Forum Moderators: open

Message Too Old, No Replies

Update Brandy Part 3

         

GoogleGuy

7:41 pm on Feb 15, 2004 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



Continued From: [webmasterworld.com...]

"Any clue as to the possible role greater reliance on semantics is playing in your never ending quest for more relevant results?"

I'd say that's inevitable over time. The goal of a good search engine should be both to understand what a document is really about, and to understand (from a very short query) what a user really wants. And then match those things as well as possible. :) Better semantic understanding helps with both those prerequisites and makes the matching easier.

So a good example is stemming. Stemming is basically SEO-neutral, because spammers can create doorway pages with word variants almost as easily as they can to optimize for a single phrase (maybe it's a bit harder to fake realistic doorways now, come to think of it). But webmasters who never think about search engines don't bother to include word variants--they just write whatever natural text they would normally write. Stemming allows us to pull in more good documents that are near-matches. The example I like is [cert advisory]. We can give more weight to www.cert.org/advisories/ because the page has both "advisory" and "advisories" on the page, and "advisories" in the url. Standard stemming isn't necessarily a win for quality, so we took a while and found a way to do it better.

So yes, I think semantics and document/query understanding will be more important in the future. pavlin, I hope that partly answers the second of the two questions that you posted way up near the start of this thread. If not, please ask it again in case I didn't understand it correctly the first time. :)

NeverHome

7:30 am on Feb 17, 2004 (gmt 0)

10+ Year Member



The www results I see just flicked back to something resembling normalcy, and with fresh listings dated 15 Feb.

[edited by: NeverHome at 7:33 am (utc) on Feb. 17, 2004]

steveb

7:30 am on Feb 17, 2004 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



Wild guess, this may be a rare opportunity to see sites ranked by only ONE algorithm factor at a time. I'm going from 2, to 57, 404, to 1, to nowhere, to number 11. Unbelievable.

[edited by: steveb at 7:32 am (utc) on Feb. 17, 2004]

djgreg

7:31 am on Feb 17, 2004 (gmt 0)

10+ Year Member



vrtlw: Not in my area. I'm seeing 64 results on all datacenters for the keywords I am monitoring. Of course there may be a change in your area.

steveb: Obviously we are not seeing the same. For me the IPs you mentioned above look all the same and like 64, no matter how often I reload or switch the IP.

Very confusing that people see so different things.
BTW this is the first time I am seeing Senior Members posting "I am seeing this on that datacenter" threads ;-)

vrtlw

7:37 am on Feb 17, 2004 (gmt 0)

10+ Year Member



This gets curiouser.

I flushed my DNS, pinged www.google.com and it returned the IP address 66.102.7.104

I typed that IP address into my browser and it searcherd for 'mytown mykeyword' and it put me where I expect to be.

Now, from www.google.com - which I have confirmed doing a 'netstat -a' is resolving to the same IP address, my listing is nowhere to be seen.

[added]This is the same for 216.239.53.99

If you search using the IP address you get stable brandy results. But if you search at www.google.com (which is resolving to the same address) the results are whacko. I am just playing with a hosts file entry and it initially appears that if you set the hosts file to an ip address for www.google.com the results are solid.[/added]

[edited by: vrtlw at 7:45 am (utc) on Feb. 17, 2004]

steveb

7:43 am on Feb 17, 2004 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



Seeing lots of examples of pages on domains like... suppose you had pages devoted to each US President. You type in Abrham Linoln and expecct to see your Linocln page in the top ten, and instead down at 407 you get your Warren G. Harding page that happens to mention Lincoln.

NeverHome

7:51 am on Feb 17, 2004 (gmt 0)

10+ Year Member



It is certainly amazing to watch. Even though my site flicks in and out of the serps I'm not concerned at all - the whacko version is SO bad it is hilarious. I look forward to hearing Google Guy's take on this one! :)

vrtlw

7:53 am on Feb 17, 2004 (gmt 0)

10+ Year Member



MMMmmm

Back to normal now. Let's hope it sticks!

Kirby

7:55 am on Feb 17, 2004 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



Using steveb's example - even worse are the pages with just links to Lincoln, but not Lincoln the president - but Lincoln towncars.

NeverHome

8:05 am on Feb 17, 2004 (gmt 0)

10+ Year Member



Forget the Semantic Web. Welcome to the Demented Web! A search for <snip>

I guess they key to this pages success is no SEO tricks, and no links to "dark corners of the web". haha.

