March 14, 2003

Resist the Oppressive Dominant Internet Hierarchy Through Link Sluttage!

I got an email last week from the thoughtful and highly intelligent Gary Farber of Amygdala that got me thinking. It read, in part:

If you're too busy to take a look, I understand.... I try not to write e-mail on my own blogging too often; I know it can be annoying. So if you'd like to never get such mail from me, please, of course, let me know. Otherwise I'll try to keep my link-whoring to, on average, well under once a week. (This is largely content-dependent, of course.) Meanwhile, a spot now and again seems, alas, the only route to getting an occasional link.... Clearly it would be a far better strategy for me to start blogging early in 2001. I'm working on a Clever Scheme to achieve that, but meanwhile....

My first reaction was that I definitely need to appease Gary Farber of Amygdala, one of the geniuses of our age. His Clever Scheme presumably involves signals sent to well-armed and powerful aliens, wars of conquest waged by said alients using technology beyond human understanding, the construction and use of alien wormhole time machine with their throats threaded with Exotic Matter, and finally with Gary Farber's beginning blogging under an assumed secret identity back in 2000. (Hmmm... who could that secret identity be?)

Such plans are hazardous to their originators, and are very, very hazardous to passers-by. They are best headed off as quickly as possible. And if throwing a few links to Gary Farber is the way to avoid alien conquest of Earth and hazardous experiments that may attract the malevolent attention of the Eschaton (No! Not that Eschaton! This Eschaton!), then put me down for a bunch o' links

And then I began thinking. Suppose we start with some process by which web sites of varying quality arrive, and people link to them, with the rate of linkage proportional to (a) the number of people who visit a website and (b) the quality of the website. The number of people who visit will be proportional to the number of links to the website times the time it is up. You might argue that as time passes and websites vanish or are reworked, that the density of links will converge to be proportional to some appropriate measure of quality. You might argue this, but for a couple of things: Google, and the normal working of positive-feedback processes.

A huge proportion to visitors to my website arrive via Google. (Indeed, I often arrive at pages on my website via Google. It's the only *&^^#@$!* way to find anything.) Each time someone arrives via Google, there's a chance they'll add a link. Sites with lots of links get high PageRanks. Sites with high PageRanks get lots of links. Will the system converge to fundamentals, or will it spiral out of control? Will positive feedback takes hold and in the end create many popular sites that are popular primarily because of their popularity? It's not just Google. There are other positive-feedback processes at work as well--other channels through which those sites with lots of links acquire more links, and thus find themselves with more channels that lead to them through the web, and more traffic.

How can we purge the internet of the positive-feedback effects of pure celebrity, and return to a link structure--and a Google PageRank, for PageRank is derived from link structure--that corresponds to the fundamental goodness and informativeness of sites, rather than their mere celebrity?

In the case of Google, it is clear what to do. When Google compares the links to a site at time t with the links at some later time t', it has to try to assess how much of that change in link numbers is the result not of the site's informativeness but of the PageRank the site held at time t. It needs to subtract from PageRank at time t' some parameter lambda times the site's PageRank at time t. How to calculate lambda? A regression of changes in PageRank on the past level of PageRank, some transformed measure of the site's longevity (because the longest-lived sites have had the greatest opportunity to acquire net celebrity), and some alternative measure of "quality" suggests itself. It is clear that Google should do this: a measure of PageRank that did not assign sites a high PageRank because they held a high PageRank in the past, but instead assigned sites a high PageRank because they were useful and informative, would be a better PageRank. Purging PageRank of the effects of the positive-feedback dynamic of celebrity should be high on Google's priority list already. (Hell, maybe they've already done it.)

But what about the rest of us? We don't aggregate up and summarize the link structure of the web the way Google does. But our link patterns are correlated with net celebrity and with past net longevity as well as with our judgments of interest and excellence. What should we do?

Promiscuous link sluttage is the answer. I call on everyone reading this with a website or a weblog to find a measure of sites that link to you--my current favorite is Technorati, but whatever you want. Rank the sites that list to you by their "authority." Find the least authoritative, look at it, and if you like it link to it. (If you don't like it, don't.) Link to it. Put it in your blogroll. Talk about it. And don't just pick one. Pick two--maybe one from the least authoritative category, and one at random.

LINK SLUTTAGE IS THE TOOL TO SMASH THE LINK-CHAINS THAT ARE THE OPPRESSIVE DOMINATION OF THE INTERNET HIERARCHY!!!

My entries this week are:

(Of course) Gary Farber's Amygdala.

Ncfocus: Ideas, issues and life in nevada county CA: Belaboring the obvious since 2003.

