- Authors
- Walter Willinger
- Abstract
- In many respects, the recent popularization of Network Science as "the next big thing" and its
impact on much of Computer Science (CS) has many similarities with "shock and awe" -- a term used by
the Bush administration for its massive hi-tech air strikes at the beginning of the the Iraq war.
Telltale signs that CS is currently experiencing some of the symptoms that are typical of "victims"
of this "shock and awe" treatment are (i) an uncritical and almost submissive embrace of Network
Science concepts, (ii) a disturbing denial of being firmly rooted in the engineering sciences, and
(iii) an acute fear of having lost its identity. In this purposefully provocative but completely tenable
article, I argue that it's time for CS to turn the tables and show Network Science its proper place.
- Documents
-
Tags: Network Science, Internet measurement, data analysis, network modeling and validation
Comments
As far as I can see, there are no rigorous proof(s) that the results from "network science" are not applicable to computer networking problems. On the contrary, more and more interesting results are appearing with provable applications.
I find the read to be at best misleading since it is only arguments, with no rigorous proofs where concepts from "network science" fail. The Internet router-level topology and the traceroutes problems are not proofs. Nobody (al least in the recent years) said that power-laws and preferential-attachment mechanisms discribe Internet router level topology. Further, I find it hard to believe that all these efforts and funding to build measurement infrastructures simply leads to "wrong data" and "network science" coincidencially discribes these observed behaviors (pls note, degree distribution is not the only metric here). It would be interesting to see a proof/empirical evidence (perhaps from the author?) showing that preferential attachment, or approximate preferential attachment-like mechanisms, do not apply to the AS Internet. As far as I can see there is currently no such proof. On the contrary, all the evidence so far says the oposite...
The fact that CS is being dominated by people from other disciplines such as physics is that people who consider themselves "authorities" of the field wasted their time opposing any fundamental innovation in their field rather than encouraging it.
The quality of research should not and will not be determined by those "authorities" but by all the people that affect the field and through proper channels (journals, conferences, etc). And the idea of war between science and engineering is just plain ridiculous.
On the positive side, the statement "Network Science has to learn from past mistakes, broaden its view, and build deeper links with domain experts across the sciences" is a valuable one and should not be forgotten.
Interesting and compelling read. One reason academics jump onto the Network Science bandwagon is perceived impact: a chance to publish in Nature, Science, or PNAS is HUGE - especially for CS people! Those journals get one tons of visibility, citations, and other academic rewards - isn't that what an academic dreams of? Who knows, maybe someone in Network Science will win a science Nobel tomorrow?
Interesting and compelling read. One reason academics jump onto the Network Science bandwagon is perceived impact: a chance to publish in Nature, Science, or PNAS is HUGE - especially for CS people! Those journals get one tons of visibility, citations, and other academic rewards - isn't that what an academic dreams of? Who knows, maybe someone in Network Science will win a science Nobel tomorrow?
Post a comment on this page
