Slashdot Mirror
Google's Search Results Degraded?
scrm writes "According to this Wired article, recent tweaks to Google's PageRank search algorithm have degraded rather than improved the accuracy of the results." I noticed this firsthand the other day, but only when I was searching for pictures of famous people, but all my technical queries came back fine.
Yes, the web should be indexed by people.
But how about not one that requires sites to pay to get in? (Yahoo)
At Google, go to the "Directory" tab, or go to DMOZ.org (Open Directory) itself. DMOZ is bigger, better organized, has fewer broken links, no ads, and is built by hand by people who know their categories and are interested in keeping them linking only to sites with meaningful content.
Semi-mindles search spiders are not all there is in finding stuff on the Internet.
I had been searching for information on US bills that are still in circulation, and I found that google did a poor job in finding what I was looking for. The catch is that I did the same search about two months ago, and I got all kinds of decent sites from google.
Well, so much for PageRank. Im afraid this is whats going on: Link base PageRank, like 1 person per vote democracy, was turning dangerous for the few with enought resources to put some pressure at google. Lest see how they solved it. Problem: http://www.google.com/search?q=%22go+to+hell%22 Microsoft was #1 at "go to hell", thanks to popular vote like linking. AOL was #3, Disney #4. Solution: http://www2.google.com/search?q=%22go+to+hell%22 They are no counting any more Anchor links from "no-authoritative" web pages. Joe Doe pages dont count any more. Only Anchor links coming from pages oficialy "recogniced" at big sites whit a superior page rank to start with, or directories like Dmoz or yahoo count now. If this is the case, we can say PageRank is DEAD. From now on, big corporation marketing rules over popular choice. take this as a example: http://www.google.com/search?q=correo+gratis "correo gratis" is spanish for "free mail". Hotmail was #1. Now, at www2, is nowhere to be found. Hotmail is pagerank 9, and hundreds of spanish web pages where pointing at it as "correo gratis". Now is not, but is still #1 if you look for "free mail". Why? joe doe pages dont count, hundres of spanish users linking at it dont count any more. Only the "official", msn network pages count now, plus the few Dmoz pages pointing at it using that text as links. Most of those pages happen to be English Only, so only the english version of the query survives. Oh well. .
There is a company called Search44 which seems to have made this kind of stuff their living.
They index lots of other sites pages and when google comes around spidering they return random content from them. If you follow one of these links from google you will be redirected to their portal.
No doubt it gives them quite a bit of traffic.
I notice that google gives a date with the results, and that this is when the googlebot, apparently, visited the link. As a test, I put my keywords in over several days, and google returned the page I was looking for with a date that was, in most cases, only a few days old, together with a short sentence from the beginning of the page. I had thought that googlebot took a month to go around the internet and list all the pages, but from this date thing, I would think that googlebot only takes a few days at the most. Try this: put "gary gray tropical discussion" in google. I just did, and the date returned is 10-1. I didn't put it there, googlebot must have, IMHO. Very nice, anyway, and google and search engines that work with a bot remain my favorite.
Rapidweather's Linux Screenshots.
This isn't Mozilla, it's Netscape 3.04 which is less prone to make things up -- when it gets nothing from a server, it usually whines "NS is unable to find the file named blah-blah".
Tho I did try in Mozilla as well, same result. THEN I checked docsource in both browsers, and saw what I noted in my previous post.
The live pages on the SAME server were behaving normally in Netscape, so it's clearly not a connexion problem (in NS, that usually produces the "connexion reset by peer" complaint). When I went to their root site and tried looking thru their links and site search, it appeared that the entire branch I'd hoped to read had been deleted (anyway, no reference to it anywhere in sight) -- assuming it ever existed!! There was no scripting involved, and no reason for anyone at this site to give a damn where they're ranked.
If you're wondering, I was looking for historical maps incrementally showing the shoreline for Devil's Lake, North Dakota, to supplement what I've found in ancient atlases.
~REZ~ #43301. Who'd fake being me anyway?
At Google, go to the "Directory" tab, or go to DMOZ.org (Open Directory) itself. DMOZ is bigger, better organized, has fewer broken links, no ads, and is built by hand by people who know their categories and are interested in keeping them linking only to sites with meaningful content.
First, I would suggest going directly to the categories at dmoz.org rather than the Google relistings. Google picks up revised RDF dumps from DMOZ whenever they please, but the lag in the cycle is pretty long. If you are looking for the "fresher" data, go directly to the source.
Second, DMOZ can become what you say it is only with proper editing. The project itself may list 50000+ editors, but they're volunteers and there is a lot of ground to cover. A large number of edits are made by those "high up" in the directory structure to "lower"/"deeper" categories less well understood. Certain branches of the project are neglected; others eat editors for breakfast with the amount of work that needs to be done. Volunteer and help out.
You may also want to investigate ChefMoz and MusicMoz, too.
I noticed that google is plural-sensative. For example, "SQL alternative" will give a bit different answer than "SQL alternatives".
It does not seem like a very good idea to me in most cases.
Table-ized A.I.
The pagerank algorithm is one of the most important reasons why it is so good in bringing up relevant and popular results. But, this is just one of the ways of searching for good results and will not always work to your satisfaction.
Google gives preference to the number and quality of links to a particular site rather than the content of the site itself. One can easily come up with cases when this is probably not the best approach.
For example, consider a portion of the web containing lyrics of songs. If you search by artist name or song name, google will return excellent results, because the pages are probably linked using the names. However, if you only know the soneg from radio, you might want to search for songs containing a few particular lines. The pagerank algorithm might not be the best fit here.
All your favorite sites in one place!
I wonder what the folks at Google are thinking when that shows up as a high-ranking source of traffic... I also wonder why somebody felt that was neccesary to even put. Ah well.
SIG: HUP