The organization I know of that’s most similar-in-shape to many existing LW meetups, that seems to do a good job of scaling the tribal leadership thing in a systematic way, is Toastmasters, which I think people have talked about here before. The structure is such that there are multiple roles to give people as they get involved, designed to scale as a local group grows (and eventually bud off into a new group).
(Though I don’t have the sense that Toastmasters quite fosters what I think Qiaochu was seeing at the CFAR mentorship workshop.)
In that case, here are some more thoughts on World of Warcraft as a case study in “leadership”:
(1) The “amount of leadership” required for a group to succeed is roughly proportional both to the size of the group, and to the difficulty of the challenge.
Corollaries:
Smaller groups require less “leadership”, and are self-organizing to a large degree.
This is partly, though not entirely, because challenges appropriate for smaller groups are (ceteris paribus) correspondingly less complex (they have fewer moving parts—because the cognitive capacity of individual group members does not scale with group size). This is an artifact of the game’s designed nature, but it does (IMO) map to certain real-world properties of projects, though imperfectly.
However, much of the “extra leadership” (i.e., effort expended on leading × competence of leader) required to effectively lead a larger group is required due to social/coordination issues (which are of the greatest variety, ranging from issues of communication and timing during execution of the collective effort, to scheduling, to the acquisition of necessary resources, to differences in communication/cognitive styles, to clashes of personality, etc.).
A fixed “quantity” of leadership (by which, again, I mean effort × competence) suffices to overcome challenges of greater difficulty in a small-group context than in a large-group context.
This also means that a smaller group with more resources, or some other “effectiveness booster”, requires “less leadership” to accomplish a given objective than a larger group does to accomplish the same objective.
(2) A critical part of what “leadership” involves is making decisions concerning the apportioning of rewards from successful task completion.
Corollaries:
The self-organizing nature of groups (even those composed entirely of very competent individuals) tapers off much more quickly (as group size increases, and as task difficulty increases) when it comes to reward distribution than when it comes to task execution. (In other words, even in cases where only minimal “leadership” is required to successfully complete the task, apportioning the rewards by spontaneous consensus is almost impossible to achieve, and quite perilous to try.)
The reason for this is as follows. Even when there are multiple possible approaches or strategies to completing the task, there is nonetheless always a clear way to evaluate those approaches, and thus there is a strong sense in which there are right and wrong ways to do it—so competent/experienced group members can, and will, to a large degree, simply do what needs to be done, without having to be instructed to do it, or having to have their job explained to them. In contrast, there is no objectively correct way to apportion rewards. (This is because the interests of group members are aligned when it comes to task execution—everyone wants to make sure the task is completed successfully, as it is a prerequisite to anyone receiving any rewards[1]—whereas, when it comes to apportioning rewards, the interests of group members are either imperfectly aligned, or not aligned at all.[2])
Puzzle:
Suppose a group performs some task which is not a one-off, but iterated (or performs tasks sufficiently similar to some previous tasks). Practice makes perfect (in various ways which needn’t be enumerated here), and thus the “amount of leadership” required to complete the task will decrease over time. Will the “amount of leadership” required to apportion rewards also decrease over time? Why or why not?
[1] Careful readers may note that this may not always be true (either in WoW or in reality). Puzzling out the consequences of this condition failing to hold is left as an exercise for the reader.
[2] What circumstances determine whether the interests of group members are aligned at all, in the context of apportioning of rewards? Two conditions must hold. First, the group must not be a “one-off” coming-together of disparate individuals, but must be a persistent institution composed of a largely fixed set of members, who expect to cooperate on tasks in the future. Second, the rewards being apportioned must affect group members’ expected performance on future cooperative tasks to be attempted by this persistent group. In this case, it will be in the interest of each group member that not only he, but all group members, get their “fair share” (where “fair”, however, is defined not according to some general principles of justice, but rather means “whatever will most effectively guarantee a member’s continued participation and optimum performance in future cooperative endeavors”).
First, this is a generally excellent comment, esp. with puzzle at the end. I think it’d make for a pretty good post in a revamped sequence of running meetups.
