User talk:70.22.128.220

From Wikipedia, the free encyclopedia

[edit] Scale-free Networks

If you plan on making large changes to articles, especially deleting content, it'd be more appropriate for you to sign up and get a user name. In particular I've reverted your changes to scale-free network. I realize there is some debate about the formal definition, but most researchers agree to Li et. al. and not with the old informal definition of a simple power-law degree distribution. You're welcome to disagree and we can talk about it, but please get a username first. Meekohi 15:52, 13 January 2006 (UTC)

[edit] Complex Networks

Out of concern I checked your contributions list, and it appears you've also made some false claims in the Complex networks article. Please make sure any edits you make are verifiable. Meekohi 15:58, 13 January 2006 (UTC)

Sorry, I should be specific. In particular although Barabasi's preferential attachment is the most well known generative model, it is not the first (which was clearly pointed out in the article before you edited it.) Meekohi 15:59, 13 January 2006 (UTC)


Sorry to say that, but the Li it al defintion has absolutelly nothing to do with the definition of a scale-free network. The concept of the scale-free network was introduced in the Barabasi-Albert 1999 Science paper, and its only defintion is that there is a power law degree distribution. And it is only the tail what matters. That is how the field uses it, and feel free to consult any text on the issue, from the book of Vespignani and Pastor-Satorras, to the book of Mendes and Dorogotsev, or any reviews, from the review of Strogatz in Nature to the review of Albert and Barabasi in Rev of Modern Physics, and so on. The Li et al paper is an interesting exception and it is far from mainstream (in fact, I do not think that anybody ever cites it, except Wikipedia :)). YOu may like it, but it is simply wrong. As for the generative models, we need to start seeing the difference between a generative model for power laws and one for networks. Simon/Yule/Price have generated the first models for power laws, and I am in agreement with you there. But not for networks. When you break it down to the bare math of a scalar process, it is similar to the one introduced in the BA model. But the network theory is not about scalar models, but network topology, and in this respect the first *network* generator was by BA and by the Ragavan et al paper from the IBM group. I hope this clarifies this issue.