68 replies to this topic

[Irony]

Thanks!

and 11 (yes, 11) is the highest

But the indicator shows PR only up to 10. How do we find out that a certain site got 11 then?

End of the day, if your PR went down or up by one, I wouldn't worry. If you are seeing a significant drop in traffic, then it may be time to start looking at your strategy - but again - don't totally rewrite your strategy in panic.

Sure not. Though our PR went down from 6 to 5, our traffic suddenly increased, and our site is often found for search terms I've never thought of. Our English site, I mean, not the Russian one. I should consider joining the Russian mob, I think

What's worrying me, people at other SEO forums are earnestly discussing on resuming obviously spamming methods. They are angry. It is a side effect of this update, and it is real. As for our PR I only need to know how to explain it to my boss next time he asks me about it

[Grumpus]

You can explain it to your boss as I explained it. It's a bit simplified, but your bosses would need to understand PR in order to get more complicated.

PR is a relative number that is done on a curve. Think of it as grading a test score.

On test #1 the highest grade is 70 the lowest is 30 and the average grade is 52. 52 merits you a C or a PR5. Last month you had a 58 which got you a C+ (or PR6).

On test #2, the highest grade is 78 and the lowest grade is 28. The average is still 52, though. 52 gets you that PR5. You still have a 58, but this time, in the curve, it's only a C (PR5) - even though it's still a 58.

The traffic to a search engine, though seasonal, is relatively constant. The number of inbound links you have is relatively constant. PR is not a constant. Your traffic, in most cases, isn't related to the PR (well it is, but not so much as some other factors), is based upon the traffic of the search engine and where you place for the terms people are searching for.

----

In respects to the PR11...

The PR numbers we see are split up into 10 numbers. I'm going to simplify this a bit. (In reality the distance between PR4 and PR5 is likely twice the distance between PR3 and PR4, but let's not worry about that). In the numeric system used by the toolbar, there are actually degrees of each PR - we just can't see them. a PR of 4.8749 is much greater than a PR of 4.000006 - yet both will show up as a PR4.

Now, as I said, there is only 1 page out there that has (penalties aside) a natural PR of 0. It is the low end of the scale. There are other pages with 0.0002 and 0.89765 that appear as 0 in the toolbar, but there is only 1 that is at the very bottom with an absolute PR0. (Assuming there are no tie scores, which could happen, I suppose).

The same is true at the top of the scale. Only 1 page can have that absolute highest value on the curve. Since there is more than one page out there with a toolbar PR10, we have to assume that there are varying degrees of PR 10, too - so one may be 10.2784 and another may be 10.87432. These numbers are greater than 10, so the scale must go to 11.0 (or 10.9999999999999999999 <- Which eventually "becomes" a number that is actually 11 if you are a mathmetician). If that weren't the case, then there would be one site with PR10, we'd know what it was, and all the other PR10 sites we now see would be a PR9 (9.9999837 or whatever, but still showing as a 9).

So, like the amplifiers used by Spinal Tap, this one goes to eleven.

G.

[Tribe]

Grumpus,

Great article...filled in some blanks for me while being easy to read.

Have a great day all!

Rebecca

[Ron Carnell]

Okay, I'm a little confused here, Grumpus. Do you have a source for any of this you can share? My understanding of PR is significantly different from this description.

Actual Page Rank is calculated on a curve only in the sense that everything statistical can be graphed on a curve. All pages on the Internet start out with the same exact PR, generally thought to be equal to unity. As "votes" are cast, those pages with the most and best links will have their PR increased. If every page on the Internet linked to this page and this page only, the PR of this page would be about three billion multiplied by the dampening factor. All the other page would remain at unity. Yea, that's a curve, but it certainly isn't a bell curve. Of course, in reality, we probably get much closer to something resembling normal distribution, but it's nonetheless NOT based on averages or means.