Posted by DeLong at March 14, 2003 06:19 PM | TrackBack

Comments

It's an empirical question to see whether people are more likely to find new blogs via blogroll links or links that show up on search engines, as you noted. Personally, I think it's just as good - if not better - if people find just one of my blog entries that is particularly relevant for the topic they are exploring as opposed to seeing my main blog page that may show nothing relevant to them on that particular day.

My blog gets a LOT of Google referals. And I can tell that many visitors to one specific entry never click through to the main blog entry page. But that's fine. I'm quite content that thousands have read my rant about a mistake in the movie "Chicago" b/c it comes up as a top five result on Google when you do a search for movie chicago. At this point I know that whatever I post on, if I pick the subject of the blog entry well enough (which then is put into the title tag of my blog entry page, information relevant to search engine ranking) it will show up quite high for related searches within days.

I suspect most people who link through to my blog via a blogroll are the owners of that blogroll (if I wanted to verify this I could.. one day when I really need something to procrastinate with). I see a surge in visitors from another blog when the blogger has featured a link to mine in an entry. And that's how I see myself exploring new blogs off of others' sites as well: if it gets a mention in an entry I might follow through. So your suggestion, Brad, to feature certain blogs in an entry could be a fruitful approach to driving more traffic to certain blogs. Then again, this may already be automatic on some blogs where we engage with other blogs' contents and link back whenever we take something from their entries (and this can act as a type of quality control).

As for Gary Farber's email about blogs, it's an interesting reverse phenomenon. Announcement lists predate blogs for many people. I started my mailing list of links six months before I started my blog. The list has over 350 subscribers. Most of these people don't read my blog unless I include links to particular entries. (Of course, you could ask whether the subscribers actually read my list and I don't have data on that, but responses to occasional queries posted on it suggest that many do.) So if I want a Web site out there to really get exposure, I better feature it on my list as well as my blog.

Point being, there are lots of ways to get traffic on a blog and I'm not convinced blogrolls are the most important although they seem to be the ones the social network mapping approach using data from blogrolls favors. I guess the question is whether you want your blog to be read by others who often visit blogs or just anyone out there. I'm quite sure that the average user still doesn't know what a "Weblog" or "blog" is (and when it comes to linking in to just one particular entry this doesn't necessarily
matter). I don't know of large scale data on this. I have tiny tiny data from a survey I did on a random sample of Internet users in my local county (n=100) which show that 50% of people say they have no understanding of the term "Weblog", 25% say they have little understanding. Then I had 39 of them (a random 39) answer a multiple choice question about what it means and none of them picked "A Web page with frequently updated posts that are arranged chronologically". (Date of survey: 2001-2002.)

I think this comment is long enough that I should post it on my own blog where I'll add some underlying links and a few additional notes.

BTW, is there a reason to only feature links to blogs that link to you? (As per your comment "I call on everyone reading this with a website or a weblog to find a measure of sites that link to you...".) There are blogs I think are great and bummer if they don't link to me, but I still want to point people to their content b/c I think they're worth reading.

Posted by: Eszter on March 15, 2003 08:49 AM

____

BTW, I realize your post may have been tongue-in-cheek, but I thought it raised some interesting issues worth exploring. (Maybe I've just been spending too much time on my dissertation which looks at how people find their way to content on the Web.)

Posted by: Eszter on March 15, 2003 01:09 PM

____

Does longevity skew beyond the ability of other factors - such as utility - to correct? Arguably not - and it's hard to resist noting that you (like most of the rest of us) use google as a synonym for search engine. Being better has been more than enough compensation for being younger in the population of search engines, it seems completely plausible that that should be true of blogs as well.

Albert-Laszlo Barabasi in his book 'Linked' - modestly subtitled 'how everything is connected to everything else and what it means for science business and everyday life' - addresses precisely this question, and concludes (at the end of chapter 8) 'In networks that display fit-get-rich behavior, competition leads to scale-free topology. Most networks we have studied so far - the web, the internet, the cell, Hollywood, and many other real networks - belong to this category. The winner shares the spotlight with a continuous hierarchy of links.'

He acknowledges the 'theoretical possiblity that in some systems the winner can grab all the links. When that happnes, the scale-free topology vanishes. So far among real systems, only the operating systems market, with Microsoft as its dominating hub, appears to fit the bill. Are there other systems out there displaying a similar behavior? Very likely. It will take some time, however, to recognise them all.'

So my guesss is that the network of blogs - still a very immature one - will settle down into a fit-get-rich pattern, not a first-get-rich pattern. Whether any of us trying to force the pace will make any difference is another matter.

Posted by: marek on March 15, 2003 03:42 PM

____

I'm interested in getting details about the Exotic Matter and well-armed aliens. Thank you in advance.

