Talk:Reed's law

make it prettier

Added frac{} to the n(n-1)/2 to make it prettier. Otherwise theres no reason to use the math tags as it's all inline. Ixtli 14:39, 3 January 2007 (UTC)[reply]

count the empty set

The formula given for Reed's law - that is, 2^N-N-1 - seems to count the empty set as a valid subgroup of network participants, if one follows the derivation given. I think that the -1 term in the law probably comes from subtracting out the empty set, not from subtracting out the total group (which clearly is a valid group of users). Physicist 18:16, 27 Oct 2004 (UTC)

The -1 term isn't very interesting, really. And the -N term would in more modern terms imply removal of the "solo" value of a community - e.g. an online game (social space) would have fun stuff for me to do alone in addition to all the group opportunities, or consider stuff like last.fm. But at the time of the HBR article, the overlap of solo and group was not as notable today (few real reference implementation of truly scalable groups). But the real challenge is that it's not clear how high Ns are actually scalable. If a social site has 10M members is there really any value in distinguishing the 10M from the 10M+1 groups? E.g. at what point do you get the "one, two, many" effect. This effect would cut into the bulk of the exponential. Maybe I should write an HBR article, too. :-) --Psm 21:47, 15 April 2007 (UTC) , o[reply]

Utility?

"Reed's law is the assertion of David P. Reed that the utility of large networks, particularly social networks, can scale exponentially with the size of the network."

What does "utility" mean in this context? --Abdull 23:12, 10 November 2005 (UTC)[reply]

It obviously means the number of possible sub-groups of network participants... --2001:620:8:2A49:2D42:6651:126D:A08D (talk) 16:58, 19 July 2012 (UTC)[reply]

Asymptotic notation spurious in this context

Point regarding exponentiality is obvious without asymptotic notation to anybody who reads and understands the term 2^N - N - 1, so I have edited it out.

Applying Reid's law to social networks is academic

2^n is not very likely or representative of a real social network. In practice, I'm not going to belong to every possible combination of group, but rather a small number of groups across my set of neighbors (friends). I'm definitely not participating in groups with all subsets of the 175 million (as of Feb 2009) faceboook users, more like 3 groups consisting of a couple hundred members. The mathematics behind the pure combinatorics is one thing, but real-life social scalability is another. The number of possible combinations in a social network is not only irrelevant, it's also impractical across both spatial and temporal axes.

For example, say I work in a company of 100 persons, and there is a group that goes drinking on Friday nights. That group might consist of 10 or so regulars. If someone can't make it one week, that doesn't constitute a distinct social group entity - it's still the friday night drinkers. Ditto if it's a different person missing next week. You wouldn't establish 1013 distinct drinking groups (2^10-10-1) based on all possible participation scenarios. Accounting for all such possible mathematical groupings does not capture the essence and continuity of the organized social activity to which the membership applies. Any absentee-related subgrouping is just an overlay. Of course, there might be intentional division and fracturing, but you're not going to get 2^n.

Furthermore, it's not possible to actively participate in all the different classes of groups. I can't, for instance, attend both the Friday night drinker's club AND Friday night bowling AND Friday movie night, etc, especially when these are in different zip codes. Sure, "groups" online have different rules, but it's going to have largely similar limitations. It still takes time for me to read and type.

Joe Hanink (talk) 06:04, 21 February 2009 (UTC)[reply]