L’adoption de l’innovation
Par Bernard Golden, Navica Software (en anglais)

How soon will mainstream IT organizations adopt open source en masse? According to Harvard Business School Professor Karim Lakhani, maybe never.
I met Karim a little over a year ago at a company advisory board meeting I attended. He did his Ph.D. thesis on where open source innovation comes from; when his research is published later this year its conclusions will shake up the conventional wisdom on this subject and significantly affect current assumptions about the common practices of decentralized innovation. Karim now is a professor at Harvard Business School focusing on innovation.

I visited Cambridge last week to attend a meeting convened by the Kellogg Foundation to address the role of technology and education and present on the open source work I've been collaborating with the Silicon Valley Education Foundation on. I arranged to meet Karim to catch up on what he's been doing.

I was particularly curious to get his take on what I view as the critical issue in open source: the rate of "official" take up by mainstream IT organizations. I say "official," to reflect open source use that is not comprised of covert downloads or opaque inclusion in projects, but instead is sanctioned by upper management and overtly recognized and tracked as part of mainstream project management. In a phrase, open source use that has come in out of the cold.

I began our conversation by describing a vivid anecdote described in a recent eWeek article, in which Jonathan Schwartz, CEO of Sun, upon hearing from a financial services CIO that her organization uses no open source, responded by informing her that 1300 copies of MySQL were downloaded by her employees during the previous six months. (In fact, I blogged about this incident twice because I found it so striking).             

"Karim Lakhani's research will shake up the conventional wisdom on the topic of open source innovation."

Turning to open source adoption, I noted that it poses a real challenge to mainstream IT, since the economics of open source mean that open source vendors don't follow established software selling activities: massive marketing, multi-visit direct sales forces, extensive pre-sales technical support, and so on. Consequently, IT organizations have been uncomfortable adopting open source because it "feels different" than what they're used to.

On the other hand, the price of open source is undeniably better than proprietary software, which, to my mind, makes the value proposition irresistible.

What keeps mainstream IT organizations, I asked Karim, from embracing what (to me, at any rate) is a self-evident powerful innovation? Rather than responding directly, he gave me a copy of a case study used by HBS that describes the adoption (or rather, non-adoption) of innovative technology at Google. Several Google employees created an internal prediction market to allow a large number of Google employees to "bet" on various important dates at Google relating to project release dates, application uptake, etc. Even though prediction markets have proven to be quite accurate in forecasting important events (even more accurate than established tracking mechanisms like expert opinion, official statistics, and so on), Google managers refused to use the prediction market tool, preferring to rely on project dashboards and the like (i.e., officially sanctioned reflections of project status that did not rely on "uninformed" opinion to predict results).

In other words, even Google, that Pantheon of genius and enlightened management, preferred to stick with accepted, though suboptimal, management tools rather than embrace ones that might be more accurate but also denigrate managerial omniscience. More strongly stated: even mighty Google demonstrates that organizations reject innovation, preferring comfortable mediocrity to uncomfortable advantage.

I noted that this phenomena is what I refer to as "impedance mismatch": open source low touch sales practices running up against senior IT managers who prefer high-touch selling. While this seems like a dilemma, I told Karim that I'm a firm believer in the ultimate power of cheap: the acid of low pricing will dissolve the edifice of established processes and will ultimately leave a decayed wreckage of obsolete practices, brought down by open source innovation.

Karim demurred. Large IT organizations prefer to interact with like organizations, or as he phrased it, economic actors that resemble themselves. Large IT organizations will, essentially, refuse to adopt open source until acceptable economic actors are available.

"Large IT organizations prefer to interact with 'economic actors' like themselves."

This makes sense - large organizations strongly prefer to deal with software vendors that resemble themselves; it's a more comfortable way to consume innovation. On the other hand, I don't see how the economics of open source support standalone economic actors of the type Karim describes. When I pointed this out to him, he responded that other types of economic actors will step in: large companies like HP or even non-profits like the Linux Foundation. While this is logical, there is little evidence that this kind of large company or foundation support is even in the conceptual stage. In fact, nontraditional actors have shown little appetite for open source support beyond Linux itself, seeming to feel that the other 169,999 open source projects can fend for adoption on their own.

Of course, there is one tranche of the technology industry that one might logically expect to embrace open source: the system integrators. The business case for them pushing open source is even more logical than the non-profits: Less money spent on software means more is available for services. However, as I noted in a recent newsletter, thus far large SI organizations have preferred to continuing working with proprietary vendors; one might say, in fact, that large SIs also have a strong preference for working with large economic actors and reject working with ones that, despite an undeniably significant (though "unofficial") presence in IT infrastructures, are still small companies.

Overall, one could interpret Karim's analysis as quite pessimistic. It seems to imply that open source will never "cross the chasm" (to pull out a phrase much beloved in technology market analysis). I prefer to interpret things differently.

I've recently begun reading a biography of Joseph Schumpeter. While he is widely known and his theory of "creative destruction" is often referred to in a kind of self-congratulatory shorthand by technology industry participants, most people fail to comprehend the magnitude of upheaval his theory posits. In short, creative destruction tears up markets, decimating vendors - but also assaulting customers. Embracing changes brought on by technology can be just as uncomfortable for users as for suppliers.

Despite the discomfort, though, no matter the personal resistance, notwithstanding the preference for familiar but suboptimal solutions, open source eventually will invade "official" IT, if for no other reason than once one user in a market embraces it, others will be forced to as well due to competitive pressure. While IT organizations may resist open source with all their might, waiting for acceptable economic actors to appear bearing open source in a suitable form, eventually they will succumb to its inevitability. My money is on Schumpeter.