In my last post I maundered about potential properties of the eTourism’s Long Tail. Today, as promised, I will extend the discussion (actually it’s a monologue still) to another aspect of Web 2.0 Design Patterns, more precisely to what Tim O’reilly refers to as Data is the Next Intel Inside, meaning that most web 2.0 applications are essentially data-base driven. O’reilly emphasizes that the challenge here is to create a unique, hard-to-recreate source of data.

Mashups and Meta-Searches

Price comparison and booking sites have a hard time to accomplish this since their USP generally is their ease of use (e.g. a fast and fancy AJAX-interface) and the sheer amount of data that can be compared. However, that data is being pulled from external sources most of the time and a state-of-the-art interface doesn’t really pose a insuperable entry barrier these days. This applies to mashed-up interfaces as well. If there is no additional data added in the process of mashing and therefore no other sustainable value created than the way the emerging results are presented you should better do some cool searching stuff.

Recommender Systems

On the other hand, web usage data can be tremendously valuable for marketing research. But this layer of information doesn’t easily translate into immediate value for the users – at least as long as there is no recommender system implemented that can harness these usage patterns and relay it to the users (or even social interaction patterns, if the site implements some kind of friendship model). However, recommender systems become truly powerful first when combined with data that reflects users’ unbiased preferences acting as fixed points to tie the fuzzy data mining algorithms to. Another thing about those systems is that they help driving the demand down the tail by enabling people to find their niches more easily. Moreover, like Chris Anderson points out, in combination with an increasing variety on the supply side the demand curve flattens out.

Bring Sexy Back!

The classic approach for travel and shopping sites looking to enhance their data is to drive people into reviewing, ranking or rating – or even creating new assets, which on the other hand bears the danger of introducing inconsistencies into the data-base.

If I think of unique, hard-to-recreate source of data the word that crosses my mind first is the word “unsexy” (a germanized Anglicism meaning “not sexy”). In a perfect 2.0 world the uniqueness of a data-source should not be based on assets in certain data-bases but on the way the data is humanized. As I mentioned, recommender systems are most powerful when harnessing direct user contributions (voluntarily revealed preferences) and web usage profiles – but what makes them smart is the way they combine these two realms and the way they draw conclusions.

Now, open those/source your data silos! Give me some APML! And then we’ll see who’s really smart.

Read more in my next post in which I will address item 3 on the List of O’reilly’s Web 2.0 Design Patterns: “Users Add Value”. That post will also feature some nice and shallow background knowledge on buzzwords like social networks, the wisdom of crowds and collective cognition. So if you’re looking for something to impress the ladies (or gents), stay tuned.