[edited by: Marcia at 8:17 am (utc) on Feb. 17, 2004]

Oops, sorry Marcia. I'll paraphrase the page content then... ERROR 404

;)

[edited by: NeverHome at 8:20 am (utc) on Feb. 17, 2004]

Rachel2

8:06 am on Feb 17, 2004 (gmt 0)

10+ Year Member



I have read some people have lost rankings and think it is googles way to boost up there adwords. I buy adwords too, but wont continue if search results keep getting worse and junk sites. Google users will drop off as well, so will put my money where the best engines are. If there is no one searching google, then adwords have no value.

Google has put to much weight on page linking. While so many webmasters have purchased links on high pr rankings sites just to get there page ranks higher, does not mean they have a high quality site. Good quality site should have nothing to do with who is linked to you.

My opinion is google had good quality sites, when you did a search you found exactly what you wanted. Then they changed it. Sad to see a good search engine keep getting worse and worse.

Hissingsid

8:11 am on Feb 17, 2004 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



Hi,

On 64.* I'm seeing some interesting results when I look close.

For my main 2 word term which was a problem from Florida onwards now on 64, 14 of the top 20 slots are pages from "real sites" and only 6 are auto generated directories.

For an example three word term which was fine until Austin and then filled with pages from generated directories at that update. On 64 just seven of the top 20 slots are taken by real sites the rest are still auto generated directories.

64 still looks like a work in progress to me and I still expect a rough ride for the next few months as the Google engineers leanr how to play tunes on their new instrument. Whatever was done at the Austin update was IMHO very bad indeed but what is on 64 is not a roll back it looks to me like the same Florida algo with significant ajustment. I'm hoping they've put a big block capital note next to the code that caused the Austin problems saying ### DO NOT TOUCH THIS OR VERY BAD THINGS WILL HAPPEN ### and we won't see that again.

Best wishes

Sid

PS I'm starting to wonder if the Guys at Google were so excited with their discovery that they could actually make the new algo create good results that, having had a big cigar and 5 star brandy they sent GoogleGuy off to his computer to tell us all about it and then realised to late that it was going to take longer than expected to roll the thing out because of all the DNS faffing that they've been doing recently.

PPS Leave off Bobby whoever it was flickering him (I was going to say flaming but it was a mear flicker) . Just because you don't understand LSI doesn't mean that it isn't having a big influence on all of the updates since and including Florida. In LSI and by extention CIRCA document means something much wider than the old Google definition of document = page.

djgreg

8:16 am on Feb 17, 2004 (gmt 0)

10+ Year Member



Google has put to much weight on page linking. While so many webmasters have purchased links on high pr rankings sites just to get there page ranks higher, does not mean they have a high quality site. Good quality site should have nothing to do with who is linked to you.

I think it has to do with who links to you.
You are right saying that there was too much emphasize on anchor text, but I think this is solved looking on the 64 results. For me they seem very clean and the Top results for example in the travel industry, where one of my sites is located are not the ones with the highest amount of backlinks but the highest quality sites.
I think determining the value of a page by the sites linking to it is the best way they can do it at the moment. In the future the LSI will be a very powerful tool to be added to the Pagerank and Anchortext criteria, if not done on 64. Not sure about that, yet.

[edited by: djgreg at 8:22 am (utc) on Feb. 17, 2004]

vrtlw

8:22 am on Feb 17, 2004 (gmt 0)

10+ Year Member



Im still seeing very different results on www.google.com

This time instead of being dropped from the serps I am at a new all time high and am very near the top of page 1. I have never been this close to the top for these terms and even on 64. I am listed on page 2

adfree

8:23 am on Feb 17, 2004 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



I have been waiting for a resolution of a specific spam type within the German Google SERPs - even this update does not take care of it.

Some sophisticated techniques that I can not figure out make Google unuseable for major keywords. Looking for something local or regional in combination with descriptive keywords that can not be narrowed down any further the SERPs are dominated (first 100 or so results) by just three guys (maybe it's just one with varying formats), cloaking like crazy.

I have switched over to other SEs for that matter, many of my friends and colleagues have too.

GG, where can I drop you examples to have a close look as your sticky is off?

Thanks, Jens

crobb305

8:33 am on Feb 17, 2004 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



Can someone ref the LSI paper? I have searched through the thread with no luck.

Thanks

This 327 message thread spans 22 pages: 327