In a past homework assignment, we looked at the notion of a gatekeeper in a network. The intuition behind our definition was that a gatekeeper in a network–much like a literal one–is in an advantageous position with respect to information flow. However, when we looked at negotiations and power in networks of human subjects, we saw cases where gatekeepers were not necessarily endowed with much power. Still, it seems that this result is only a consequence of the rules of the experiment; that is, the network was not modeling paths along which there could be flow, but rather just the exchange relationships between the subjects. As it stands, we have not seen an example of a network modeling information flow that illustrates when gatekeepers are not necessarily powerful. Could it be, then, that gatekeepers are always powerful?

Consider the following thought experiment: in a small, sparse network, identify a gatekeeper (or even a local gatekeeper); does this node have power? The intuition is “yes,” since these are the kinds of examples that we have seen. Now, what happens as we increase the number of nodes and maintain the relative number of edges? We still have a sparse network, only it has become bigger; additionally, we can still probably identify gatekeepers (perhaps the local variety is now more common), so gatekeepers are still relatively powerful. However, consider what happens as we add more edges to the network while maintaining the number of nodes–it creates more paths along which information can flow. The intuition is now that gatekeepers are harder to find: nodes are generally more connected, able to bypass others more easily.

This thought experiment seems to suggest that the power of gatekeepers is inversely proportional to the connectedness of the network. As a concrete example, we can look at one of the most connected networks that we know–the internet.

A few weeks ago, the whistleblower site Wikileaks.org had been censored by a California judge, who enacted an order proposed by a disgruntled bank that saw some of its sensitive documents detailing fraud and other criminal activity appear on the site. This incident has been covered on various sites around the web, so I’ll defer the actual story to these links:

http://www.thenewfreedom.net/wp/2008/02/18/wikileaks-dns-entry-removed-by-us-district-judge/

For more links and a colorful discussion, see http://yro.slashdot.org/article.pl?sid=08/02/18/043211

What’s interesting is the method of censorship that was proposed (and ultimately enacted). The lawyers for the bank needed Wikileaks to be unreachable as quickly as possible, so they opted to delete its DNS (domain name system) entry. In order to understand how this decision ties in with the topic of this post, we need to look at the basic implications of the DNS paradigm. In very general terms (but find out more about DNS and root nameservers at Wikipedia), whenever you type in a domain name into your browser, it actually performs a lookup to translate the domain into an ip address. Traditionally, there are 13 well-known root nameservers that constitute the Internet’s “official” DNS. So unless you proactively use other DNS services (like OpenDNS, for example), then chances are that your requests are handled by these root nameservers.

The consequence of having these root nameservers is that they are very powerful nodes in the network. Just about everyone’s request for a website would need to go through them–they are the gatekeepers of the internet, in some sense. It’s reasonable, then, that this was the form of censorship the bank chose to pursue, but the question is, “was it effective?”

Recall the results of our thought experiment: in a highly connected network, the power of gatekeepers should be relatively low–this is indeed what happened in the case of the attempt at censoring Wikileaks through its DNS entry. As various other nodes on the internet (news sites, blogs, forums) realized what had happened, many provided links to the explicit IP address of Wikileaks, effectively bypassing the deleted DNS entry. Clearly it would be an impossible task to censor all sites which linked in this clever way to Wikileaks, and so the court order (later overturned) that tried leverage the power of gatekeepers was but a slight inconvenience for the motivated. This is of course only one example that supports the intuition and hypothesis for the power of gatekeepers in dense networks–other examples (or even counter-examples) are of course welcome as comments.

Finally, in the spirit of this post, should our glorious and most wonderful school website ever lose its DNS entry, you may find it at http://128.253.161.179 :)