6 ms·

So... I think Dunbar's number exemplifies a failing in our social science culture, or maybe our epistemological tendencies. .. or maybe it just shows what hap

So... I think Dunbar's number exemplifies a failing in our social science culture, or maybe our epistemological tendencies. .. or maybe it just shows what happens when MBA texbooks adopt something. We don't really know what to do with a piece of insight or knowledge like Dunbar's number. When is it "true?" When is it relevant? What does it say about how we should organize? What does it say about how organisations of different sizes function? What kind of successes or failures does Dunbar's nunber suggest we encounter. I'm not saying that Dunbar's number is wrong. I'm suggesting that for it to be profound in a scientific sense, it needs to have consequences. Prediction, prescription, insight about existing structures, something. We intuitively agree that working at an 80 person company is different from an 800 person could company. I don't see Dunbar's theories adding much beyond that, no predictions or insights about how such organisations work. Mostly it's a piece of theory that can be thrown onto slides, in support or objection to something as needed. Is there a there, there? Do I have this wrong?

bjelkeman-again 5y ago
My experience from the army, a startup growing to 100 or so (majority remote, or local offices), hanging out in online communities which grow beyond 120, is that around 100-130 it gets hard to keep track of all the people and have some emotional relationship. Without that relationship it gets harder for me to care about the progress the group makes as a unit. Interestingly enough, for me it is based on context. It is ok to have 100 people in the company, 100 people in an online community and 100 people in an off line community, all at the same time, and still keep track and have some kind of relationship. When a context grows too large it doesn’t seem to affect the other contexts.
- dalbasal 5y ago
  I agree that there are elements like this that make intuitive sense. What I am saying is: After at these years on military/boardroom slides... what does Dunbar's tell us? How has it influenced how we do anything? What should or shouldn't be organized in smaller/larger groups? How? It just feels like a lot of popular social science tidbits give us a great introduction with no follow through. What was the point of all those slides?
  - bjelkeman-again 5y ago
    To me it fits into the experience that as a team goes through growth, you need to adjust your team communications/management as you grow. Roughly: 7 people, 15, 35-42, 100ish. Where the Dunbar number is just the top number. (Which tells me I need to pull out Vern Harnish’s Scaling Up, again).
    - dalbasal 5y ago
      Ok, but where does Dunbar add insight. What would be different in our approach without Dunbar?
      - bjelkeman-again 5y ago
        It confirmed my observations about teams of 100+ people. It serves a purpose for me as a way to start discussing the issue of communications and management within a team. Not a huge thing, like many things. But it is a useful pice of a puzzle that works, for me.
      - WorldMaker 5y ago
        That's a bit of a chicken-and-egg question here, too. Armies have been organized around no more than 100/150 people in a single "unit"/"team" for millennia, prior to any of Dunbar's research much less "Dunbar's Number". There were a lot of observed practical concerns that led to that number, some of which are unrelated such as top-down command structures, but some of which would be direct observations of similar (if not the exact same) phenomenon that Dunbar studied. In this case Dunbar's Number gives us far more insight into why armies have organized that way for millennia than the other way around (telling them how to organize). One group discovered it practically and Dunbar theorized about if that was just a practical local maxima or if it was a more global maximum. What becomes different in our approach post-Dunbar's Number is the added perspective of: not only has this been proven to work in millennia of military practice, but in theory we know it is a theoretical barrier in far more social situations than just those that also have confounding factors like top-down management.
- ItsMonkk 5y ago
  I've been trying to figure this out too. My insights actually started from a post made here, where someone talked about post 'weirdness'. There are so many posts that we skim all of the time. If there is a post that is weird, we skip it. If there is a thread that is weird, we collapse it. What I realized however, if that weirdness is coming from the people who are here all the time, at the top of the karma charts, and people see their name all of the time and respect their opinions, that you can become more weird. And so as the community grows and as you communicate with more people, as you see the same person less and less often, the ability to be weird drops. The larger a community, the less it can move as a cohesive unit, the less it can find new ideas. So what we're really looking at is how Metcalfe's law clashes with the Overton window. That's why Dunbar's number is a thing. As we communicate we subtly change our opinions, and when a group becomes to large, the communication scaling at a factorial rate eventually meets its limit. You end up with a group that believes different ideas, and therefore the best thing to do is split. We don't see the splits in existing social media because even though the Overton Window is smaller and smaller with larger and larger numbers, more and more people are being pulled into it, and exponentially more people are tossed aside.
