Florida theories are wrong?

PhilC · 12-03-2003, 07:28 AM

I've been doing some extensive study into Florida, and I am now becoming convinced that it is *not* an seo filter.

The results that we see when we use the -garbage search method are the 'normal' results that we would have seen had the Florida changes not been implemented. They are current and have been updated, although many serps will look the same as before Florida.

The results that we see when we don't use the -garbage search method are a different set entirely. Pages are not dropped from the 'normal' results, as an seo filter would do. It's a completely different set.

I'm basing this conclusion on my examination of all the results that Google will show for both 'normal' (-garbage) and 'Florida' searches for a number of searchterms. For instance, one searchterm produced 869 'normal' results, of which 624 did not appear in the 'Florida' results. Another has 619 missing results in 'Florida' from 924 'normal' results. That's a hell of a lot of missing listings. Judging by the URLs of some of the missing listings, e.g. a BBC page, all the missing URLs cannot have been due to tripping an seo filter, and we still see some highly SEOed pages in the top 10 of some search results.

Further evidence is that I see wild changes in the rankings between the 'normal' results and the 'Florida' ones. E.g. a BBC page is ranked over 800 in the normal results, but turns up at #1 in the Florida results. That cannot be the result of dropping some SEOed pages ahead of it. Yes, the searchterms are different when adding the -garbage element, but it's a hell of a big swing just the same.

Other searchterms also have high percentages of URLs that appear in the 'normal' results, but don't appear in the 'Florida' results, and some of the missing URLs are such that they must be squeaky clean. So I have to conclude that the Florida results are a different set entirely, and are not the normal results that have been modified by the dropping of SEOed pages.

If it's a completely different set of results, then it's not LocalRank, because LocalRank works on the original set, it doesn't produce a set of its own, and it doesn't drop pages from the set. It's much more likely to be Hilltop, a developement of Hilltop, or something along those lines.

If the change is along the lines of Hilltop, then it makes it even harder for us than we'd thought. LocalRank would have been easy to deal with but a Hilltop-like system isn't as easy, although it can be done. The quickest way to deal with it is to seo as normal, but for minor searchterms that Hilltop's system can't produce a set of results for. To understand it, see this thread:- http://www.internet-marketing-resear...pic.php?t=1993 and read the paper that John points to.

webmastery · 12-03-2003, 08:20 AM

A dictionary of money words has been added. To me it's the only likely answer.

People are not returning to their former positions because their domain now carries a penalty for those particular phrases. I bet those positions will be lost forever.

PhilC · 12-03-2003, 08:52 AM

No. It's not a filter - money-words, or any other kind. It's a completely different set of results. A Hilltop-type, expert-based system would account for several things that people are seeing:-

(1) The idea of a list of searchterms. It would appear like that because that type of system cannot create a set of results for all searchterms, but only for some of them. That's why people conceived the idea of a list - because the serps for some searchterms were the same in both 'normal' and Florida results, but not for others.

(2) It would account for people seeing fewer commercial pages at the top of serps, because expert pages (not ranked themselves) would tend to point to information pages rather than to commercial ones.

(3) It would account for many home pages missing from the serps, because expert pages would tend to point to specific information pages within a site rather than to home pages.

dpam · 12-03-2003, 11:29 AM

Here come the Pseudo-Experts!

I, Brian · 12-03-2003, 12:24 PM

Good call on the research - and Hilltop does explain certain features - see the article Phil wrote here, especially about uni sites ranking better: http://www.search-this.com/google/go...lgo_update.asp

However, I'm still not convinced that any single explanation will work. Remember Googleguy effectively stated on WMW that 2 updates were being carried out - the initial one and then the big blast that flattened everybody? If Hilltop was the second datadump, then what was the first?

PhilC · 12-03-2003, 12:49 PM

That's the seo filter theory (it's Barry Lloyd, btw, not me

). I liked that idea because, if I were Google, I would have done it.

It may still be a combination. All I can say is that I am convinced that the 'normal' and 'Florida' result sets are completely different sets from the start, which means that it isn't an seo filter that's caused the disaster - at least not only an seo filter.

It may be that an seo filter is applied to the Florida results, but I still don't think so. I see well optimized pages ranking in the top 10. Certainly the results have been updated in the -garbage serps (the 'normal' results) and it may be that GG was refering to that as being the other update. I think that either GG got it wrong or maybe you used a word that he didn't use, because whether the change is an seo filter, a Hilltop-like or LocalRank implementation, etc. it's a change and not an update.

At this point I am convinced that the two result sets (normal and Florida) are completely different sets that are derived in different ways. The normal set does not transpose into the Florida set. I'll put some of my studies on the web for you to judge for yourself. You'll see what I mean. It'll be a few hours before I can do it, but I'll post the URLs here when they are up.

PhilC · 12-03-2003, 01:03 PM

The searchterms are stated at the top.

There are 2 tables on each page and at the bottom of each table are the figures for it. You'll see what I mean.

The Normal column shows the 'normal' rankings (e.g. search term - qweasdzxc). The Florida column shows the Florida rankings (without the -garbage element). Have a look at the results from top to bottom and tell me if you agree that the Florida rankings are not based on the 'normal' rankings. I said "from top to bottom" because the "internet marketing" report shows a good deal of similarity near the top, but that fades away and is coincidental.

I, Brian · 12-03-2003, 02:03 PM

Quote:

Originally Posted by PhilC

That's the seo filter theory (it's Barry Lloyd, btw, not me

). I liked that idea because, if I were Google, I would have done it.

