I do have a suggestion for you: One of the major problems in getting corporations to adopt Open Source or any technical innovation is the issue of an inability to predict things like technical trends and ship dates. The Web page at www.ideosphere.com is an experimental example of the kind of techniques that might go a long way towards alleviating these kinds of problems. The insurance business has used these kinds of techniques for a long time (Lloyds of London is a very old example). The Ideosphere implementation is primitive, and doesn't work well for long-term projections, but it _is_ a _start_.

For example, right now, I have a client that would love to convert their clients to Linux, but they really don't have a good idea on when the Wine interface for Linux is likely to be ready for prime-time-and they really do need compatibility with Microsoft Office. One of the projects I'm working on a means to assess for them when these products will be ready.

In the commercial software world, these aren't such huge problems, people are used to Microsoft making announcements and slipping predictably. In comparison, in the Open Source world, the analysis is much more tricky. I think that if a lot of decision makers had a more accurate picture of what is coming,they could make much better decisions.

RJB

Randall,

You're right that inability to predict ship dates is something of a problem in the Open Source world, but I'm surprised that you think it's any better in the commercial software world! Yes, it is predictable that NT 5 will slip, but that's not much use for planning purposes.

I think the major issue is really that commercial vendors like Microsoft do a better job of providing information to the mainstream about what they are doing and thinking. Microsoft has an army of PR people making sure that, whether the release information is true or not, at least the process is managed. So they are good at creating the illusion of predictability.

Those of us who are working on commercial releases of Open Source (companies like Red Hat, O'Reilly, ActiveState, or Scriptics) do realize that we need to provide more information to corporate America about the state of the software.

A related issue is the fact that Open Source release names are often not created with marketing in mind. A good example of this is Perl 5.005, which was a far more substantial update to Perl than, for instance, Windows 98 was to Windows 95. The name makes it sound like a minor release, when in fact it was quite significant. We're trying to persuade the core developers to make the next release 5.6 rather than 5.006, just because it will sound more acceptable to corporate buyers.

While the indifference of the Open Source community to marketing hype is part of its charm, it is nonetheless important to take the shaping of customer perceptions seriously.

One way to do this is for the release numbers of commercial versions not necessarily to track the underlying software release. Thus, we have Red Hat 6 on its way, while the Linux kernel is at 2.2. Some of the other software packages need similar differentiation between what the hackers are using and packaged releases that may be more suitable for corporate America.

We're actually starting to do this with Perl. This summer we're going to be releasing a shrinkwrapped Perl distribution, the O'Reilly Perl Toolkit, containing Perl itself, plus some of the most important modules, precompiled for a number of popular architectures.

Back to your question about Wine--another important point is that as more companies become involved with Linux, you can expect more resources explicitly devoted to important projects, and particular companies leading the charge on specific developments. Wine is one of those aspects of Linux where I think we'll be seeing a lot of accelerated development soon.

Regarding Ideosphere, it's an intriguing concept, but I'm not sure it's all that relevant to your question. It tracks people's opinions about what will happen, but (except for things like stock market prices, which are heavily influenced by what people think will happen) it isn't what outsiders think that matters. It's what the people who are doing the work think. Most Open Source software advances are fairly predictable once you get in touch with the people closest to the project, so no probability estimates are needed.

That brings me back to my initial point: what we need is better communication between the Open Source developers and mainstream users. That's one of the things we're trying to provide at O'Reilly. I'm looking seriously at providing a market research service that would allow corporate users to get the kind of assurance and analytical support that they can now get from a variety of market research firms for commercial software and technology trends. (Any corporate users out there who might be interested in such a service should let me know.)

A final point about Ideosphere: are you aware that it was Christine Peterson of the Foresight Institute (which runs Ideosphere) who came up with the term "Open Source" in a meeting with Bruce Perens and Eric Raymond?

--Tim

Return to: Ask Tim Archive