The website I work for is suffering some serious consequences because we've dropped into Google's "supplemental result" index (commonly referred to as "Google Hell"). I don't know if it really is Google hell but I wasted a couple work-days theorizing and researching on how to pull our pages out of purgatory. I gleaned these ideas from some other websites which I unfortunately have long since forgotten so cannot provide attribution (sorry). This is all conjecture, however, because I don't have any proof that these methods actually work.

1. Varied and semi-random META descriptions
2. increase quantity of META keywords
3. put important keyword combinations closer together
4. try to fake in some random natural language text

Varied and semi-random META descriptions

One thing you may notice is that Google really does pickup META descriptions. Lots of people say that META doesn't matter anymore but I think it does because sometimes META descriptions appear in the match summaries for result links.

If all of your pages have identical or substantially similar META tags, you appear lazy and pages become obviously generic. Why would anyone search for generic bullshit pages written by a machine? Thus, I think having overly generic and identical content in your pages (semantic and META markup) hurts you.

increase quantity and quality of META keywords

I figure if you can put in a crapload of good keywords why the hell shouldn't it help? As long as it isn't the same keywords and the keywords are actually in your pages it can't hurt to have a bunch of them in your META.

put important keyword combinations closer together

This is about having the right words near each other. If you can possibly craft your language around using the same phrase a few times within the page it has to match better than just having the words spread out at random across a page. Duh. I'm guessing its better to have related variations of the phrase within the page than having only a single phrase repeated over and over. That would make it more natural sounding AND give you better keyword coverage on related terms. Hellz yeah.

try to fake in some random natural language text

As mentioned in the points above we try to increase our keyword coverage by using meaningful variations of target phrase. At the very least it makes the page interesting to read and hopefully search engines are understanding context of a page through some algorithmic magic. If this is really the case -- and I think it is -- the less your pages read like a robot wrote it, the more likely a robot will like it. I would characterize this as self-discriminatory by robots against other robots. whoa -- that has a wierd deepness to it that makes me surprised I'm not high or something.

An Adpative Path blog pointed me over to this site called Making Life Easy which features several everyday real-world user experience quirks. It is part of the World Usability Day initiative I saw earlier this week.

Portland Airport's Water-saving Toilet is one of those no-brainer ones where clearly, if you just thought about it for 5 minutes you would have thought of a better way to get the job done. Fortunately not all urinal designs are so poorly conceived.

Link via Adaptive path

The UPA is going to have a conference titled "Patterns: Blueprints for Usability" June 11-15, Austin, Texas. This would be something I would like to see.

From the website:

As usability professionals, our ability to observe users and to discover their patterns of interaction is integral to our work. By defining these patterns we can then leverage that knowledge to create usable interfaces that are familiar and useful to our users.

Link

While browsing the Raible Designs blog, I came across this interesting article regarding input label placement, at UXMatters.com. Both of these sites are new to me but they have some useful things to say about HCI and interaction design.

The results of the article:

• Form label above input field has least saccade time (I.E. virtually no eye movement, thus faster)
• Right-aligned label next to input field is next fastest
• Left-aligned label with input field took longest (500ms saccade time)

The Information Architecture Institute has a salary survey from 2006 and 2005. It appears the sweet spot is around the $85-95k/year range.

However, that may not be so high afterall considering 54% of respondents live/work in the Northeast or Western United States. A whopping 95% of respondents are located near large Metropolitan areas. I'm guessing those regions are New York, Boston, San Francisco, and Los Angeles.

Link