Is Google bomb-proof now? May 7, 2002 12:09 PM   Subscribe

Is Google bomb-proof now? After seeing "verisign" posts all over the weblog world, I decided to check the progress on Google. The article calling for the bomb is #79 in a search for Verisign as of right now, and the link itself within the post doesn't show up in the first 100 results for me. Perhaps Google has gotten wise to everyone's shenanigans?
posted by mathowie to General Weblog-Related at 12:09 PM (43 comments total)

This is interesting because I've seen dozens of well-indexed weblogs pointing to textism, all ranking very highly for searches of just about anything, but there is no sign of it on Google. What's up with that?
posted by mathowie (staff) at 12:11 PM on May 7, 2002


The assigning of value for a search page to a search term is an iterative process. It might take a couple iterations (up to a few months) for the page to reach it's maximum value. A google search (strangely enough) for PageRank will find pages that go into greater detail, although if someone is really interested, I could expand a bit...
posted by websavvy at 12:13 PM on May 7, 2002


Could it be that:

a) the Google spider isn't that quick?

b) VeriSign has too much of an established Net presence for one Google bomb to work?

c) VeriSign pays to be up top? Wait, Google doesn't work like that, right?

Google-bombing, much like the real-life version, doesn't get the job done. Only a long, protracted campaign will remove Saddam Hussein, er, VeriSign from power. This is a good start, but it'll mean nothing if it just becomes a meme for the day, and bloggers' attention spans quit as they are wont to do.

The media needs to get in on this shit.

Then again, the stock price is tanking anyway...
posted by solistrato at 12:17 PM on May 7, 2002


websavvy, it took less than 48 hours for my bomb to hit #1, but after 4 days nothing has surfaced on Google, even though many blogs that get indexed daily had linked it.

Something is up, there's no reason why the post being linked would not show up, but the post asking for the link would. I suspect the Google algorythm has changed (it probably has a failover now, so that if 30 references to a new page suddenly surface, it is ignored rather than indexed).

I bet John Hiller is writing something right now about his.
posted by mathowie (staff) at 12:29 PM on May 7, 2002


Google reorganizes their indices periodically. If you bomb near a re-indexing, and you have a popular site, and many other popular sites involved in the bombing (popular sites are crawled more frequently), then the effects can show up very quickly.

Sometimes, the system will require more time for the net effect to emerge.

Which is not to say that there isn't a conspiracy...

Lots (and I mean lots) of interesting info here.
posted by websavvy at 12:36 PM on May 7, 2002


And another thing!

Google bombing is just exploiting the essential nature of google. They rank higher for inbound links, especially from highly-ranked sites. Of course there's a chicken-and-egg problem in there (how do they know that the referring sites are highly ranked? Their referrers, of course, ad infinitum).

What I'm trying to say is that you can't take the bombing out of Google without stripping them of their ace-in-the-hole, a search engine that provides good links without a lot of human intervention. I'm betting that they won't be drastically altering the fundamental behaviour of their engine anytime soon.
posted by websavvy at 12:40 PM on May 7, 2002


websavvy, if they haven't made any changes, there is no reason why something similar to what occurred a couple months ago, which only took 48 hours, has not happened in four days. We have essentially repeated the exercise, having blogs that are highly ranked and indexed at least once a day, all pointing to the same place, and there is no net effect days later. There is little difference in variables (Though verisign is much more popular than some company called Critical IP, all other things are almost equal), and the fact that the post calling for links is showing up, but not the effects of the links leads me to suspect something is in fact wrong, and that Google has changed things.

I fully understand what Google bombing is, and I am simply asking everyone why it hasn't worked in this case. Odd, no? Something must be up.
posted by mathowie (staff) at 12:59 PM on May 7, 2002


I don't see that article in the results mathowie links at top. Actually, I'm not seeing a Textism page anywhere in the results now. Is that just me? (Not that I know how it could be.... I loaded 100 results at a time, and used Find to search each results page 'text'. No dice -- on any of them.)
posted by mattpfeff at 1:09 PM on May 7, 2002


There is a big difference between crawling and indexing. Crawling may happen daily, but indexing may only happen monthly. There are also several servers at google, which will give you differing results for the same queries, since they use varying parameters for the standard google algorithm.

New entries are entered into the google database during their indexing, when the crawled pages are processed and their pagerank values are updated. Some people refer to this as "dancing", because values move around. More info about the dance, and its frequency here.

You will NEVER EVER see changes in google rankings unless an indexing has taken place. Lately, indexings have happened every 3 weeks or so. So if you bomb the day after an indexing, you've got 3 weeks to wait. If you bomb the day before, you have one day to see your results.

