How the Digitalpoint Co-op weight works-my guess-How to get the most benefit

[t2dman]

How the Digitalpoint Co-op weight works is something not officially stated by Shawn. However, max 4000 pages, multiplier of around 3 and PR have all been mentioned.

So how about some actual numbers to back that up?
And what does it all mean for how we should structure our sites to make the most of this amazing network?
Co-op weighting (several of my sites)

site..PR..pages in Google..DP weighting*..multiplier
a----0-------4------------------14------------3.5
b----0----5740--------------13500--------- 3.375 (based on 4000 pages)
c----6-----595------------- 23975---------- 3.375 x 2 (approx)
d----5---31000-------------67500-----------3.375 (based on 4000 pages)
e----4------16----------------432-----------3.375 x 2
f-----4------60--------------1613-----------3.375 x 2 (approx)


* - excluding referral bonuses

What I can't figure out, is why the additional multiplier of 2 - sites b & d are both vbulletin forums, the other sites use no specific package. So it looks like if you specify that your site is an 'other' (ie non vbulletin), then you get additional weighting. This makes sense, as there are lots of 'Google filler' pages in vbulletin that don't get much traffic - ie site 'b' has around 2000 threads, yet 5740 pages in Google, despite trying to take out as many as possible via the robots.txt.

I do not have any of the other types of sites blogs etc, so I don't know what the formula is for them.

Benefit of one site v two or more sites

It is easier to get a Pr5 than a Pr6. To go from Pr5 to Pr6 is a factor of around 6 (Pr6 is around 6 times more powerful than Pr5 etc.) However, the co-op only gives you a weighting of one for every increase in Pr.

Therefore, better to have two sites of PR5, both sites having 4000 pages, than one site with 8000 pages that you have worked hard to get to say PR6.

PR5 x 4000 (4000 pages) x 3.375 = 67500 x 2 sites = 135000
Pr6 x 4000 (8000 pages) x 3.375 = 81000

Possible Co-op improvement

I consider that the weighting should be based on the PR of the front page based on the approx Google formula of Pr being a factor of say 6. ie Pr5 is 6 times a Pr4, and 36 times a Pr3 etc. This will give some very large weightings, but since everyones will be increasing, it will overall be more fair than the current method.

With the changed weighting method, it would probably be better to have the one site that you worked hard to get to Pr6/Pr7, than two smaller sites at lower PR's.

Referral Bonus

The referral bonus is not reduced from a referrers reported weighting. The referral bonus for site b 13500 is 1350, and on site f 1613, is 162. And the referral bonus has not not messed with the formula.

So what does it all mean

To get the maximum co-op weighting, get all your sites to at least 4000 pages per the Google link:www.domain.com (http://www.domain.com/)
Where possible, better to have multiple sites at max 4000 pages, rather than a single site with over 4000 pages.
Get the highest PR possible - goes without saying
Non VBulletin sites get x 2 weighting
If you have a vbulletin forum, have it on a separate url - ie a subdomain rather than on your main url

means your main non vbulletin site can have the x 2 multiplier.
means that you have an additional domain - easier to get a PR5 x 4000 x 3.375 = 67500 on both sites rather than just one site at 67500

It would make sense to have a weighting factor based on the PR log factor say 6, than the current PR log factor on the Co-op of 1.
I can't wait for the next PR update when my forum sitting at coop weight of 13500 goes to 67500.





I would love some comments and see some calcs for the non vbulletin sites ie blogs phpbb etc.

[t2dman]

...Taken me a good hour+ to write and figure this all out. Learnt lots in the process. Then I saw the another thread asking the question I have just answered....


What weight PR0 is given

My analysis begs comment as to how Pr0 is treated - I totally agree that it should be treated the same as a PR1, since in one case, my so called Pr0, is 6th for a very competitive term, and will probably be Pr5 in the next update.

[joeychgo]

I disagree with "Non VBulletin sites get x 2 weighting"

the reason I say that is this.

when google reads a vb page - one page of 30,000+ - it isnt likely to read it again for quite some time. So as far as google is concerned, the links are still there. This doesnt happen on a 12 page mini site. So my link on that 12 page site - will be there for a week or two - tops - but a link that google scans on a vb post - will likely be there (according to google) for months or longer.

if anything, VB should get a lil higher ranking because they contribute a longer life link to its receiver.

[t2dman]

With many of the pages Google has for VBulletin forums, the pages will have low PR, much lower than the PR on non VB websites. For instance, on my non VB site, most of my 595 pages are PR5, with several Pr6's (many external inbound links).

VB has so many navigation pages - ie page1 page2, ... that while they are separate url's, they are very unlikely to get found in their own right on Google, and with a PR minus 1 for each page in, are unlikely to have much PR left at the end of it.

I have done many SEO hacks to the standard VB package, (see Search engine optimizing VBulletin (http://forum.time2dine.co.nz/showthread.php?t=98)) so Google should have less pages, and each of those pages has much more PR. However, I still consider that the extra weighting for non VB sites is more than fair.