  - bjelkeman-again 5y ago
    I’m still surprised the discourse at HN is reasonably decent still. I do think we end up with majority issues where a cultural bias from a larger population will define “how it is”. But it isn’t too bad.
aaron695 5y ago
wahnfrieden 5y ago
Graeber and Wengrow dug into it recently and generated compelling research that shows it is indeed untrue nonsense, simply inaccurate given the evidence
- tomcatfish 5y ago
  I tried to look for this research, and am referring to what I found. If there's more, I'll be wrong below. Why do so many commenters here throw away the many many Dunbar's number studies in favor of what appears to be one clump of new research? What do Graber and Wengrow have that convinces you everyone except for them is making some error?
  - dalbasal 5y ago
    Wengrow has it out for Dunbar's number for fairly esoteric philosophical reasons. Dunbar is often taken as meaning that above a certain group size, direct, intuitive or emergent democratic organisation isn't possible. You need kings, presidents, parliaments , courts and whatnot. He's mad at Yuval Noah Harari for similar reasons. It's somewhat of a straw man, imo. As for Dunbar, what do all these Dunbar's number studies amount to? I mean sure, they confirm something about how many relationships we can maintain, and such. What does any of that imply though. Dunbar's number studies haven't really proposed, tested or studied any consequences to their theories. What happens when group sizes increase beyond the number? That makes it pretty empty theory, imo.
    - wahnfrieden 5y ago
      It's promising validation for fans of there being alternatives to capitalist realism. You also can't deny there's piles of shit discourse built on top of the number, there's a good amount to unwind there.
      - dalbasal 5y ago
        Perhaps, but personaly I found it disingenuous. Sure, Dunbar has a lot of influence in shallow realms: MBA textbooks, managerial self help books, military slideshows and such. Re: umwinding... I think most of what Wengrow unwinds is a strawman. The steelman version of Dunbar is pretty small and unimportant. The strawman is pretty much made up by Wengrow. It all amounts to cheesy discourse, IMO. It has a "what the media don't want you to know" quality to it.
brightball 5y ago
Scaled Agile Framework (SAFe) uses it to define the maximum size of a their organizational units (Release Trains as they are called). Everything else in the framework is geared towards helping the units work together more effectively. Been rolled out with great success at much larger organizations too.
- slowhand09 5y ago
  Used at my work. While lauded as great, its the bane of our existence.
  - brightball 5y ago
    Would love to get your input on this Ask. https://news.ycombinator.com/item?id=30948201 https://news.ycombinator.com/item?id=30948201
- jzackpete 5y ago
  Someone at my last job threatened to quit if they implemented SAFe. My current employer has implemented it and now I'm actively looking for another job.
  - brightball 5y ago
    I probably need to do a blog post on this exact subject. As a dev of 20+ years, I got into SAFe because it's IMHO the ideal work environment for developers. It's supposed to put developers in control of the schedule, prevent you from being handed arbitrary deadlines, avoid constant changes in priority by letting teams plan out for a quarter at a time, avoid over planning the time you have (2/3 of time planned per 2 week iteration and 2 full weeks at the end left entirely open) and make sure everybody is clear up front about what the risks are to being able to deliver on the plan. There's supposed to be 2 backlogs, one for the product side to prioritize and one for the dev side to prioritize without any influence from product so that you can ensure all of the technical stuff that needs to be done can be done without having to spend a ton of time explaining why to the non-technical side of the house. End of the day you should end up with less meetings and a more realistic schedule that you, the developer, have essentially created. If leadership won't give up the control to let the process work, everything goes to crap. Every place I've heard of bad SAFe experiences, that's been the cause fwiw. Usually the most frustrated people in a SAFe are the micromanagers (and it's supposed to be that way). Would you be willing to share why it's going so wrong? My keybase contact info is in my profile.
michael1999 5y ago
He makes concrete predictions about expected troop size in primates based on neocortex volume. You can criticize that he cherry picks the volume measure used after playing with different options, but there are real predictions that you can check against non-human animals.