Posted by: vachon on March 15, 2003 04:05 PM

____

OK I assume that google has solved the google problem already but how did they do it ? It is very hard to tell if high page rank at t-1 causes high page rank at t or high quality (which probably changes slowly) causes high page rank at t-1 and t. They could correct page rank for age of site, but that is assuming that new, say, blogs are the same average quality as old blogs. Since everyone is making a blog (including me) and since, by now, people assume no one will ever look at their blog (well anyway I assume that) new blogs could be lazy sloppy me too-ing (mine certainly is check if you doubt it).

One thing google can do is compare google page rank persistence with use of google. Use of google has shot up, if page rank persistence has increased it is reasonable to suspect google based positive feedback. So page rank at t-1 effect separate from quality causes page rank at t-1 and t is from (page rank at t-1)(use of google).

Finally, if they really want to know, they can do experiments adding a random term to some page ranks and seeing what happens. I don't think any law would stop them.

Posted by: Robert on March 15, 2003 07:14 PM

____

Re: "google as a synonym for search engine" in the above comment, the highest figure I've heard of backed up by data is that Google refers around 36% of all searches (other sources point to ~30%). So that leaves a lot of searches to other engines. If anyone has higher figures - not just anecdotal evidence - please share.

Posted by: Eszter on March 16, 2003 09:04 AM

____

I've recently started using "AddFreeStats" as one of my hit counters, and and among its many useful functions, it keeps track of search engine hits by percentage. Though this only covers a couple of weeks, here's what it says:

Google 537 59.40%
Yahoo 151 16.70%
Msn 79 8.74%
Other 75 8.30%
Lycos 14 1.55%
AOL 13 1.44%
Netscape 9 1.00%
AddFreeStats 6 0.66%
Dogpile 6 0.66%
Altavista 6 0.66%
Excite 4 0.44%
MetaCrawler 3 0.33%
Looksmart 1 0.11%

Posted by: Gary Farber on March 16, 2003 05:37 PM

____

Gary, thanks for sharing those stats. I'm assuming these are based on referrals to your site? I'm afraid I need something a bit more general than that. But this is definitely a helpful data point.

Posted by: Eszter on March 17, 2003 08:09 AM

____

thanks for the info Gary, great job

Posted by: zip codes on April 25, 2003 10:12 AM

____

I'd say so what a great line!

Posted by: mikelee on September 8, 2003 04:20 PM

____

Solo para saludar

Posted by: netflix on September 11, 2003 11:44 AM

____

I agree

Posted by: phentermine on October 19, 2003 10:05 AM

____

Interesting information to think about

Posted by: activex on November 3, 2003 01:56 PM

____

Interesting information to think about

Posted by: activex on November 3, 2003 02:00 PM

____

Interesting information to think about

Posted by: activex on November 3, 2003 02:03 PM

____

Hello. I just wanted to give a quick greeting and tell you I enjoyed reading your material.

Posted by: Atkins Diet on November 10, 2003 07:44 PM

____

Hello. I just wanted to give a quick greeting and tell you I enjoyed reading your material.

Posted by: Atkins Diet on November 10, 2003 07:52 PM

____

This is a very informative website! Very interesting!
I enjoy your site very much! Keep up the good work!

Posted by: hgh on November 13, 2003 01:26 PM

____

This is a very informative website! Very interesting!
I enjoy your site very much! Keep up the good work!

Posted by: hgh on November 13, 2003 01:29 PM

____

This is a very informative website! Very interesting!
I enjoy your site very much! Keep up the good work!

Posted by: hgh on November 13, 2003 01:29 PM

____

Point being, there are lots of ways to get traffic on a blog and I'm not convinced blogrolls are the most important although they seem to be the ones the social network mapping approach using data from blogrolls favors. I guess the question is whether you want your blog to be read by others who often visit blogs or just anyone out there. I'm quite sure that the average user still doesn't know what a "Weblog" or "blog" is (and when it comes to linking in to just one particular entry this doesn't necessarily
matter).

Posted by: liv.52 on December 7, 2003 08:08 AM

____

Point being, there are lots of ways to get traffic on a blog and I'm not convinced blogrolls are the most important although they seem to be the ones the social network mapping approach using data from blogrolls favors. I guess the question is whether you want your blog to be read by others who often visit blogs or just anyone out there. I'm quite sure that the average user still doesn't know what a "Weblog" or "blog" is (and when it comes to linking in to just one particular entry this doesn't necessarily
matter).

Posted by: liv.52 on December 7, 2003 08:12 AM

____

Thank you, I just wanted to give a greeting and tell you I enjoyed reading your material.

more..

Posted by: black on December 9, 2003 09:13 AM

____

____

HI, Wow.. this is a very informative website!
I enjoy your site very much! Keep up the good work!