Hmm. This is an interesting answer… I can’t quite say it’s wrong, because there are a whole lot of assumptions and frameworks lurking behind it, some of which I recognize only dimly. However, I do recognize some of what I see clearly enough to say that in my experience, running “task groups” (like WoW raid groups/guilds) in this way is… unstable, and not really reliably workable. (The reasons for this are related to the “corrective” vs. “selective” dichotomy which I have obliquely mentioned in the past, and also parallel certain real-world political dichotomies.)
I don’t know how much further it’s appropriate to take this comment thread, as we’ve sort of digressed from the OP, but this particular subtopic (i.e., what I say in the preceding paragraph, as well as your answer to the puzzle) is, IMO, an important and potentially quite fruitful one to explore.
But all of that aside, this part is definitely correct:
Bs pbhefr, guvf nffhzrf gur yrnqre(f) qvq rfgnoyvfu nalguvat fgnaqneqvmrq (vs ebgngvat yrnqref punatr gur erjneq fgehpgher nyy gur gvzr, vg zvtug rira orpbzr uneqre)
but this particular subtopic (i.e., what I say in the preceding paragraph, as well as your answer to the puzzle) is, IMO, an important and potentially quite fruitful one to explore.
I think I’m fine with the conversation continuing, but preferably in a way that ties it back to the original context (i.e. meetups and local rationalist communities), if that’s possible. (That said, as noted, I think these comments could be reworked into a top level post).
I’m not actually sure what you mean by “running a group in this way”, since I don’t think I really specified a way. Any kind of doling out awards seems like it’d congeal into a status quo if you did it roughly the same way each time.
I think I’m fine with the conversation continuing, but preferably in a way that ties it back to the original context (i.e. meetups and local rationalist communities), if that’s possible.
It certainly is possible, and in fact I’ve already had a few discussions (outside of LessWrong) of precisely this application of what I’ve learned from my WoW experiences.
I’m not actually sure what you mean by “running a group in this way”, since I don’t think I really specified a way.
Indeed, you didn’t, which is why it’s tricky to tease out the (apparent-to-me) divergence in our views. It’s not that you specified a way of running a group, but rather that what you said encodes certain assumptions about how a group will, necessarily, be run.
That’s cryptic, I know; I’d like to make my meaning very explicit, but doing so will require a diversion into some more actually-talking-about-WoW, and also many more words. I actually started writing out my answer as a comment, but it got very long very quickly. So I think I’m going to take your suggestion, and try to write this up as a top-level post or two (at which point, I hope, we’ll either come to agree, or at least our disagreement will be clarified completely).
In some cases, “leadership” also consists, in part, of being a Schelling point at which key individuals come together to collectively undertake a difficult task.
Some tasks are so challenging that they require the participation of multiple very competent individuals—all of whom, however, have other things that they could be doing with their time. Such a task may involve substantial investment on the part of each member (which may be simply an investment of time, or it may also require an investment of material resources of some sort).
It would be wholly intolerable, in such a case, for an individual, who is qualified to participate in an attack on such a difficult objective, to lend his participation, to expend his resources—only to see that expenditure wasted, on account of some or all of the other participants lacking a similar level of competence. Similarly, such an individual has little interest, no doubt, in personally going around and attempting to round up other qualified folks, to enlist them in a collective attack on the objective—he has better things to do with his time.
The leader, in this case, is one who can, and does, say: “Alice, come do a 45-minute Baron run with us.” “Do you have a good group,” Alice asks—meaning, of course, “can you promise me that the three other members of the group that you assemble will be at least as competent as you and I are?” The good leader is the one who can credibly answer “Yes” to that question. She (the leader) then goes to Bob, and invites him, and fields the same question, and gives the same answer, and then Carol, and Dave, and now leader+Alice+Bob+Carol+Dave go and do the thing.
Thus the leader manages to put together an unusually effective task group, by virtue of being able to bring together several unusually competent (and suitable) individuals.
This requires some (very basic) “project management” (i.e. scheduling, etc.) skills, and some (slightly less basic) communication/persuasion/personnel-management skills, but most importantly, it requires personal competence and the ability to judge competence in others. The latter is absolutely critical, because when the leader says “I have [or can get] a good group”, each prospective group member must be able to trust that the leader can convince each other prospective group member to participate—which is perhaps not trivial, but more or less surmountable—and that the leader can properly select prospective group members. Note that a failure of the latter skill is far more catastrophic than a failure of the former! If the leader can’t get prospective group members to participate, well—hopes are dashed, a bit of preparation time is wasted, perhaps opportunities to do other things have been unnecessarily turned down, but… it’s not the end of the world. If the leader has misjudged the competence of the group members, then the project proceeds and fails, which is a much greater waste, much more frustrating, etc.
