenient. -- BayleShanks
So, I started thinking about this in terms of utility theory. I tried to quantify some of the utility values for the different agents, or get a start. For agent A:
- The utility of leaving the trash around is 0 if it's been around for less than 5 days (no biggie)
- The utility of leaving the trash around is -15 if it's been around for 5 days or more (it stinks, it overflows, etc.)
For agent B:
- The utility of leaving the trash around is 0 if it's been around for less than 7 days (no biggie)
- The utility of leaving the trash around is -15 if it's been around for 7 days or more. (it stinks, it overflows, etc.)
The step-function utility is simplistic -- probably a linear or exponential function would be better, but we can leave it as it is.
For both agents:
- The utility of taking out the trash is -10 (it's a dirty and uncomfortable chore)
Notice that it becomes worth it to take out the trash after 5/7 days, respectively.
It'd be pretty straightforward, if these guys lived alone. However, they live together, and there's a relationship factor that's kind of complicated. The best I could come up with is:
- The utility of having your roommate's utility lowered in any way is -3 (they complain, they get angry, they stomp around, etc.)
In other words, it costs B some utility each time A takes out the trash, and it costs B some utility each time A takes out the trash. It also costs B some utility when A doesn't like having the trash around, and vice-versa. There's probably some better way to couple utility between agents -- I'm not wise enough to know it. All the utility values are rough; I just pulled them out of my ass.
Anyways, I wonder if there's a way to calculate A's and B's respective strategies for dealing with the situation, based on these utility values. Again, I'm kinda dumb about this stuff. But here's my analysis:
On days 0-4, nobody takes out the trash. It's not worth the effort (-10).
Now, on day 5, A should take out the trash. That costs him -10, and B -3. If A doesn't take out the trash, and B does, that sets A's utility at -3, and B's at -10. If B doesn't take out the trash either, A is at -15, and B is at -3. So for B, it doesn't make sense to take out the trash, and for A, it does. So A takes out the trash.
For the short term, this is optimal for everyone. But in the long term, it's not an optimal strategy for B. After 4 cycles, for example, A has -40 and B has -12. If B had taken taken out the trash one of those times, he'd have utility -10, and A would have -33 -- better for both, but notably better for B. If he'd taken it out twice, he'd have -20, and A would have -26 -- better for A, but worse for B than not taking it out at all. So, I think the optimum strategy for A is to take out the trash on day 5 if B doesn't take it out, and for B, it's to take out the trash on day 5 once every four cycles (20 days).
Note that this isn't particularly "fair" for A, but that's not B's problem. A consistently has lower utility -- the price of being a neat freak.
The N in the "once every N times" strategy for B depends on the relative utility of taking out the trash versus having your roommate lose utility. If the utility of getting your roommate mad were, say -6 or higher, then B would need take out the trash every other time. The whole thing revolves around how much it sucks to have your roommate be angry.
Expanding this analysis to 30+ agents is left as an exercise for the reader. My guess is that we should all let the neat freaks do as much work as they want, and throw in some effort whenever it seems they're going to get mad. --EvanProdromou
While a very good analysis, that assumes that the perceived utilities are constants, which is false since they are merely perceptions. Consequently, one can also change the perceived utilities so they are closer in line with each other. This is done by internally altering one's own utility as well as convincing others to change theirs. Easier done than said. RewardReputation as required.
I agree that it's a simplistic analysis. One thing that's not in this analysis, and that probably should be, is the utility of moving out. For an apartment, I'd probably put it around -2000 -- it's a pain in the ass to move out, at least 200 times harder. But it does factor in.
For a Wiki, however, the utility cost of moving out is much, much lower. Especially for drifters and casual members -- the people who don't care about the trash building up, anyways. Those who actually take out the trash usually have more invested in sticking around. --EvanProdromou