You must also keep in mind that the pagerank (google ranking) of the referring sites) may rise over time, giving them a more important influence to google. This takes time, and may require several indexings for the values to fully have their effect.

This is a page showing the pagerank algorithm, in simplified form. There is also a good explanation of the dynamic nature of pagerank in this paper. If you look around page 11, you will find how the pagerank value grows iteration by iteration. If you keep in mind that iterations happen during indexing, you can see why it sometimes takes a while to hit your peak on google.
posted by websavvy at 1:15 PM on May 7, 2002


I think it's a question of the ratio of Google bomb links on a word or phrase to other occurrences of the same word or phrase on the Web.

Even now, after Matt's campaign against them, searching for the phrase "Critical IP" only produces 1,230 hits. Searching for "Verisign" produces 885,000 hits.

Textism's Google bomb simply has a much higher hill to climb.
posted by timeistight at 1:16 PM on May 7, 2002


I can validate that re-indexing hasn't occured since at least April 16th when I launched a revised version of a client's website. Should be any day now.
posted by machaus at 1:23 PM on May 7, 2002


But like any search engine that relies on automated indexing, Google's process can be taken advantage of, which birthed the conceot of Google-bombing.

So perhaps Google has established a human intervention process where a human identifies when a Google-bomb is happening and adds it to some kind of exceptions file that is maintained for the sole purpose of preventing Google-bombing.
posted by camworld at 2:10 PM on May 7, 2002


Google could very well have excluded the page based on its content. It wouldn't take much to kill from the index any page containing the phrase "Google bomb," either by hand or computer.
posted by me3dia at 2:32 PM on May 7, 2002


Interestingly, the textism article [ ./505/ ] shows up at #5 on a search for google bomb.
posted by me3dia at 2:36 PM on May 7, 2002


The target article [./494/] shows up third in a search for "hoopla.com" and first in a search for both "hoolpa" and "verisign".

It's seventh in a search for the phrase "Un fucking believable."
posted by timeistight at 2:52 PM on May 7, 2002


I'm pretty sure Google has a "this is a googlebomb" logarithm or switch or button the admin might press. I remember when I bombed Google for "Zora Neale Hurston" I quickly fell off the charts (literally - I wasn't anywhere. As soon as I took the bomb down, I went straight back up.

Maybe it picks up the word "googlebomb" or "google", and ranks accordingly.
posted by Neale at 3:17 PM on May 7, 2002


I'm with timeistight. There's nothing fishy going on. It's a bit of hubris to think that a few hundred sites with perhaps a few hundred links into them, by adding a the search term once to a page, would affect a search for a term that appears on nearly 900,000 pages, a few dozen of which probably have thousands of links in to them (how many "partners" does verisign/netsol have?) and a few hundred of which probably have a few hundred links into them and the search term repeated multiple times (i'm seeing IBM, Netscape, Microsoft, ZDnet, news.com in the first few pages). I do find it interesting that a Kuro5in article from early March appears within the top 50. But in that case, verisign appears on the page more than 50 times.
posted by dchase at 3:30 PM on May 7, 2002


We could run a test! My site isn't currently found in Google; neither is the phrase "shameless link whore."

If a hundred metafilterians put <a href="http://www.timeistight.com">Shameless Link Whore</a> on their pages and I become the number one "shameless link whore" according to Google, then Google bombing still works, no?
posted by timeistight at 3:57 PM on May 7, 2002


It's a bit of hubris to think that a few hundred sites with perhaps a few hundred links into them, by adding a the search term once to a page, would affect a search for a term that appears on nearly 900,000 pages

I'm not so sure about that. From what I've noticed a link from a metafilter or a slashdot is worth 1,000 links from a lowly ranked site. I don't think page rank is just about raw numbers.

I don't know exactly how google's system works and considering they're always tweaking or changing it I wouldn't doubt it.
posted by skallas at 3:57 PM on May 7, 2002


As an aside, I'd like to become the editor of this directory...
http://directory.google.com/Top/Society/Issues/Business/Allegedly_Unethical_Firms/
posted by FreezBoy at 4:12 PM on May 7, 2002


for a long time /505/ was ranked higher than /494/ on daypop, then /494/ went to the top for a good while, so that could explain why /505/ is coming up higher right now. (?)

/494/ is now only #25 on daypop though, and falling, i guess people are losing interest.
posted by rhyax at 5:15 PM on May 7, 2002


From what I've noticed a link from a metafilter or a slashdot is worth 1,000 links from a lowly ranked site.

