It’s not often I find myself kerfuffled, but that’s what happened when I was permitted to be a guest to discuss Windows Phone 7 on Windows Weekly, hosted by Leo Laporte and Paul Thurrott.  I’ve been a long time listener of several of the TWiT network podcasts, and fondly remember staying up late in grad school to catch Dev Null on “The Site” on MSNBC.  It was a lively conversation, and I have received many emails from people I don’t know with their thoughts on the show and WP7.  I love hearing from the community, and really look forward to meeting more developers building cool apps and games.

We also sponsored a couple of .Net Rocks shows with Carl and Richard (show 1) so that we could talk about the tools release, but more importantly so that we could enable 6 developers to talk about what they are building for Windows Phone 7.  We really wanted to give a microphone to devs doing cool stuff; specifically guys who are working without the aid of PR firms and other means to get the opportunity to promote themselves.  We know that there are going to be many apps built for Windows Phone 7 by small teams, and we really want to help get them noticed.

I have been thinking more and more about the behavior modifications which are likely to come as a result of the new Google Instant.  Whereas Google positions their technology as an assistant to the search process, I wonder if it ends up acting as a drain on creativity.

The feedback loop which they have created is certain to have impact on search terms people type in.  I read somewhere that 20% of the searches at Google have never been seen before.  I need to track down source on that, but if that *was* true, I wonder if that is *still* true now that the search results page updates in real time.  Further compounding the potential loss of serendipitous searching is the reduced suggestion list.  A searcher is only presented with 4 real choices.  Will people continue to have flights of search fancy?

Despite the fact that the results are based, presumably, on the most likely search terms, will people tire of the lack of diversity?  I wonder what type of world my children will have when they are programmed from youth to believe that what Google suggests is “the most likely” or “most popular” search query.  Will they limit their choices because of the massive constraints placed on their minds by the manifestation of the information as presented by Google?

With this insistence on showing you only what’s popular, is Google running to a model not too dissimilar from Top40 radio programming?  I find that strange given how consumers have opted for the long tail model access to music via services like iTunes and Zune Pass.  I have discovered a ton of great artists in the last couple of years, and haven’t listened to Top40 radio in ages.  I don’t even know what’s on Top40 anymore, but I do have a ZuneHD full of awesome music.

Helping me find what I want is great.  Insisting that what I want is a narrow list is not.

I have always heard that Google is ruthless in their testing and use of data to make decisions, so I have every reason to believe that they know what they are doing.  However, their new Google Instant is something which is a new animal all together.

Steve Rubel posted today that he thinks that SEO as we know it is dead.  His reasoning is based on the notion that because of the prompted feedback, all of the web will look different for everyone.  The process of prompting these searches will in fact cause a massive balkanization in search queries, which blows up traditional SEO strategies.

I’m not sure I agree with this.  Because Google is curating the suggested search terms, presumably based on likely terms gleaned from use data, there is a discoverable path for search terms.  Clever SEO marketers can figure out these terms and optimize for them.

Further compounding this problem is that searchers are no longer committed to many search queries.  Before Google Instant, a searcher was promoted with likely terms, but they still had to click the “Search” button to get the results.  At that point, they were committed to the results page, and would likely consider all links, and, more importantly, ads on that page.  If they didn’t find what they were looking for, they would click the back arrow, and conduct a new search.  With Instant, the results change in real time.  So do the ads.

Let me say that again.  The ads change in real time.  What. The. Hell.  The way that Google worked before was that they optimized inventory for revenue attainment.  It didn’t matter if you were willing to pay a $100 CPC.  If your ad didn’t perform, it was removed in favor of those which garnered more clicks per impression.  So the first question, what now counts as an impression?  If the ads are changing in real time, isn’t Google now creating two types of ads?

A PRIME ad would be that which is at the end of a search query chain.  An ALT ad would be one that may only appear for a millisecond as the search query is being typed?  Is it possible a searcher considers those ads?  Do they get completely ignored, and are thus worthless?

UPDATE: From Google, there are several things which will cause an impression event to fire.  Selecting anything on the page, which makes sense.  But stopping typing for 3 seconds also causes an “impression” event to fire.  Interesting.

Allow me to demonstrate with an example.  Here’s a query I started.

As you can see, there are no ads displayed.  Simply adding a space (i.e. “Specialized “ vs. “Specialized”) generates ad impressions.

Strangely, the ad inventory doesn’t change as I type the word “bikes.’”

So they are certainly optimizing the ad inventory based on the likely pathing through the search query tree.  You can see that if you type “allez” instead of “bikes” that the ads do change.

Again, back to “bikes” as the root search.  If you add a space to it (“Specialized bikes “), the ads shown are those associated with the next term “Seattle.”  So Google is trying to at least show you ads based on where they think you are going.

So what Rubel suggests is partially true.  Search becomes a unique experience – making it hard to exploit by the SEO marketers.  The inclusion of “Seattle” as the likely next term for me is probably based on the fact that I live in Seattle, have an IP address in Seattle, and likely have done many searches with Seattle in them.

Will SEO marketers be able to figure out the suggestion tree in a programmatic or general way?  It seem that with this new model of Google Instant, the generalized search terms are no longer important, but rather it’s the leaf nodes on the Google suggested search term tree (long tail) which become infinitely more important.  Suggested search terms with the immediate feedback of a changed SERP will likely cause general keywords to plummet in value.

What does this mean for Google’s revenues?  Did they in fact just dramatically reduce the number of searches, thus impacting their revenue opportunities?  Will clever marketers be able to discern the likely pathing through Google’s suggested query terms, and bid up the PRIME ad inventory, offsetting any lost revenue due to reduced searches?  What is the value of ALT inventory versus PRIME inventory?  Will searchers note the change in ads and consider them in turn, meaning that any bidding up is a short term phenomena, causing a temporal spike in revenues?

Again, I assume Google has tested this and thought through the issues.  The challenge, however, is that with any large scale system which influences, and is influenced by, human behavior, the results often differ from the expectations.