A leader who has this skill set—that of bringing together unusually competent individuals (all of whom are in high demand) to successfully tackle an unusually challenging task—can rack up a very impressive record of accomplishing many very difficult things. It is thus a much more valuable skill set than it might, at first glance, appear to be.
Suppose a group performs some task which is not a one-off, but iterated (or performs tasks sufficiently similar to some previous tasks). Practice makes perfect (in various ways which needn’t be enumerated here), and thus the “amount of leadership” required to complete the task will decrease over time. Will the “amount of leadership” required to apportion rewards also decrease over time? Why or why not?
In larger groups, the leader must devote more effort to leading than to participating in the task execution per se.
This is necessary because the “amount of leadership” required for the successful completion of large-group tasks is greater than that required for the succesful completion of small-group tasks.
This is viable because larger groups can more easily “pick up the slack” from a leader who must devote much of his effort to leading (compared to small groups, where everyone must contribute substantial effort to task execution, else the group fails to achieve its goal).
Nevertheless, when the tasks a large group faces become more challenging, and thus require more of group members in order to ensure success, it becomes increasingly important that the leadership role be taken up by someone whose contribution to the group effort aside from leading it, is least important (relative to the contributions of other group members).
(Delegation of leadership responsibilities can alleviate this somewhat, but ultimately it only delays this effect; at some point in the task difficulty curve, these concerns return in full force despite any amount of delegation—which, in any case, will also be necessary.)
(None of these insights are new, of course—these points have been noted before, in the context of, e.g., startups. It is, in any case, interesting to see them arise in this very different context.)
Huh, this is a pretty neat point.
The organization I know of that’s most similar-in-shape to many existing LW meetups, that seems to do a good job of scaling the tribal leadership thing in a systematic way, is Toastmasters, which I think people have talked about here before. The structure is such that there are multiple roles to give people as they get involved, designed to scale as a local group grows (and eventually bud off into a new group).
(Though I don’t have the sense that Toastmasters quite fosters what I think Qiaochu was seeing at the CFAR mentorship workshop.)
In that case, here are some more thoughts on World of Warcraft as a case study in “leadership”:
(1) The “amount of leadership” required for a group to succeed is roughly proportional both to the size of the group, and to the difficulty of the challenge.
Corollaries:
Smaller groups require less “leadership”, and are self-organizing to a large degree.
This is partly, though not entirely, because challenges appropriate for smaller groups are (ceteris paribus) correspondingly less complex (they have fewer moving parts—because the cognitive capacity of individual group members does not scale with group size). This is an artifact of the game’s designed nature, but it does (IMO) map to certain real-world properties of projects, though imperfectly.
However, much of the “extra leadership” (i.e., effort expended on leading × competence of leader) required to effectively lead a larger group is required due to social/coordination issues (which are of the greatest variety, ranging from issues of communication and timing during execution of the collective effort, to scheduling, to the acquisition of necessary resources, to differences in communication/cognitive styles, to clashes of personality, etc.).
A fixed “quantity” of leadership (by which, again, I mean effort × competence) suffices to overcome challenges of greater difficulty in a small-group context than in a large-group context.
This also means that a smaller group with more resources, or some other “effectiveness booster”, requires “less leadership” to accomplish a given objective than a larger group does to accomplish the same objective.
(2) A critical part of what “leadership” involves is making decisions concerning the apportioning of rewards from successful task completion.
Corollaries:
The self-organizing nature of groups (even those composed entirely of very competent individuals) tapers off much more quickly (as group size increases, and as task difficulty increases) when it comes to reward distribution than when it comes to task execution. (In other words, even in cases where only minimal “leadership” is required to successfully complete the task, apportioning the rewards by spontaneous consensus is almost impossible to achieve, and quite perilous to try.)
