|
|
|
|
|
 |
Sunday, March 3, 2002 |
|
Reed's Law has been getting press lately; for instance it's referred to in this Jon Udell article. It says that the value of a network is not just proportional to N2 as in Metcalfe's Law; rather, it's actually proportional to 2N (which is much bigger for large N). The difference is that Reed considers subgroups; Metcalfe doesn't. But Reed is probably wrong. His justification assumes that every subgroup of a large group has a unique separate value. For instance, suppose you have group of 100,000 people and a subgroup of 12,456 people. Is another subgroup of 12,455 people, exactly the same as the first subgroup but with one person missing, really by definition a completely different group with its own completely unique value? Come on. That's silly. Luckily there appears to be a way of looking at the problem of calculating the value of the possible subgroups that really does make sense. I've only done the clacs in my head though so far so I'm not posting any details now. But if I'm right, the 2N will go away and (surprisingly to me) Metcalfe's Law will reign supreme once again, even when subgroups are considered. More on this when I get a break from testing and debugging.
|
 |
 |
|
 |
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||