Whoops - I saw your name up in the articles, but was pretty rushed at the time.

My point in linking to the article isn't the conclusion - it's the example, where he finds a particular keyword with unusually high rankings of "authority" sites:

http://www.search-this.com/google/go...o_update_2.asp

That, to me, fits into Hilltop.

**John Scott** · 12-03-2003, 04:14 PM

Damn, Phil, that is an awesome tool you made.

PhilC · 12-03-2003, 04:31 PM

Thanks John

Quote:

Originally Posted by I, Brian

My point in linking to the article isn't the conclusion - it's the example, where he finds a particular keyword with unusually high rankings of "authority" sites:

http://www.search-this.com/google/go...o_update_2.asp

That, to me, fits into Hilltop.

It does I, Brian. Authority pages are what the Hilltop system produces. Everything that we see appears to fit a Hilltop-type system.

Does Google own the Hilltop system? I know that the address of the guy who wrote it was Google's address in 1999, but do they actually own it? If they do, have they patented it?

**John Scott** · 12-03-2003, 07:24 PM

If it isn't an SEO filter/penalty, why did IMR drop from 26 to not_in_top_1000?

If it were a re-ranking, I could understand a drop from 26 to, say, 100 or 200. But it's gone entirely. It seems like a penalty in that way.

PhilC · 12-03-2003, 07:36 PM

For the same reason that loads of other pages vanished - they aren't linked to by more than one non-affiliated 'expert' page, or if they are, they are not linked to by enough of them of good enough quality (even the expert pages are internally ranked for the searchterm), to get into the set of 1000.

Mel · 12-03-2003, 07:56 PM

I can see some Hilltop like attributes to this new algo - like many more results from directories and edu sites and less from commercial sites, but I think there is more than that to it, at the very least it appears that if a site has excessive on page SEO it suffers more than most.

PhilC · 12-03-2003, 08:07 PM

Can you provide one or two examples Mel?

Have you looked at the 2 report pages that I posted? If so, would you agree that the 'normal' and 'florida' results sets are completely different sets that have been derived in different ways, and that the 'florida' sets are not derived from the 'normal' sets?

PhilC · 12-04-2003, 01:41 AM

I've noticed something too in the serps that's caught my attention. I've followed the phrases Adult Web Design and Adult Content Provider pretty closely in the past few yrs.. and on the old pre-florida serps, small sites came up on the first page from just optimizing their index pages.

Now, all I see there are webmaster resources capturing the top spots. All webmaster resources... got me thinking.. these sites are all optimized, but they all have something in common that small sites don't have.. they have a lot of links to specific resources, such as adult web design and adult content providers. In essence, they're link farms. They are a directory for adult services. So not only do they have a ton of incoming links, which small sites have too.. but a ton of outgoing links, which small sites don't have many of.

Mel · 12-04-2003, 07:55 AM

Hi Phil
Take a look at the SERPs for economical search engine optimization (which I used to have #1 on) and look at the PR4, two page site that ranks #1 out of 32000 results, And note that it does not have the full phrase in the page title, in the H1 tag or contiguous in the page text, nor does it seem to appear in the inbound links. Then append -googlefart to the search and compare it to my page that used to rank #1.

FWIW Phil I do believe that the two SERPs (with and without -googlefart) are different ranking sets, but I have not yet figured out what triggers the different sets.

dpam · 12-04-2003, 12:29 PM

By my read of the 'Hilltop' paper, and it doesn't seem like directories themselves should appear highly ranked, they are only used as the source of 'votes' for the authority pages. So the facts listed above about Adult stuff, and other comments that directories and .edu's appear highly ranked, seems off. In this world, you get rank by linking out to relevant pages - Google results show places to go get results, not results themselves. Makes no sense.

PhilC · 12-04-2003, 02:43 PM

Yes. The expert pages don't appear in the serps, not directly from being expert pages, anyway. They are merely the source of the pages that do appear in the serps.

But don't forget that expert pages are not selected by hand. They are any pages that have at least n number of links pointing to unaffiliated pages. They can point to commercial pages and they can point to information pages, but the majority may point to the latter.

PhilC · 12-04-2003, 02:48 PM

Quote:

Originally Posted by dpam

By my read of the 'Hilltop' paper, and it doesn't seem like directories themselves should appear highly ranked, they are only used as the source of 'votes' for the authority pages. So the facts listed above about Adult stuff, and other comments that directories and .edu's appear highly ranked, seems off. In this world, you get rank by linking out to relevant pages - Google results show places to go get results, not results themselves. Makes no sense.

Actually it does make sense in a way.. not saying I like it .. but it does make sense to me.. it looks like (to me anyway) what google is doing is becoming a direcory of directories. I only know the areas that I watch.. and this is what I see happening.. the search words/phrases that I've been watching for years have been taken over by places that contain an abundance of links to places that fit into the search. It seems very clear to me what's happened.. these webmaster resources have never listed that high in the serps for one of their categories.. and if you'll take a look, the links go to the exact page that the particular category is on.. not the front of the site

PhilC · 12-04-2003, 02:59 PM

Also remember that the Hilltop idea is just a theory. All that I am confident about is that the -nonsense serps and the Florida serps are completely different result sets, and that the Florida results are not a modified -nonsense set of results, as all the other theories believe.

The Florida results are aquired another way, and a Hilltop-like, expert system appears to account for all the changes that we see since Florida began. If if really is Hilltop, then it must surely have been developed a lot since 1999, although the basic principles are probably the same.