Atkins Diet plan is the most poppular low carb diet today. Lose weight while you sleep!

Posted by: Atkins Diet Plan on December 9, 2003 03:34 PM

____

____

____

http://www.posters-land.com http://www.posters-land.com/2pac.htm http://www.posters-land.com/adults.htm http://www.posters-land.com/animal.htm http://www.posters-land.com/anime.htm http://www.posters-land.com/antique.htm http://www.posters-land.com/artprint.htm http://www.posters-land.com/beach.htm http://www.posters-land.com/britney.htm http://www.posters-land.com/buy.htm http://www.posters-land.com/bw.htm http://www.posters-land.com/car.htm http://www.posters-land.com/che.htm http://www.posters-land.com/christmas.htm http://www.posters-land.com/degas.htm http://www.posters-land.com/dinosaur.htm http://www.posters-land.com/dog.htm http://www.posters-land.com/dolphin.htm http://www.posters-land.com/egypt.htm http://www.posters-land.com/fantasy.htm http://www.posters-land.com/fish.htm http://www.posters-land.com/flower.htm http://www.posters-land.com/football.htm http://www.posters-land.com/frame.htm http://www.posters-land.com/hawaii.htm http://www.posters-land.com/hiphop.htm http://www.posters-land.com/horse.htm http://www.posters-land.com/inspiration.htm http://www.posters-land.com/ireland.htm http://www.posters-land.com/jazz.htm http://www.posters-land.com/kids.htm http://www.posters-land.com/lopez.htm http://www.posters-land.com/lotr.htm http://www.posters-land.com/marilyn.htm http://www.posters-land.com/matrix.htm http://www.posters-land.com/model.htm http://www.posters-land.com/moon.htm http://www.posters-land.com/motivate.htm http://www.posters-land.com/motorcycle.htm http://www.posters-land.com/movie.htm http://www.posters-land.com/movies.htm http://www.posters-land.com/music.htm http://www.posters-land.com/nature.htm http://www.posters-land.com/nude.htm http://www.posters-land.com/owl.htm http://www.posters-land.com/playboy.htm http://www.posters-land.com/propaganda.htm http://www.posters-land.com/rap.htm http://www.posters-land.com/safety.htm http://www.posters-land.com/scarface.htm http://www.posters-land.com/shark.htm http://www.posters-land.com/soccer.htm http://www.posters-land.com/space.htm http://www.posters-land.com/sports.htm http://www.posters-land.com/starwars.htm http://www.posters-land.com/tiger.htm http://www.posters-land.com/travel.htm http://www.posters-land.com/tupac.htm http://www.posters-land.com/vintage.htm http://www.posters-land.com/wanted.htm http://www.posters-land.com/wildlife.htm http://www.posters-land.com/wolf.htm http://www.posters-land.com/woman.htm

Posted by: www.posters-land.com on December 15, 2003 06:37 PM

____

____

Once again, I've found a wonderful website. I really like these blogs. How did you install this on your website? Thanks!

Posted by: Diet Reviews Zone on December 16, 2003 07:51 AM

____

Thank you very much,I like this website! Do it good.

Posted by: south beach diet on December 19, 2003 02:08 AM

____

I search for forum like this long time.You website is very good!I will come next time !

Posted by: ultimate hgh on December 21, 2003 10:39 PM

____

very interesting site, great read.

Posted by: free casino on December 22, 2003 12:13 AM

____

Thank you, I just wanted to give a greeting and tell you I like your website very much.

Posted by: pro hgh on December 22, 2003 12:36 AM

____

please post more comments, I will visit this site again!Happy New Year!!!

Posted by: HGH SPRAY on December 23, 2003 08:10 PM

____

excellent weblog service!great service - thanks

Posted by: online casino on December 27, 2003 07:20 PM

____

just wanted to give a greeting and tell you I like your website very much.HAPPY NEW YEAR!

Posted by: buy hgh on December 28, 2003 06:20 AM

____

great blog

Posted by: online casino on December 29, 2003 10:36 PM

____

great blog

Posted by: online casino on December 29, 2003 10:36 PM

____

Hoo great stuff!

Posted by: anal bondage games on January 4, 2004 06:29 AM

____

Hoo great stuff!

Posted by: anal bondage games on January 4, 2004 06:33 AM

____

I assume Helen Thomas is referring to the current President's father when she says "Bush." But saying "Ike" would probably make her senile dementia even more obvious.

Posted by: SexTracker Statistics on January 5, 2004 08:45 PM

____

Blackjack Roulette Slot Machine Baccarat

Posted by: gambling on January 11, 2004 04:04 AM

____

Post a comment
















__