Back in the early days of Red Hat, Bob Young, one of the founders and the first CEO, used to say, “What we really sell to our customers is control.” His argument was that one of the benefits of open source software is freedom from vendor lock-in. If you don’t like the product, you can get under the hood and change it yourself, or switch to another vendor offering a comparable product based on the same underlying technology.
I had this thought come back around in a Web 2.0 context this past week, with an eerie twist. It was two proprietary Web 2.0 software companies — enterprise search company FAST and enterprise mapping company Placebase — that told me, in two independent meetings, that one of the main benefits they sell to customers is control.
How can that be? The common wisdom is that Web 2.0 is all about openness. Of course, anyone who has thought long and hard about why “data is the Intel Inside” is one of my key Web 2.0 maxims understands that successful Web 2.0 companies each control a vast user-contributed (or at least user-enhanced) database that gets progressively more valuable via network effects. And control over that database belongs to the vendor, not to the user. Another way to put it is that every long tail is ultimately attached to a large animal. (I can’t remember which wag :-) originally came up with that line.)
FAST, at whose Fastforward user conference I’m speaking later this week, points out that Google claims more than 80 different factors that control search relevancy. But the knobs and levers controlling those factors are largely opaque to users, even corporate users who are private labeling Google’s search technology. By contrast, FAST aims to provide a platform under which the customer controls all the knobs and levers, and can add their own relevancy factors as well. So, for example, at Dell or Best Buy, price and availability are additional factors that need to be seamlessly integrated into search. At CareerBuilder, personalization and profiling (e.g. “job seeker” vs. “job searcher”) need to be just another factor in the search. (Frankly, that’s the same reason we use MarkLogic’s xquery search at O’Reilly — we have control over context and semantics that we can’t get from Google, even though they are better at search than we are in so many other ways.)
I heard the same story from Placebase. Why, I asked, do people pay you when it’s so easy to put your own data on a map mashup with Google Maps? There were two answers — first, that there are many additional types of map-data integration than pushpins. Placebase offers a full suite of GIS-style map visualizations, so that the customer can figure out what type of mapping layers most suit the data they want to show. The second answer was a lot like FAST’s: we give the customer more control over what they want to do. It’s not one-size fits all.
Now, I’m not claiming that Fast or Placebase is somehow more open than Google. I’m simply pointing out that openness occurs at different layers of the stack — and so does closedness — and that a proprietary software application might give more control over your own data than a free-as-in-beer Web 2.0 equivalent. What you’re actually paying for is that control. My original argument in The Open Source Paradigm Shift (the precursor paper to What is Web 2.0?) was that much as open standards in hardware had moved the locus of proprietary advantage from hardware to software, open source software would move the locus from software to data, via what we now know of as Clayton Christenson’s “law of conservation of attractive profits.” Companies like Google and Yahoo! have done huge amounts to contribute back to and support open source software, but they jealously guard their own proprietary advantage, which is in data and algorithms and business processes for gathering and managing that data.
And I’m pointing out that a new class of entrepreneurs are finding ways to “open up” the closed applications of Web 2.0. If you’re thinking in the old boxes of open source vs. proprietary, you’ll miss companies that are surfing new edges between the two.
And what’s more, thinking further along the lines of how to give control back to the users, we can ask ourselves what will be the real “open source” answer to proprietary Web 2.0 databases. We’ve started to hear about “open data“, and companies like Wesabe even have open data policies. (Note: I am on the board of Wesabe.) But open data is only part of the new openness we need to explore. What companies like Fast and Placebase do is give users control over the algorithms that are used to manage the data. As Nate Treloar from Fast said during our call, “Relevancy isn’t just an algorithm. It’s a platform.”
And at least so far, companies like Fast and Placebase risk giving their users control over their algorithms by taking a step backwards from another Web 2.0 principle, that of building systems that get better the more people use them. Companies that build their own local data islands may get more control, but they may lose in other ways. In a world driven by network effects, the system that is most open to user contribution will get better faster than one that is more closed, even if the latter system gives more control over algorithms. In my conversations, I urged both Fast and Placebase to think about encouraging their customers to share data with each other, so that each application on their platform benefits from all of the others. Exactly how to do this, given that companies that choose private label search often have proprietary data that they don’t want to share with others, is a difficult exercise, but an important one, not be overlooked just because it’s hard. Choosing defaults for what data will be made public is probably the single most important architectural decision for any Web 2.0 company.