bjelkeman-again 5y ago

My experience from the army, a startup growing to 100 or so (majority remote, or local offices), hanging out in online communities which grow beyond 120, is that around 100-130 it gets hard to keep track of all the people and have some emotional relationship. Without that relationship it gets harder for me to care about the progress the group makes as a unit. Interestingly enough, for me it is based on context. It is ok to have 100 people in the company, 100 people in an online community and 100 people in an off line community, all at the same time, and still keep track and have some kind of relationship. When a context grows too large it doesn’t seem to affect the other contexts.

dalbasal 5y ago
I agree that there are elements like this that make intuitive sense. What I am saying is: After at these years on military/boardroom slides... what does Dunbar's tell us? How has it influenced how we do anything? What should or shouldn't be organized in smaller/larger groups? How? It just feels like a lot of popular social science tidbits give us a great introduction with no follow through. What was the point of all those slides?
- bjelkeman-again 5y ago
  To me it fits into the experience that as a team goes through growth, you need to adjust your team communications/management as you grow. Roughly: 7 people, 15, 35-42, 100ish. Where the Dunbar number is just the top number. (Which tells me I need to pull out Vern Harnish’s Scaling Up, again).
  - dalbasal 5y ago
    Ok, but where does Dunbar add insight. What would be different in our approach without Dunbar?
    - bjelkeman-again 5y ago
      It confirmed my observations about teams of 100+ people. It serves a purpose for me as a way to start discussing the issue of communications and management within a team. Not a huge thing, like many things. But it is a useful pice of a puzzle that works, for me.
    - WorldMaker 5y ago
      That's a bit of a chicken-and-egg question here, too. Armies have been organized around no more than 100/150 people in a single "unit"/"team" for millennia, prior to any of Dunbar's research much less "Dunbar's Number". There were a lot of observed practical concerns that led to that number, some of which are unrelated such as top-down command structures, but some of which would be direct observations of similar (if not the exact same) phenomenon that Dunbar studied. In this case Dunbar's Number gives us far more insight into why armies have organized that way for millennia than the other way around (telling them how to organize). One group discovered it practically and Dunbar theorized about if that was just a practical local maxima or if it was a more global maximum. What becomes different in our approach post-Dunbar's Number is the added perspective of: not only has this been proven to work in millennia of military practice, but in theory we know it is a theoretical barrier in far more social situations than just those that also have confounding factors like top-down management.
ItsMonkk 5y ago
I've been trying to figure this out too. My insights actually started from a post made here, where someone talked about post 'weirdness'. There are so many posts that we skim all of the time. If there is a post that is weird, we skip it. If there is a thread that is weird, we collapse it. What I realized however, if that weirdness is coming from the people who are here all the time, at the top of the karma charts, and people see their name all of the time and respect their opinions, that you can become more weird. And so as the community grows and as you communicate with more people, as you see the same person less and less often, the ability to be weird drops. The larger a community, the less it can move as a cohesive unit, the less it can find new ideas. So what we're really looking at is how Metcalfe's law clashes with the Overton window. That's why Dunbar's number is a thing. As we communicate we subtly change our opinions, and when a group becomes to large, the communication scaling at a factorial rate eventually meets its limit. You end up with a group that believes different ideas, and therefore the best thing to do is split. We don't see the splits in existing social media because even though the Overton Window is smaller and smaller with larger and larger numbers, more and more people are being pulled into it, and exponentially more people are tossed aside.
- bjelkeman-again 5y ago
  I’m still surprised the discourse at HN is reasonably decent still. I do think we end up with majority issues where a cultural bias from a larger population will define “how it is”. But it isn’t too bad.