No, it's not hubris to think that a link from metafilter or slashdot is influential (they clearly are). It's hubris to think that such links are more influential than links from news.com and zdnet and ibm and icann and microsoft and... No, it's not hubris to think that a link from a few hundred weblogs is influential. It's hubris to think that such links are more influential than links from thousands of other sights (how many pages are out there with partner links to verisign? how many pages out there link to netsol's whois?). And don't forget that the number of times that a search term appears on a page is part of the equation: It's hubris to believe that a single instance of the search term on some sites is more relevent than multiple instances of the search term on others.

It really is all about raw numbers, and the numbers include among other things, a combination of all of the above (links in, reputation of sites linked from, prominence of search term). Indexing by it's very nature is all about numbers. The magic is what the indexing algorithm does with the raw numbers. As timeistight noted, it's one thing to affect a search that produces 50,000 results. It's another thing to affect a search that produces 900,000.

I do think neale has a good point, though. Google probably has tweaked how pages that contain phrases like "google bomb" are scored, especially if the phrase appears near the search term.
posted by dchase at 5:26 PM on May 7, 2002


If I notice a creepy person in my building, having entered by some surreptitious means, I call the landlord and the locks are changed.

Of course they changed their algorithm.

This issue, being discussed with the number of page hits for this site in the open media raises any eyebrows? Come on folks, this site is on the 10 most wanted list for propagating this googlebombing thing and this action by Google comes as no surprise?

Now Now...Why is this even being discussed on open?

posted by lampshade at 5:59 PM on May 7, 2002


I'm more apt to go with the big animal/bigger trap theory. You forget that NetSol is only part of the whole Verisign picture. You are not only competing with links on the web that reference Verisign's domain registration business, but every single reference to PKI in the last five years or so. When you bombed CritialIP, you were using a small trap for a tiny animal. Verisign has been in business for a (web-wise) long time and must have millions of links in high profile pages on the web. I don't think that Google has changed their algorithm as much as the Textism G-Bomb underestimated the pervasiveness of Verisign links.
posted by eyeballkid at 9:19 PM on May 7, 2002


(arrgh) What I meant is the pervasiveness of the word Verisign linking to their front page or otehr pages on their site.
posted by eyeballkid at 9:22 PM on May 7, 2002


For added influence, participating bombers can also try linking misspellings of the name: Verising, Veriisgn, Versiign, Vresign, Verisngi, etc. Every bit helps, right? ;)
posted by brownpau at 9:31 PM on May 7, 2002


I dont know. There's some smart people working at Google.

According to SearchEngineWatch, in a recent article titled 'Google Bombs aren't so scary':

Google combats these [Google Bombs] by identifying what it considers to be "artificial" link structures and adjusting or eliminating their influence in the rankings. Google has also recently taken action against reciprocal link pages, link "farms" and guest books, downplaying their importance in its link analysis algorithms. And there's no doubt that Google will take action against weblogs, if those weblogs are seen as manipulating results in a way that doesn't correspond with user expectations.

posted by vacapinta at 11:38 PM on May 7, 2002


Oh well. Back to the class action law suit idea....
posted by ParisParamus at 1:30 AM on May 8, 2002


"For added influence, participating bombers can also try linking misspellings of the name: Verising, Veriisgn, Versiign, Vresign, Verisngi, etc. Every bit helps, right?"

Not if you're googlebombing for "Verisign" entries.
posted by websavvy at 6:16 AM on May 8, 2002


I love the idea that Google "takes action" against sites for not conforming to their ranking algorithms in the desired manner. It makes it sound like the world should conform to Google, not the other way around. Ernest Hollings should propose a bill in which it's a crime to link to something in a manner that's not pleasing to the search engines.
posted by rcade at 6:37 AM on May 8, 2002


while google rebuilds its overall index only every few weeks, they definitely incorporate information from frequently-crawled sites more often than that. as i recall, there was an interview with eric schmidt at about the time he took on the ceo position about how he wanted google to be more real-time so that searches on current events would return more useful results.

i think that also explains why googlebombs fade out fairly quickly -- a site like metafilter with a good amount of 'googlejuice' will get ranked highly on a search for something that otherwise doesn't have a highly-ranked page, but as the search-term-specific pagerank scoring kicks in, it quickly drops off the map. (when i mentioned some of the propositions on the recent california ballot on my weblog, i ended up in the top-10 google results for a little while because my site is generally well-linked. but i quickly fell off the chart as the pagerank algorithm 'realized' that nobody who linked to my site was linking to me from pages that mentioned those search terms.)

verisign probably just has an unassailable position on a search for its name by virtue of the number of sites pointing to it by name. i think a coordinated googlebomb against any major company, particularly an internet-based one, is not going to gain much traction.

