Welcome to WebmasterWorld Guest from 184.108.40.206
Forum Moderators: phranque
This was something that always troubled me about PageRank. In some circumstances, links aren't votes, they're votes *against*. This is certainly the case in footnotes to academic articles, where Serge and Larry got the idea of PageRank. Although computer science may not have them, historians, literary critics, etc. are fond of expressions like "see Carter (1982), _contra_ Bligmann (1973)." Read the fully-online Bryn Mawr Classical Review some day. Footnotes aren't votes; they're daggers.
In my own experience, I've developed sites that aimed to be comprehensive in their subject. I listed all sites on a topic, good and bad, labelling—okay, skewering—the bad ones. But Google sees all my links as recommendations. The same goes for those old-time "worst site" directories.
An anti-link would surely help with Google's hate-site problem—the results for "Jew" recently pointed to "jewwatch." In that case the word was Google-bombed toward Wikipedia. Of course, the Googlebombers took infinite care not to link to the offending websites, making it rather difficult to figure out what they were upset about.
So, W3 listen up! Until search engine's "understand" the pages they read, let's all adopt: <a href="http://www.nazi-creeps.com" vote="yuck">.
Creating websites promoting a strong social or political stance is a well known method of getting people to link to you, especially by bloggers who disagree with the point of view being published (hey, get a load of this jerk who says...).
There are also other methods of manipulating people to link to you that I won't mention here (ot).
Google isn't counting links as "votes for approval", it's counting links as "votes for relevance". If you link to it, you must consider it relevant to the subject at hand, even if you disagree with it.
PageRank takes its philosophy from academic citation analysis. Citation analysis isn't meant to tell you whether a cited work is right or wrong, it's mean to tell you if the work is referred to often enough to be worth knowing about. Wrong ideas can be worth knowing about.
Different sorts of connection-based ranking systems are differently valuable. It turns out that the mere fact of being linked to is a good indicator of quality. This is because we have a cultural prediliction for citing web pages that are about relevance and quality. In some cultural contexts, eg., boxing, the guys you defeated are the guys you reference. On the web, "Here's a list of sites I'm outranking" has never caught on. Another example: some of us tattoo words on our body—metadata about whom we wish to be linked to. But a complete analysis of American tattoos would would be a perilous guide to whom to date (apparently "mother"), or what band to list to ("Guns N Roses" anyone?).
As I said, counting all links in one way works pretty well. Having every linker give a detailed evaluation of the site they're linking to—and getting that into a computer system in a way that makes sense—would surely be better. To some extent, this is already done. Googlebombing works because Google evaluates keywords within link text.
It seems probable to me that the "yuck" attribute opens the system to gaming that would produce bad effects more substantial than the good ones. Here and elsewhere, gaming is the hard part for Google to solve.