Yes, I'd noticed the discrepancy between Richard's description (0-150) and the Wikipedia page on Dunbar's number. Moreover, I doubt that the distribution is uniform in the range (whether 0-150 or 0- >230).
On Sun, Sep 15, 2013 at 6:29 AM, Raul Miller <[email protected]> wrote: > On Sat, Sep 14, 2013 at 1:56 PM, Roger Hui <[email protected]> > wrote: > > I failed to do an example with 1e7 members. The expected number of > friends > > is 75 per the Dunbar requirement. For n members, the size of s is > > (75*n),2. On J64, integers are 8 bytes each, so the total number of > bytes > > is 8**/(75*n),2 and for n=1e7 equates to 12 GB. > > Note also that the dunbar requirement in this thread is different from > the dunbar number expressed at the wikipedia link (150 there was the > average and expected averages were something in the range of 100-230 > (and I'm not going to verify those numbers because (a) my mouse is > acting up, and (b) the reasoning behind them was just too silly for me > to make sense of)). > > That said, note that we have similarly structured information > available in social networks (facebook graph, for example, but there > are others). So hypothetically speaking, we could play with that data > and see how similar those averages were to the dunbar averages. But, > to do this, we'd undoubtably need greater than 12GB to represent the > data. (I wonder if we have enough interest to put together enough of > a community to implement J for infrastructures such as hadoop?) > > But note also that these toy problems - real problems at that scale > are probably biological (or pornographic?) in character. (Genome > project, was a recent example I noticed.) > > Thanks, > > -- > Raul > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