aaron695 5y ago

wahnfrieden 5y ago

Graeber and Wengrow dug into it recently and generated compelling research that shows it is indeed untrue nonsense, simply inaccurate given the evidence

tomcatfish 5y ago
I tried to look for this research, and am referring to what I found. If there's more, I'll be wrong below. Why do so many commenters here throw away the many many Dunbar's number studies in favor of what appears to be one clump of new research? What do Graber and Wengrow have that convinces you everyone except for them is making some error?
- dalbasal 5y ago
  Wengrow has it out for Dunbar's number for fairly esoteric philosophical reasons. Dunbar is often taken as meaning that above a certain group size, direct, intuitive or emergent democratic organisation isn't possible. You need kings, presidents, parliaments , courts and whatnot. He's mad at Yuval Noah Harari for similar reasons. It's somewhat of a straw man, imo. As for Dunbar, what do all these Dunbar's number studies amount to? I mean sure, they confirm something about how many relationships we can maintain, and such. What does any of that imply though. Dunbar's number studies haven't really proposed, tested or studied any consequences to their theories. What happens when group sizes increase beyond the number? That makes it pretty empty theory, imo.
  - wahnfrieden 5y ago
    It's promising validation for fans of there being alternatives to capitalist realism. You also can't deny there's piles of shit discourse built on top of the number, there's a good amount to unwind there.
    - dalbasal 5y ago
      Perhaps, but personaly I found it disingenuous. Sure, Dunbar has a lot of influence in shallow realms: MBA textbooks, managerial self help books, military slideshows and such. Re: umwinding... I think most of what Wengrow unwinds is a strawman. The steelman version of Dunbar is pretty small and unimportant. The strawman is pretty much made up by Wengrow. It all amounts to cheesy discourse, IMO. It has a "what the media don't want you to know" quality to it.

brightball 5y ago

Scaled Agile Framework (SAFe) uses it to define the maximum size of a their organizational units (Release Trains as they are called). Everything else in the framework is geared towards helping the units work together more effectively. Been rolled out with great success at much larger organizations too.

slowhand09 5y ago
Used at my work. While lauded as great, its the bane of our existence.
- brightball 5y ago
  Would love to get your input on this Ask. https://news.ycombinator.com/item?id=30948201 https://news.ycombinator.com/item?id=30948201
jzackpete 5y ago
Someone at my last job threatened to quit if they implemented SAFe. My current employer has implemented it and now I'm actively looking for another job.
- brightball 5y ago
  I probably need to do a blog post on this exact subject. As a dev of 20+ years, I got into SAFe because it's IMHO the ideal work environment for developers. It's supposed to put developers in control of the schedule, prevent you from being handed arbitrary deadlines, avoid constant changes in priority by letting teams plan out for a quarter at a time, avoid over planning the time you have (2/3 of time planned per 2 week iteration and 2 full weeks at the end left entirely open) and make sure everybody is clear up front about what the risks are to being able to deliver on the plan. There's supposed to be 2 backlogs, one for the product side to prioritize and one for the dev side to prioritize without any influence from product so that you can ensure all of the technical stuff that needs to be done can be done without having to spend a ton of time explaining why to the non-technical side of the house. End of the day you should end up with less meetings and a more realistic schedule that you, the developer, have essentially created. If leadership won't give up the control to let the process work, everything goes to crap. Every place I've heard of bad SAFe experiences, that's been the cause fwiw. Usually the most frustrated people in a SAFe are the micromanagers (and it's supposed to be that way). Would you be willing to share why it's going so wrong? My keybase contact info is in my profile.

michael1999 5y ago

He makes concrete predictions about expected troop size in primates based on neocortex volume. You can criticize that he cherry picks the volume measure used after playing with different options, but there are real predictions that you can check against non-human animals.