DMOZ & www, non-www site-splitting

[Lightfoot]

I apreciate there have been a number of posts about site-splitting in the past, but it has still not been resolved.

2 or 3 times over the past couple of years my site has been split into www and non-www, and at the same time the DMOZ description and back link have disappeared. This can take several months to sort out again. I generally just wait patiently, but the re-occurence means I really need to
sort it out.

The DMOZ link is to the www. version, and all the backlinks I have found are also to this. I don't know why the non-www keeps being listed.

Does anyone have any advice as to how to get the site 'resolved' back into one again - and why it is connected to loss of the DMOZ backlink?
<>

Many Thanks

Lightfoot

[edited by: ciml at 11:00 am (utc) on Oct. 19, 2003]
[edit reason] Please see StickyMail. [/edit]

[dirkz]

What do you mean by site-splitting and www and non-www?

[Lightfoot]

A single site, at a single IP address, becomes split into www and non-www versions by Google, as if they are separate content. Whilst the content is exactly the same.

Happens to me all the time - I have no idea why - perhaps its just bad luck.

But is divides my PR.

I'm making some progress though after some helpful sticky-mail sugestions.

[ogletree]

The reason is that other sites are poining to you using differnt URL's. Get them to change the link and it will fix itself.

[flicker]

Google has been doing some strange things lately with different forms of the same URL. I've noticed this with a site some friends and I have on Geocities: Google can't seem to make up its mind whether to use the old "neighborhood" URL they used to use (www.geocities.com/TimesSquare/Dungeon/1234/xyz.html) or the new form that's been used in recent years (www.geocities.com/xyz/xyz.html). No one's linked to the "neighborhood" address of our site since Yahoo bought Geocities years ago, so I have no idea why Google's remembering that all of a sudden.

It's a pointless hobby site, so in my case I don't care (both URLs pull up the same page, so it's not as if anyone is getting lost). I can see why people with business sites are upset about that though.

[martinibuster]

You have to wonder if you're getting your fair share of PR when you see a PR 5 with one version of your url and a PR 6 with the other.

Don't bother asking the dmoz editor to fix your listing as they are uptight about fixing things that from their perspective is not broken.

[flicker]

This doesn't sound like a DMOZ thing, martinibuster. Lightfoot's site is listed with the www in the ODP, and the Geocities site I'm talking about isn't in the ODP at all. It's something to do with Google. I wonder if they're experimenting with really old Google archives to identify how longstanding a site is for ranking purposes, or something, and this is causing old forms of a URL to pop back up?

I'd guess that the fact that Google dropped and picked back up on DMOZ backlinks is probably due to all the downtime DMOZ had last month. Don't know why Google's fluctuating so much about it, but that's why I'd bet the backlinks dropped in the first place.

[Lightfoot]

Dear Martinibuster,

I concur with your DMOZ caution advice - I would *never* request a DMOZ change - once you're in, for heaven's sake stay in!

I've looked at all incoming links - some very strong indeed - but all are to www.

An expert and helpful sticky-mailer has suggested that if Google 'thinks' I'm 2 different sites, it will respect that - makes sense. But I'm not, so consequently I've ensured all my incomings are to www. But if there's a weak non-www link out there somewhere, and the site content changes between the frequent (www) crawl, and a rare (non-www) crawl - this could explain my unwanted double listing. I guess GoogleBot will keep seeing different content each time.

Well, I've bitten the bullet - my highest incoming link (one of my own sites) has been adjusted to point to both www. and non-www. - in the hope that Google realises it is crawling the same site.

p.s. I think it all comes down to a very slow DNS cache update with Google. Heck! My browser resolves both sites -why can't Googlebot!

p.p.s.

Flicker - what an insightful and excellent poster you are!

[dirkz]

A single site, at a single IP address, becomes split into www and non-www versions by Google

So what you mean is example.com and www.example.com?

Why don't you just redirect example.com with a 301 to www.example.com?

[plasma]

As long as you don't have a 301 redirect google will see these as 2 distinct sites (pages).
Even if nobody links to you: G will take the url from the toolbar, even typos will be followed (That's why you have to agree to the privacy statement if you want to see Toolbar-PR).

If the site is reachable (e.g. wildcard domains) it will be crawled.

Short Answer: use 301 permanently moved / don't use symlinks

[Lightfoot]

Hi,

If the site is obtained with the URL appended with www. and also without the www., then clearly the DNS look-up is working - and this is indeed the case, the non-www is being redirected to the www!

You see, as soon as re-directs are mentioned on this site, a dozen (or more) e-mails follow about how to do it! (most of which give conflicting advice!)

Why fiddle with a server that is doing it's job - it is clearly re-directing already. If I'm being an idiot, please advise (and throw custard pies as required ;-) - but it seems fairly straightforward to me.

If GoogleBot follows the URL it will find exactly the same IP address whether www. or non-www.

(assuming Google's DNS is up to date!)

[edited by: Lightfoot at 7:40 pm (utc) on Oct. 19, 2003]

[dirkz]

If the site can be obtained with the URL appended with www. and also without the www

This need not be a proper 301 redirect. Could just be the same virtual domain. For Google, these are two separate ones. Use the server header checker on searchengineworld to check whether a 301 is taking place.

As a rule of thumb, if you type in the URL "example.com" and it remains in the URL bar, it's definitely NOT a 301 (don't conclude the opposite from that though).

Or sticky me the URL and I'll check.

[Lightfoot]

Dirkz,

You've mentioned a virtual domain - and now it starts to make sense. You may be on to something.

I will sticky an e-mail to you, and please disregard previous pomposity - blame it on my Englishness

Cheers

[dirkz]

Lightfoot, thanks to my Englishness I didn't even realize any pomposity :) I will look at your URL.

[Lightfoot]

Dear Dirkz,

You see, due to my Englishness, I always assume I've been pompous due to my extreme politeness, which you have been polite enough to politely disregard, ..er..., we'd better leave it there. That Blaine chap is coming out of his greenhouse tonight apparently. Generally, when I walk out of mine I'm on terra firma, and not 60 feet off the ground (although I am generally looking for something to eat.)

[edited by: Lightfoot at 8:05 pm (utc) on Oct. 19, 2003]

[dirkz]

I have discovered a potential problem: example.com is in fact redirected to www.example.com, but via 302, which is a temporary redirect. In order to transfer all PR to a single one domain, you need a 301 (permanent).

I've checked this with the header checker on searchengineworld, you can easily access this tool with a link from the main page.

If you need further help feel with this free to ask.

The explanation for the DMOZ backlink is easy: Google doesn't show all backlinks all the time. Sometimes only 50% or less, or all from PR4 or higher are shown. Very unstable.

Hope this helps.

[Lightfoot]

Thanks,

I've checked out the tool and will forward the results to my host. There's no reason for a 302 direct, it is a permanent one. We were previously on a different server, but this too was a virtual one and may well have had the same settings - hence the problems we've had for a long time.

Many thanks for the heads up - I'll look into it.

Cheers