Thursday, March 20, 2008

4th wave of platform wars

Thursday I was fortune to hear Brent Williams of The Benchmark Company talk at EclipseCon about investing in open source. However, the former IDC/Gartner analyst most piqued my interest when he argued that signs are ripe for a 4th wave of platform wars.

Here is one of the slides summarizing his views.

1st-3rd Generation Proprietary Platforms (1965 -~1995)
  • Mainframe, mini, Windows PC
  • Extreme architectural & vendor lock-in
  • Lower complexity enabled fantasy of “one vendor could do it all.”

    Galactic Architectures (a comical detour, 1987-1995)
  • Everything to everyone, solving all problems
  • IBM SAA, Digital AAS, Apple VITAL, ACE Consortium, Sun ONE, etc.
Brent was obviously trying to be provocative, but his point about platforms often being driven by vendor needs — not user needs — rang very true.

He sees the signs that web-based platforms (like Google) will be the next wave of platforms. Even if that’s not who will eventually win, it makes the topic worth studying. Meanwhile, his hopes for involvement of user organizations in standardization seem fanciful at best.

I’ve long wanted to meet Brent, a friend of my co-author Siobhán O’Mahony. He was even more entertaining in person, and he also offered some interesting ideas of how to estimate company performance that would be useful for my students in researching competitors.


Brent Williams said...


Thanks for the kind words about my EclipseCon presentation in your blog.

I would like to clarify that my comments about the opportunity for (and need for) increased end user participation at all phases of the life cycle of a new infrastructure platform are not some sort of wishful "Summer of Love" vignette, where vendors and users will sit down in a big circle singing "Kumbayah" and holding hands.

Rather, I was thinking that the economics for many users of being able to influence the direction of products is compelling, especially if (or maybe "only if") there's a governance mechanism in place to ensure that user input is on par with vendor input into the requirements, features, etc. Customer advisory boards and other current mechanisms for providing feedback to vendors don't guarantee that customer input is treated as a priority.

Even if you devote one or two headcount to working with an open source project that's important to you, the potential savings from having a product that does exactly what you need could easily overshadow the cost of the developer resource assigned to the project, making it a very tangible economic benefit, rather than the "touchy feely" experience you hint at.


Joel West said...


Sorry I wasn't more clear on my point. I agree with you, user involvement is not kumbayah but economically rational. The technology would be better and all users would get more benefits if users (whether end-users or user organizations) were more directly involved.

My skepticism comes from the research of my friend Kai Jakobs, who has studied user involvement in standardization for years (if not decades). What happens is — exactly as you hint at in “Galactic Architectures” — the vendors are pushing the efforts to meet their goals, and not user goals.

Since they have more money at stake, the vendors supply more resources to standardization and their goals dominate. I see Eclipse and other OSS activities as susceptible to exactly the same pressure. Theoretically it could be fixed, but I think over the long run vendor self-interest will dominate.