The reason for this is as follows. Even when there are multiple possible approaches or strategies to completing the task, there is nonetheless always a clear way to evaluate those approaches, and thus there is a strong sense in which there are right and wrong ways to do it—so competent/experienced group members can, and will, to a large degree, simply do what needs to be done, without having to be instructed to do it, or having to have their job explained to them. In contrast, there is no objectively correct way to apportion rewards. (This is because the interests of group members are aligned when it comes to task execution—everyone wants to make sure the task is completed successfully, as it is a prerequisite to anyone receiving any rewards[1]—whereas, when it comes to apportioning rewards, the interests of group members are either imperfectly aligned, or not aligned at all.[2])
Puzzle:
Suppose a group performs some task which is not a one-off, but iterated (or performs tasks sufficiently similar to some previous tasks). Practice makes perfect (in various ways which needn’t be enumerated here), and thus the “amount of leadership” required to complete the task will decrease over time. Will the “amount of leadership” required to apportion rewards also decrease over time? Why or why not?
[1] Careful readers may note that this may not always be true (either in WoW or in reality). Puzzling out the consequences of this condition failing to hold is left as an exercise for the reader.
[2] What circumstances determine whether the interests of group members are aligned at all, in the context of apportioning of rewards? Two conditions must hold. First, the group must not be a “one-off” coming-together of disparate individuals, but must be a persistent institution composed of a largely fixed set of members, who expect to cooperate on tasks in the future. Second, the rewards being apportioned must affect group members’ expected performance on future cooperative tasks to be attempted by this persistent group. In this case, it will be in the interest of each group member that not only he, but all group members, get their “fair share” (where “fair”, however, is defined not according to some general principles of justice, but rather means “whatever will most effectively guarantee a member’s continued participation and optimum performance in future cooperative endeavors”).
First, this is a generally excellent comment, esp. with puzzle at the end. I think it’d make for a pretty good post in a revamped sequence of running meetups.
My own answer:
Vs lbh’ir qbar lbhe yrnqrefuvc *evtug*, gura lbh fubhyq unir orra rfgnoyvfuvat n cnenqvtz bs ubj gb nccbegvba erjneqf gung crbcyr unir obhtug vagb.
Yrnqrefuvc erdhverq gb nccbegvba gur erjneqf jba’g qebc gb mreb (lbh cebonoyl arrq gb chg fbzr punevfzn/cbzc/prerzbal/rzcngul vagb gur erjneqf sbe gurz gb jbex). Ohg lbh (be rira nygreangr yrnqref jub ner ebgngvat) fubhyq “bayl” unir gb chg va gur rssbeg sbe gur erjneq vgfrys. Gur cebprff bs rinyhngvat crbcyr fubhyq orpbzr eryngviryl fgnaqneqvmrq nf crbcyr funer vaghvgvbaf ba jung’f inyhnoyr, naq gur ohl-va sbe gur erjneq fgehpgher fubhyq fgnovyvmr.
Bs pbhefr, guvf nffhzrf gur yrnqre(f) *qvq* rfgnoyvfu nalguvat fgnaqneqvmrq (vs ebgngvat yrnqref punatr gur erjneq fgehpgher nyy gur gvzr, vg zvtug rira orpbzr uneqre)
Hmm. This is an interesting answer… I can’t quite say it’s wrong, because there are a whole lot of assumptions and frameworks lurking behind it, some of which I recognize only dimly. However, I do recognize some of what I see clearly enough to say that in my experience, running “task groups” (like WoW raid groups/guilds) in this way is… unstable, and not really reliably workable. (The reasons for this are related to the “corrective” vs. “selective” dichotomy which I have obliquely mentioned in the past, and also parallel certain real-world political dichotomies.)
I don’t know how much further it’s appropriate to take this comment thread, as we’ve sort of digressed from the OP, but this particular subtopic (i.e., what I say in the preceding paragraph, as well as your answer to the puzzle) is, IMO, an important and potentially quite fruitful one to explore.
But all of that aside, this part is definitely correct:
I think I’m fine with the conversation continuing, but preferably in a way that ties it back to the original context (i.e. meetups and local rationalist communities), if that’s possible. (That said, as noted, I think these comments could be reworked into a top level post).
I’m not actually sure what you mean by “running a group in this way”, since I don’t think I really specified a way. Any kind of doling out awards seems like it’d congeal into a status quo if you did it roughly the same way each time.
It certainly is possible, and in fact I’ve already had a few discussions (outside of LessWrong) of precisely this application of what I’ve learned from my WoW experiences.
