For the last year or a number of SEOs have speculated that google is using a new approach to ranking as part of its algorithm.
This has been confirmed for me now by people I've seen exploiting this mechanism to rank highly (e.g. #2) with virtually zero backlinks and a completely new domain.
The mechanism is google's own internal click through data which they have been building up especially with people who are logged in to their services (which is becoming the norm).
The mechanism works like this:
1. You search for 'reticular activating system'
2. You click on one of the listings
3. you don't like what you see - and you click "BACK"
4. You either use the same search on google or you re-search for a similar term
5. You click another listing
6. You like what you see and don't go back to google to re-search or look at another listing
Google simply tracks all of the searches that you weren't happy with - you went back to them to find something better. These sites get a negative impact on their SEO. Sites that have lower 'click back' ratios get more highly ranked.
The implication is that "backtrack bounces" are now more critical to eliminate if you want google seo traffic on that landing page.
Now to the exploit! 
1. Buy a domain name with exact match keywords for the term you want to rank for (.net, org are best) - but you can get it to work with others
2. Create a super simple page on the domain with this HTML (your index.html):
<html>
<head><title>[KEYWORD]</title></head>
<body>
<a href="[link to any landing page you want to promote]">[KEYWORD]</a>
</body>
</html>
Replace [keyword] with the keyword you want to rank for.
Here's an example: http://www.reticularactivatingsystem.net
You should make sure that there is some roughly relevant content on your landing page of course, so that they don't back up all the way to google. But the vast amount of people click the link on your page - so you trick google's clickthrough mechanism into thinking it's a valid page.
I've seen this rank for keywords that it just shouldn't be possible for - at some point google will fix it, but for now it's got them trumped.
I don't see your example ranking even in the top 5 pages of Google?
PUAhate.com = WTF ????

Interesting trick. Your site is #8 on Page 2 of results for me...
huh eh wtf?
Here's a high ranked URL: barrykirkeyradioshow.com
--> #1 for keyword Barry Kirkey
@julien - yeah, WTF++
hahaha.... barry... he's an old friend of mine.
Do you throw any backlinks at these sites?
Can you also do this dynamically?
In other words, get the search query variable and echo it on your page?
Something like this: http://www.blastedthing.com/php-tric...-on-your-page/
Then echo the search term into your title and anchor text.
Seems like that could make you rank for several long tail keywords as well.
Or would that dilute what we are attempting to do?
One more suggestion... how about using javascript to click the link for them? If that sets off a red flag you could always do random delay before the click. That would essentially give you a redirect right to your target page.
Sadly, I'm pretty sure that wouldn't work. It's not about what title tag, anchor etc the user sees - it's about what title tag, anchor and so forth Google sees.
You could, however, capture search terms from an Adsense placement or similar, then generate static pages with each of those terms. That might work.
I see absolutely no reason why this would possibly work, but I'm willing to give it a shot
I'm sorry about the noob question but, does the initial traffic HAVE to come from google search for the relevant keyword? If so, how would you get search traffic if your site is brand new?
Would it be possible to do it with paid traffic from 3rd party websites?
. What do you think about clicking the link w/ JavaScript?Registered a domain 9 days ago and wrote in code for the site I actually want the user to end up on
Do I just sit back and wait now?
worth a try, dunno what to promote tho
mass affiliate links to amazon?
@jfarley - depends how hard you want to work. Personally, if I was doing something like this (which I generally don't do, because I prefer white-hat SEO) I'd set up a bunch of Web 2.0 properties pointing to an index of all the pages I'd made (or some sub-indexes with appropriate a tags for each of the individual pages), then start running all the usual dodgy link sources to those web 2.0 properties.
Just waiting will be slower.
@thomashbm - That's an interesting question. My suspicion is, given that we know the Googlebot parses Javascript, that Google are aware of the possibility of scripts to auto-click links, and discount them. It's what I'd do if I was building the Googlebot, and I'm not as smart as Matt Cutts.
Guys, just to be clear - the URLs I put up here are not ones I created to use this - they are just examples I pulled off the net of others using this hack.
Obviously I have better things to do like make $$$ instead of spending my time retarding on a hate site 
just skimmed through the OP, i think you mentioned using an EMD or PMD for this test, how do you know it's not the value G sees in using a keyword rich domain as a ranking factor instead of this bouncerate theory?
when a searcher is researching, they might go through a few pages that are ranking for their queries very quickly
shooting for the 'long click' is definitely a good idea though
i don't know, i just haven't seen many reports from authority sources on whether this is a big factor
Nice tip! To apply this technique on any website check out microworkers.com, I heard it's good for search+click type gigs.