(and i seriously doubt there's any human intervention going on. a few googlebombs from webloggers are nothing compared to what i'm sure are constant assaults by the "search engine optimization" crowd.)

need more evidence that google indexes more often than once every three weeks? search for subversive subliminal messages today, and inertial existence tomorrow. my weblog is #1 for the former today, and i'd put good odds on being #1 for the latter tomorrow.
posted by jimw at 9:41 AM on May 8, 2002


Some of you seem to be missing the fact that google indexes a large set of weblogs each and every day.

I seem to remember a few weeks ago a couple articles coming out in places such as Wired News and CNet about google-bombing. Does anybody think Google didn't do something about it right away? I was certain they were.

It makes it sound like the world should conform to Google, not the other way around.

It's a funny way to say it, but they're just trying to keep their search results accurate and useful. Can one really find fault with hypothetical measures they may have taken to lessen the effects of google bombing?
posted by sudama at 9:56 AM on May 8, 2002


There may be another reason why webloggers start out with a high PageRank for a particular term:

I believe that Google ranks the home page of a domain higher than subpages and it ranks content high on a page higher than content further down. (I can't remember where I read that.)

When add a new item to your weblog, it's usually at the top of your home page. As you add more items, it slips down the page and then off into an archive.
posted by timeistight at 10:10 AM on May 8, 2002


Home pages do well on google because they usually the target of inbound links. Those wishing to googlebomb would be well served by linking from their main page to their googlebomb page, using the desired text.
posted by websavvy at 10:28 AM on May 8, 2002


I don't think Google adjusted for this bomb. People forget just how many parked domains point at Verisign, and how many sites with 9-level PageRank (any major news site) have linked to Verisign for years. Either one of those alone would overwhelm even the biggest and most popular weblogs, all working in concert.
posted by anildash at 11:10 AM on May 8, 2002


I'm with John on why it's probably not working. Bad title tagging. Update the first textism page to include "Verisign" prominently in the title tag, and see what happens.
posted by rusty at 11:07 PM on May 8, 2002


John reports that
Textism's Verisign Google Bomb has over 229 weblogs pointing to it, many of which have significant traffic.
but a link search on Google returns 18,200 pages that link to www.verisgn.com.

If we assume that the PageRanks more or less equal out (I know they don't, but there's no way for us to calculate which group's PageRanks are higher) then wouldn't Dean's bomb need another 18,000 weblogs to sign on to be effective?
posted by timeistight at 9:03 AM on May 9, 2002


okay, i was wrong. i only hit #6 for inertial existence.
posted by jimw at 9:36 AM on May 9, 2002


Hey Arthur... actually, those page counts are pretty funky. I tried to work that into my post on the Google Bomb, but the data seemed weird to me.

I did that same link search... if you click into it, it turns out that there around only 515 pages after all?

That doesn't seem like an unstoppable Google Bomb to me... especially since I bet most of those pages have much lower page ranks than most blogs.

But what got me most about the Google results is that some of the links I clicked on didn't have links to Verisign.com. Still trying to figure that one out...
posted by kaname at 10:02 AM on May 9, 2002


Yeah, there’s definitely weirdness behind the scenes in the link search thing.

John: how did you get 229 links to www.textism.com/article/494/? When I search Google for links to that page, I get “did not match any documents” even though Google has the original call to arms page cached.

Ultimately, though, doesn't the effectiveness of a Google bomb depend on the ratio of Google bomb links to other references? Won't it be a lot easier to Google bomb "Willie's Wacky Widgets" than Microsoft or General Motors?

posted by timeistight at 10:50 AM on May 9, 2002


Screwed up the link search link. Should be:

http://www.google.com/search?as_lq=www.textism.com%2Farticle%2F494%2F&btnG=Search
posted by timeistight at 11:23 AM on May 9, 2002


Google's "link search" consistently gives me strange results. Anyway, I used Blogdex to get at the number of blogs bombing Verisign:

http://blogdex.media.mit.edu/linkinfo.asp?url=http://www.textism.com/article/494/

It's up to 239 now... pretty impressive display of blogger power!

As for the ratio of Google bomb links to other references... that's certainly a major driver. The other one would be the relative Page Ranks of each page linking to the bomb. That's one place where Blogs really shine, thanks to blogrolling and cross-blog linking.

Google Bombs are pretty damn powerful, which is why it looks like Google took some action to filter the big bombs out of their search results.
posted by kaname at 11:42 AM on May 9, 2002


You conspiracy theorists may care to know that the specified link now comes 15th in Google's list, behind some heavyweights.

Gee whiz, maybe it did take a month or two for the results to show up after reindexing. Who ever would have thought that?
posted by websavvy at 12:08 PM on June 7, 2002


« Older commentary-cutting link view pony?   |   Please add *.met4filter.org to the DNS tables Newer »

You are not logged in, either login or create an account to post comments