Indeed, you didn’t, which is why it’s tricky to tease out the (apparent-to-me) divergence in our views. It’s not that you specified a way of running a group, but rather that what you said encodes certain assumptions about how a group will, necessarily, be run.
That’s cryptic, I know; I’d like to make my meaning very explicit, but doing so will require a diversion into some more actually-talking-about-WoW, and also many more words. I actually started writing out my answer as a comment, but it got very long very quickly. So I think I’m going to take your suggestion, and try to write this up as a top-level post or two (at which point, I hope, we’ll either come to agree, or at least our disagreement will be clarified completely).
Addendum II
In some cases, “leadership” also consists, in part, of being a Schelling point at which key individuals come together to collectively undertake a difficult task.
Some tasks are so challenging that they require the participation of multiple very competent individuals—all of whom, however, have other things that they could be doing with their time. Such a task may involve substantial investment on the part of each member (which may be simply an investment of time, or it may also require an investment of material resources of some sort).
It would be wholly intolerable, in such a case, for an individual, who is qualified to participate in an attack on such a difficult objective, to lend his participation, to expend his resources—only to see that expenditure wasted, on account of some or all of the other participants lacking a similar level of competence. Similarly, such an individual has little interest, no doubt, in personally going around and attempting to round up other qualified folks, to enlist them in a collective attack on the objective—he has better things to do with his time.
The leader, in this case, is one who can, and does, say: “Alice, come do a 45-minute Baron run with us.” “Do you have a good group,” Alice asks—meaning, of course, “can you promise me that the three other members of the group that you assemble will be at least as competent as you and I are?” The good leader is the one who can credibly answer “Yes” to that question. She (the leader) then goes to Bob, and invites him, and fields the same question, and gives the same answer, and then Carol, and Dave, and now leader+Alice+Bob+Carol+Dave go and do the thing.
Thus the leader manages to put together an unusually effective task group, by virtue of being able to bring together several unusually competent (and suitable) individuals.
This requires some (very basic) “project management” (i.e. scheduling, etc.) skills, and some (slightly less basic) communication/persuasion/personnel-management skills, but most importantly, it requires personal competence and the ability to judge competence in others. The latter is absolutely critical, because when the leader says “I have [or can get] a good group”, each prospective group member must be able to trust that the leader can convince each other prospective group member to participate—which is perhaps not trivial, but more or less surmountable—and that the leader can properly select prospective group members. Note that a failure of the latter skill is far more catastrophic than a failure of the former! If the leader can’t get prospective group members to participate, well—hopes are dashed, a bit of preparation time is wasted, perhaps opportunities to do other things have been unnecessarily turned down, but… it’s not the end of the world. If the leader has misjudged the competence of the group members, then the project proceeds and fails, which is a much greater waste, much more frustrating, etc.
A leader who has this skill set—that of bringing together unusually competent individuals (all of whom are in high demand) to successfully tackle an unusually challenging task—can rack up a very impressive record of accomplishing many very difficult things. It is thus a much more valuable skill set than it might, at first glance, appear to be.
V’q thrff gung vg qbrf. Vs abguvat ryfr, cerprqrag graqf gb pneel abamreb jrvtug.
Good; but—jung pbafgvghgrf cerprqrag? Vf “nalguvat gung guvf tebhc unf rire qbar orsber” gur nccebcevngr pngrtbel, be vf vg fbzrguvat ryfr?
Addendum I
In larger groups, the leader must devote more effort to leading than to participating in the task execution per se.
This is necessary because the “amount of leadership” required for the successful completion of large-group tasks is greater than that required for the succesful completion of small-group tasks.
This is viable because larger groups can more easily “pick up the slack” from a leader who must devote much of his effort to leading (compared to small groups, where everyone must contribute substantial effort to task execution, else the group fails to achieve its goal).
Nevertheless, when the tasks a large group faces become more challenging, and thus require more of group members in order to ensure success, it becomes increasingly important that the leadership role be taken up by someone whose contribution to the group effort aside from leading it, is least important (relative to the contributions of other group members).
(Delegation of leadership responsibilities can alleviate this somewhat, but ultimately it only delays this effect; at some point in the task difficulty curve, these concerns return in full force despite any amount of delegation—which, in any case, will also be necessary.)
(None of these insights are new, of course—these points have been noted before, in the context of, e.g., startups. It is, in any case, interesting to see them arise in this very different context.)