BTW: Great post - I’m self-learning SEO & this helped me a lot!!

@steve: Best of luck with your web design and SEO business. I’d expect you could take that background of yours and combine it with your SEO learnin’ and generate some decent returns as a web entrepreneur.  

• We need to separate domains from URLs, so we can better discuss the impact of keywords in the URL (such as www.johnon.com/seattle/seo-in-seattle.html) from domain (www.johnon.com vs. www.seattleseo.com)
• “user experience” is a nice, friendly descriptor, but it is ill-defined and relies on Google’s interpretation. In the past Google has said that pages which rank well and deliver very strong return for advertisers are actually spam (e.g. when they have little content of their own). Rather than accept Google’s pronouncement of the user experience as valuable, we need to get Google to describe what exactly is meant by the user experience.
• If I (john andrews) say that keyword SEO works differently in different markets, and under different levels of competition, then suggesting that keywords in the domain name is helpful because it supports a positive user experience must also be market dependent, no? My Answer is yes, that is correct. BUT, we don’t know if Google’s handling of keyword domains is that sophisticated. For example, if a word has multiple meanings depending on context, does Google consider context or just presence of word? My vote is Google falls back to the predetermined contxtual relevance used for the content, when considering the domain name. But I say that simply because it is easier to do that than analyze the domain name for its market niche… something Google can add later when it is worthwhile to do that extra work.Does this answer the question for long-term domain speculators? Nope… which means consider that in your risk reward planning.

@peter: yes, that is what I am saying, and I say that not justbased on what matt said but personal experience. Of course only Google knows what is truly happening… but matt did say the domain name value was partly based in the contributionof the domain name to the user experience, not just the fact that there was exact match. 

My perception is that linking is changing from where it was when Google started. Long link pages with the domain names used as the anchor text are becoming less frequent and contextual links within the article, (i.e. SERP above) are more prevalent. Now if your link building strategy revolves around submitting to 1000 directories which will use the domain name as the link, you may see more of a bump from such links to keyworded domains. However if your links are more natural from sources that are discussing what is on the domain rather than the domain itself, the links will naturally contain better anchor text and be less dependent on the domain name.

Searching for an [online auction] brings interesting results as well. With ebay bouncing around the top 10 somewhere, even though they probably have the most links of any site in the top 10, onlineauction.com is always as #1. If that is from keywords in the domain, the anchor text, or on-page factors, I’ll never know for sure, but makes for a good study.

Assuming no brand awareness at all (tough to do, but imagine we’ve got a Martian using Google today for the first time) onlineauction.com may be the one he’d choose before ebay.com even though onlineauction.com doesn’t even bother to have a snippet, the domain itself screams that it’s relevant to the query, though in reality ebay is a much bigger brand and site for the query.

For me personally I don’t have any “run 10 super-bowl ads” kind of budgets to deal with so branding of domains isn’t going to be as easy and will almost always choose the keyworded domain if I can get it, to the limit that online-auctions-that-sell-books-cheap135.info looks like junk and will never go anywhere.