[Kadence]

I consider that the weighting should be based on the PR of the front page based on the approx Google formula of Pr being a factor of say 6. ie Pr5 is 6 times a Pr4, and 36 times a Pr3 etc.
But having the Co-op scale be as extreme as the Google scale understates the fact that the point of the network is to have links across many domains.

Also, given that PR is not accurate--the toolbar is out of date--it makes sense to try to diminish the impact of PR on the weighting algorithm. Making the weighting algorithm more sensitive to toolbar PR--which is what changing the scale to 6 would do--is undesirable.

This will give some very large weightings, but since everyones will be increasing, it will overall be more fair than the current method.
Not really true, since weightings are relative.


So my link on that 12 page site - will be there for a week or two - tops - but a link that google scans on a vb post - will likely be there (according to google) for months or longer.

if anything, VB should get a lil higher ranking because they contribute a longer life link to its receiver.
This also means that since Google spiders the pages less often, it is less likely to pick up on new Co-Op network links. Thus vBulletin links are actually less valuable in that sense.

[hurricane_sh]

Is this the first research on coop? Nice post.

[Kadence]

Is this the first research on coop? Nice post.
I agree, providing that data was a very nice thing by t2dman.

It is much appreciated, forgot to mention this in my earlier post :o

[t2dman]

But having the Co-op scale be as extreme as the Google scale understates the fact that the point of the network is to have links across many domains.The point of the co-op should be fairness as well. If I can break up an 8000 page site into two, and get twice the amount of weight, then thats not what I consider proper. If it takes 6 times the effort to get a PR6 than a PR5, and a person only gets rewarded with a factor of one, then its better to have many lower PR'ed sites say PR5, than aim for the higher PR - that's also not proper.

Bottom line, if its based on the Google API factors, then in my opinion its needs to be as close to the Google formula to stop any manipulation.

The co-op is powerful enough to warrent making FULL use of it!

While the Google PR is certainly old at any one time, at least it does provide some form of rating.

ps - I appreciate the thanks guys

[schlottke]

Well, think about it this way...

if you have three PR5 sites, with 6,000 pages each and one PR6 page w/ 1,000 pages, it is not only better for YOU to add the pages from one of the PR5 sites to the PR6 but it is better for the network as it grows also.

[thebassman]

You can only have 1 account per domain, regardless of subdomain, though, right? So even if you had your archive in a sub domain, the way it's currently set up, you can't list them as seperate sites unless the archive is on a different domain all together, right?

[schlottke]

I believe so bassman- I think *.domain.com falls under www.domain.com at the moment.

[Kadence]

The point of the co-op should be fairness as well. If I can break up an 8000 page site into two, and get twice the amount of weight, then thats not what I consider proper. If it takes 6 times the effort to get a PR6 than a PR5, and a person only gets rewarded with a factor of one, then its better to have many lower PR'ed sites say PR5, than aim for the higher PR - that's also not proper.

Bottom line, if its based on the Google API factors, then in my opinion its needs to be as close to the Google formula to stop any manipulation.
Exactly--the point should be to replicate the Google algorithm as closely as possible.

Problem is, we don't know the specifics of the algorithm; especially in terms of how it weights links at different domains versus, say, sitewide links. The point of the network displaying randomized ads is to address the belief that there are steeply declining marginal returns for sitewide links.

Reducing the weight scale for PR also addresses this indirectly, and might make the contribution<-->reward relationship more accurate, rather than less.

As for the arbitrageurs, if they're trying to split existing sites, they're going to lose existing PR--unless they make lots of 301 redirects. Arbitraging the current Co-op algorithm is not a free of transaction costs of its own.

[t2dman]

Problem is, we don't know the specifics of the algorithm; especially in terms of how it weights links at different domains versus, say, sitewide links. The point of the network displaying randomized ads is to address the belief that there are steeply declining marginal returns for sitewide links.
The network is brilliance in terms of it not being sitewide links. It is certainly delivering an amazing service. Not to say it can't be tweaked or that it doesn't still deserve analysis. Sitewide links can be a killer to sites in that the "one site" issue starts to kick in.

As for the arbitrageurs, if they're trying to split existing sites, they're going to lose existing PR--unless they make lots of 301 redirects. Arbitraging the current Co-op algorithm is not a free of transaction costs of its own.The benefits are potentially huge.

My recommendation is that for people with an existing site that want to add a forum, that they use a subdomain for it rather than adding as a subdirectory for their site - since the non VBulletin pages get an extra weighting of x2, and they will very quickly get to the max 4000 based on their VB site only, and easy to get PR5+ on both sites rather than try for Pr6 on having as one site. In fact, having two lots of weighting rather than one, will probably put both to Pr6+ anyway (?).

And for people with large sites with multi topics - consider breaking it up into a subdomain. The co-op power may well be sufficient to break the new site out of the sandbox and have two+ powerhouse sites. Just be careful about the "one site" issue.

[thebassman]

you can only have 1 account per domain, regardless of it you split your site up with a subdomain or subdirectory, I believe...