The GTB is simply a representation of actual PR, designed to help us "see" actual PR, and it's curve is generally thought to be pseudo-logarithmic. We don't really know what the PR is of a site with a GTB rank of 5, we only know that it corresponds to other sites with a GTB rank of 5. Go to the Google Directory and look at the Page Ranks depicted there. They, too, are only a representation of actual PR, and they do NOT correspond exactly to the Tool Bar scale (instead of going from 0 to 10, the Directory PR runs from 0 to 7). A logarithmic scale only tells us the hill is going to grow increasingly steeper, making it much more difficult to go from 4 to 5 than it was to go from 3 to 4 (many think four times harder, but no one outside Google really knows the scale). To suggest that "the average site on the web is PR5" essentially means there will be more PR5 sites than PR4 sites and that simply contradicts the math.

It is possible, and maybe even probable, that the number of pages in Google's index will affect the GTB because it is, after all, a scale with a beginning and an end. That does NOT, however, mean that the number of pages in the index will affect actual PR. Nothing about Page Rank depends on averages or means.

If we think of Google's PR as a voting system, which was the way it was designed, it becomes more clear, I suspect. No matter how many people who turn out to vote next November, we will NOT rank the Presidential candidates on a curve. Fortunately for the American public, this analogy quickly breaks down because ABC, CBS, et al, haven't yet come up with a Tool Bar to help us "see" the election results.

So, like, what do you know, Grumpus, that I'm apparently missing? Please tell me your post was just the result of not enough coffee this morning, 'cause this confusion you're causing me is very uncomfortable.

[Grumpus]

Nothing really. Tis true that it's not really a curve, but without the reader understanding the complexities of how it all works, it's easiest to use that as an example they can relate to. When you really break it out, it's divided among 10 sectors (representing each level of PR) And what sector you fall in is affected more, in reality, by the size of the sampling and not the high/low/mean of that sampling is.

I suppose maybe it's a poor example as it's somewhat misleading, I've just never really heard another explanation of it done in a way that the average person can understand. So, I've used that example just to show that it really is a relative score and not a fixed one.

G.

[designbug]

Very good information, thank you. I'm on the panic list as I dropped from #1 to #17 - OVERNIGHT!!! -

Been keeping an eye on things since Saturday morning when I was #1 and then Sunday down to #17 and fluctuated a bit 'tween #17 and #14. But things seem to have stopped moving around when I check.

How do we know when it's over and things have truly settled to where they're going to be until the next round? I don't want to jump into messing with things as they were so good before.

Problem is, I had redesigned my site and put the new version up like 2 days prior to this dance mess. Had moved a few pages around, didn't do much content changing as I didn't want to mess w/what was working. I did Permanent Redirects to any moved pages. This dance picked up the new site and so that's the one it seems to have been working off of. Because I had to many glitches to work out in the time I had left, I then put my old site (that had such great results going for it) back up, removing the new version. I hoped Google would then pick up the old version and put me back where I belong. So far, it's not.

Now, I don't know what to do. What are your suggestions for keeping the old for now or putting up the new? Again, no Real Content changes, just additional pages, new graphics and layout, added title tags to images. Nav moved from across the top getting rid of the javascript, went to new CSS nav and nav links on the right instead of the top. The pages that I moved, I put back where they originally were because the Google PR didn't show up on them and they had good PR.

Open to any and all suggestions. I had been getting some nice estimate requests off my site prior to this dance.

Thanks,
Deb
<edited colored font. Please use color for emphasis only! thanks- >

Edited by Scottie, 28 January 2004 - 02:04 PM.

[smc_online]

Grumpus,

Thanks for the easy to follow explanation.

But, most importantly, thanks for taking the time to educate us and give us peace of mind!

[Grumpus]

Thanks for the easy to follow explanation.

Bear in mind that Ron is exactly right - and that I probably did underqualify my statement about PR being "on a curve". That is really inaccurate when it comes down to the big picture. It works on a general level though - if you aren't interested in the specifics. There are far more PR0 and PR1 pages than there are PR2 pages, and so on. So PR5, as used in my example as the "middle" isn't even accurate.

Think of that explanation as a metaphor more than an analogy.

G.

[Irony]

Grumpus and Ron, thanks for the explanation. It is really easy to understand, and I feel more sure of my grounds. Simple math always brings consolation.

[eCommando]

How does google determine which site to put the new spam filter on?
I am still seeing some sites with many spam. For example, there's a site with hidden links (1 point dot as anchor) on the page and it's ranked #1 for a very competitive word.

