update sequence

Will watching how our sites move during the update tell anything?

[Tartan75]

The dance will last several days. During this time, many of us will be watching their sites closely, and our sites will probably move up and down a few times during the dance. Is there anything useful we can tell from when/how our sites move during the dance?

For example, I have a theory (based on very limited experience), that pagerank changes may not 'kick in' until near the end of the dance.

[2_much]

Interesting question Tartan, I've often wondered about this but have never seen it discussed.

Often what I've seen happen is at the beginning, I have a lot of sites doing well. Then, towards the end of the update, some of them get booted.

To me, this indicates there is some manual intervention.

Any other ideas?

[rfgdxm1]

I find it exceedingly unlikely this is manual intervention, unless you happen to be checking on some extremely notoriously spammed keyword SERPs. For the most part on the SERPs I check they don't change that much from beginning to end of the update. I do have reason to suspect PR isn't fully factored in until the end of the update.

[entropy]

What I noticed during 1 update was that my sites appeared very high on the SERPs on www2/3 and then dropped towards the end of the dance, giving way to pages with higher PR. I can't be 100% sure of this, but I think what I was seeing was SERPs generated before the backlinks and PR had been calculated.

If this is the case, then observing SERPs during the dance can give you an idea of how good your pages are excluding the factor of non-circular backlinks.

[Tartan75]

Thanks for your thoughts. Interesting others think PR may only kick in at the end - and particularly that they think it, haveing observed it in a slightly differant situation to me.

The situation that got me thinging about it was:
On launching a website, I spent alot of time getting others to link, which over a 3-4 months they did. This meant for the next few updates my PR was rising. During most of these updates I didnt make any significant changes to the contents of my pages (with one or two exceptions - more on that in a minute). The patern of movement during every dance was virtully the same - Little movement, then shortly before the end, a significant rise.

Now on a handful of pages, I made some small but signifiacnt changes - for example on one dynamically generated page, I noticed I had made an error and mossed of a heading - that heading was a key key phrase for the page. Now that page rose thanks to freshbot during the month, but it also rose much earlier during the dance.

If others can verify this appears the case for them, it appears we can get a feel for how much influence on page changes are making, seperate from how much influence PR changes are making.

[daamsie]

Last dance I noticed the changes at the start. The only difference with the end was that the results were more consistent.. my pages never ranked any higher than they did at the very start of the dance.

I think the PR is factored in quite early from my own experience, but still being quite new to this game, that's not much to rely on :)

Last month's dance saw my site's PR jump a point and keywords that hadn't been doing exceptionally well when freshbot picked them up, surged to the top 2 pages. This was right at the beginning of the dance and not the end.

[Tartan75]

daamsie
your increase in page rank - was it due to high PR plinks, or low PR links, and what sort of PR did you have to start with?

My reason for asking is I wonder if perhaps there is some 'order' for working working out/factoring in PR.

[daamsie]

A combination no doubt.. I had one very high ranking page(PR8) linking to my site, a new listing in DMOZ (PR7) and a whole ton of pages around the PR5/6 as well. My backlinks increased from about 40 to 442. The ranking was boosted from 6 to 7.

This month that PR8 link has dropped (it was a monthly feature), so I'm curious to see whether the new links to our site will be enough to maintain the PR7. I am also curious whether the internal links within my site help maintain that PR. I wonder if more pages will be listed this month, because my PR has been higher and if so, presumably that could also help affect the PR of the front page. I am also hoping for a Google directory link to help fill the gap :)

[DaveN]

Just a theory

When I ran a data processing dept for a large computer company we would run a set of reports on the same DB depending on priority of the filters would dictate the final results,

example

1) index of parts used in the last 12 months
2) remove any parts of internal usage
3) group by manufacturer
4) apply filter 1
5) apply filter 2

i tend to think of googles dance has

apply new algo
apply spam filter 1 -> compare data sources
apply spam filter 2 -> compare data sources
apply spam filter 3 -> compare data sources

within % threshold's release new database
check WebmasterWorld re do results
check WebmasterWorld okay

DaveN

[shrewsbury matt]

Another Theory ...

I've been watching dances for 12 months on all datacenters not just www2 & www3, once the index has been copied to the first datacenter, those will be the results you see on www.google.com at the end of the week.

The results you see on www2 & www3 can vary depending on if they are pointing to a datacentre that has the new index or not.

Well it make sense to me!

[The Subtle Knife]

It's quite clear that from PR published info,
the google algo is "iterative".
So, over time it's get's more accurate.

So initially the PR will be way, out and
get more and more accurate.

Infact it' annoying - we want to believe we
have a high PR.... google is messing up our brains...

[Tartan75]

It could stand to reason that they process higher PR first. High PR will have more effect on itterative processes than low PR. My experience was generally arround the PR4/5 mark.

I havent been watching individual datacentres - will need to read up on then and start watching.

I appreciate PR is an iterative process. Is there evidence that we actually see the iteration, or do we just see it factored in once its all calculated? I am no expert, but I would be suprised if the comment 'So initially the PR will be way, out' is correct. I would have thought that they would start with each sites previous PR. That would make itteration become accurate far quicker. On the basis that most sites dont change substantially from month to month, then most sites would not be way out.