It does not affect rankings and pagerank in any way..... Just try to validate http://www.google.com and you will be surprised to see the results.... It has abt 60 errors... So I don't think it makes any difference but it basically helps or ease spiders to browse your site.....
A lot of people here are comparing apples to hand grenades! Ebay has amazing PR and rankings because the have a bazillion backlinks! Same with Google .. come on folks!
Valid code may NOT effect rankings but clean code sure does and I'll argue that point until enternity. I have proven this a thousand times over since I went to tableless CSS driven sites. Yes my sites are almost always valid but that's only 1/2 of it. Getting rid of all the GARBAGE code the spiders have to wade through is the real trick.
I do websites that conform to w3c too, unfortunately.. I am yet unsure how these two do well in the engines, wether a nicely one ranked well compared to that of the table type made site.
Although I understand that css based websites are more search engine friendly, and is read more efficiently. That is one reason I can support apart from that. I am quite unsure, I would really like to hear what you have in mind dc dalton. As I know deep inside of me, something worth must've been telling me well coded can rank higher than that of a non css version.
More than happy to explain. I started my testing over 2 years ago with one of my old site, dcddesigns.com (which I no longer own btw) ... we were in one of those categories from hell where we couldnt find any listings in anything above page 100 of the search engines. Being that at that time this site was my only source of income I started looking into alternative methods to enhance my site. I did a bunch of reading about how spiders actually worked and realized the more crud they had to wade thru the harder it was for them to get to the content. I also found some interesting reading on code / content ratios on a page.
Being a "table guy" myself at that point I made the decision to learn tabless CSS layout and flip the site into a new layout. After a month of beating my head on the desk learning it I finally got it done and redid the site. This was Dec of 2004. I set the site up with the exact same content but in the new layout. The only other difference was I added a mod_rewrite for the urls.
I didnt bother to resubmit so I sat around and waited. In Feb 2005 I got a contact form from a potential client about a large app and after talking to him I asked how he found us. He said he typed something in to Google (forget what it was) and we were #1. (got the job BTW, a $10k job)
Needless to say I RAN to the browser and started testing our keywords and found that within a month Google had brought us up into the top 10 on 6 out of our main phrases and the other 4 were within the top 30.
I have since become an advocate for this design methodology in the SEO area. I have replicated these exact same type rankings in over 30 sites now ... one of my more famous was a client whose primary keyword went from "not found" to #9 out of 91 million results.
Plain and simple it works! Yes I use other things like mod_rewrites and good clean html attribute usuage (headings, alt and title attributes) but the bulk of the success I attribute to the super clean "spider friendly" layouts.
I most recently proved this again with my new site (in my sig) ... the site was brand new in March of this year and I have pulled 15 top 10 listings in Google for my primary keywords. The one I love the most is "css driven website" where I am ranked #4, only behind sitepoint and the w3c and "tableless css driven website" where I am #1 in front of Sitepoint and the w3c.
Folks you can argue all day long but I am telling you, You want to stop having to kill yourself at SEO? Get into tabless css driven layouts, a mod_rewrite and some great content and 3/4 of your work is already done!