[smc_online]

Think of that explanation as a metaphor more than an analogy.

Got it! I am of the mind set that PR is BS anyways!

By that I mean "beyond stressful" and without enough "balanced substance" to weigh it out. I go nuts just trying to read a small blurb on it! There are other things that is within our grasps that I can spend the energy on.

So I'm with you!

[designbug]

Thanks for all the input, but I would really still like some help/answers with my two questions, please?

1) How do we know when it's over and things have truly settled to where they're going to be until the next round?

2) What are your suggestions for keeping the old for now or putting up the new?

These 2 questions seem to have been overlooked.

Thank you!
Deb

[Grumpus]

1) How do we know when it's over and things have truly settled to where they're going to be until the next round?

It never really completely settles down. Your sector may be sooner or later than others. When most of the spam is gone and the results aren't hopping wildly around for a full 48 hours, then you can be pretty sure your sector is going to be good for a bit. Looks like several if not all of the sites I watch are still very much in play, though. There's no easy answer - primarily because there's a lot more going on in the last two updates than the ones we were used to.

2) What are your suggestions for keeping the old for now or putting up the new?

What are you putting up the new for? Your visitors? To make the site better? Or to rank well in search engines?

And, if you don't put the new up now, when would be a good time?

If you're making changes to make things better, more logically set up, and easier for the users, then any time is a good time to do this (though I'd qualify that with making sure folks aren't getting any 404's and other problems if they are coming in from links on search engines and other sites). If you're doing the revamp because you've changed your tack on SEO, then make sure that it's the tack you want to be on and commit to it - and now is as good a time as any. If you are making the radical changes because you're panicked about what's happening at Google and you're just doing it to try something that might save you, you're already sunk. Not only is it important to have a plan, but if you don't stick with your plan and just make slow and steady changes to make the site better, you're going to spend your whole life playing catchup and be fixing problems rather than learning to anticipate and be ready for the problems.

G.

[Ron Carnell]

I think Deb's second question can be clarified a little by this thread.

You correct me if I'm wrong, Deb, but I get the sense you changed your site design (not the content) a few days before the Google hit the fan, and then panicked a little when you lost your rankings in the SERPs?

I think your real problem is one of expectation. I'm guessing you don't yet realize that there is a lag, sometimes a very BIG lag, between your actions on your site and what Google does as a result of those actions. With over 3 billion pages in its database, things just don't move that quickly at Google.

I'm guessing your new site design was probably never even spidered. The changes you made likely had nothing at all to do with your loss of rankings, therefore putting back you old pages also had no effect. You're just stuck on the same merry-go-round as so many others, and like it or not, patience is the only way to survive the ride.

Grumpus gave you some good advice, I think. If you sincerely believe your new design is an improvement over the old one, you need to just go for it. It's probably going to be at least a month, and possibly two months, before you find out if Google agrees they're better. You can check your server logs to see when Google stops by to spider the new pages, and you can check the Google cache to find out when they're indexed, but frankly, even those are a waste of time. Put up your new pages, then move on to the next project. If you have more than one thing in the pipeline, the waiting gets a whole lot easier.

[designbug]

Re: spidering

If the links that show up in the searches picked up the new pages, doesn't that mean it was spidered/indexed?

The site design was changed, not the content. New graphics, moved things around a bit but left the content pretty much untouched except for a couple of places in the meta description where I found unintentional repeats of words and thought of spamming so removed the duplicates. I was doing just fine in the SE's before this so no, my intent was not to do better in the SE's just to stay constant. The old design was like 2 years old, got sick of looking at it and wanted a fresh new look. That's it, period.

Ok, thought of another pertinent question ... during this upheaval and then the background work at Google's servers for placement ... is there just one big sweep or do the spiders continuously revisit sites daily until they're "done?" I thought they kept running around for several days and then settled into doing the ranking part of things. But now I'm confused on this issue.

So far, the pages of mine that are showing in the SERP's are the new ones even though I put the old site back up. So, the lag thing mentioned may not be accurate? I have found not only on my site but others I've done that there generally isn't much of a lag once the site has been picked up by Google. I can do a site or page, let Google do it's thing and then make a small change and it shows up w/in the week usually.

Thanks,
Deb