Once More unto the Breach

OSCON 2006: IBM, Microsoft, and Open Source Strategy

Bob Sutor gave a fine talk at OSCON 2006 in Portland last week called "The Future of Software is a Blend, but of what?" (slides here).  It was a classic IBM talk.  He developed an architecture of thinking about using free and open source software in the enterprise.  It is obviously a mixed environment between free/open software and closed software.  The additional questions of what is commercial or not and what standards-based software is used opens the discussion into a multi-variable landscape that reflects the real world of most enterprise software architectures.  Of course, when discussing how one might consider these options against closed product decisions, IBM software choices were made. 

Bob's use of language was perfect.  For example, when one needed to "step up to a scalable enterprise database", it was of course DB2.  (Look for the transition between slides 13 and 14.) This is exactly the sort of perfectly pitched rhetoric we've seen from IBM over the years, back through initial engagements with the Apache and Linux communities, and on through the development of the Eclipse community.

Before the reader thinks I'm bashing at IBM, let's look at Microsoft's OSCON 2006 engagement.  Danese Cooper (Intel) interviewed Bill Hilf (Microsoft) during an Executive Briefing session on Tuesday before the main conference opened.  Bill is director of the Microsoft Open Source Lab, and helped create Port25, which is a careful view into Microsoft's Shared Source world similar to Channel9. 

Danese asked lots of difficult questions, and Bill gave lots of pro forma answers that typically involved: "That's a good question." + a restatement of the question + the scripted response from the Rude Q&A + "Microsoft is still learning from the open source community." 

[A Rude Q&A is the document Microsoft teams develop with PR before announcements and events that covers all the questions you DON'T want to be asked by the press and analysts, and what the best response can be.  There is an art to writing these documents and it helps keep the message delivery sharp.]

Danese didn't probe too hard because it would have been impolite to beat up a Silver Sponsor, which is pretty acceptable.  And in Bill's defense, he's worked extremely hard with Jason Matusow this past few years to get the message to the point that the Microsoft executive team isn't still running around offering personal opinions based on conjecture.  So the fact that we didn't learn anything other than the Microsoft rhetoric is being buffed to a very high shine isn't surprising. 

But here's the simple difference between the IBM and Microsoft open source strategies.  IBM participates.  Microsoft talks about "still learning from open source."  Port25 is interesting.  Re-casting and re-branding the plethora of licenses to the few Shared Source licenses is nice.  Cutting a few partnership deals with companies creating products based on open source is just business.  Opening a portal where other people can share code is fine.  But it all pales in comparison to IBM employees' direct and constant participation in project communities around the open source world, including the creation of their own, for what has to be going on eight years. 

Whether it's a lack of executive will, the frenzied fear of IP loss or litigation from Legal Affairs, or the cultural inability to collaborate because of the historical competitive nature of the company, Microsoft professionally talks up a weak story and refuses to actually engage and participate.  Maybe it's just pride (which is really just another word for fear).  IBM's practiced rhetoric is much much more acceptable and credible, because at the end of the day they participate and contribute. IBM deeply understands the competitive advantage and business credibility it brings.

In the end, Microsoft is going to have to DO something. Be Bold.  Contribute.  Participate. Risk something.  At one point in the interview with Danese, Bill said that Gates and Ozzie support his team's work.  There's a huge difference, however, between supporting a mean-well market message and supporting the cultural shift it takes to actively participate. 

Vive la Société libre! [With appropriate apologies to de Gaulle.]

Post to del.icio.us

IBM, Patents, and Open Source Software

There is something wonderful about watching IBM execute.  Terrifying perhaps if you're the competition, but wonderful nonetheless.   Last week they announced several initiatives around the U.S. patent system and open source software.  Bob Sutor covers the three initiatives in his blog entry. 

The Open Patent Review initiative sounds brilliant.  Let the public participate in patent review.  By registering an interest in particular domains (presumably via RSS feeds), people that have experience in an area can contribute and keep track of what's happening that might impact them.  It wouldn't take much imagination to see reputation systems wrapped around this such that different "community examiners" begin to stand out based on the ratings of their peers and the usefulness of their contributions.  (Likewise cranks can be weeded out.)  Beth Novak is running a series of workshops on the initiative.  You can find out more information at her website: http://dotank.nyls.edu/communitypatent.

Then there's the Open Source as Prior Art initiative.  This work is being done in conjunction with the Open Source Development Lab, Novell, Red Hat, Eclipse, and SourceForge.net.  Obviously this has a bit of a Linux-centric view of the world, but by adding SourceForge they can test the environment well beyond the Linux world and it presumably wouldn't be too difficult for the Apache Software Foundation, or the Gnome Foundation to get involved.  Hopefully Google gets involved before Yahoo or Microsoft wake up and smell the coffee here.  You can find more information at this website: http://developer.osdl.org/dev/priorart/information.html.

The Patent Quality Index sounds intriguing.  I'm a little more skeptical here perhaps, but look forward to seeing where it goes. 

While the initiatives are each interesting in their own right, it is the business execution on IBM's side that remains brilliant.  It is how IBM is attacking the problems of a congested system, when they clearly continue to patent intellectual assets to manage vendor-to-vendor competitive situations.  Consider:

• HP, Sun, Oracle and Microsoft don't appear to have been invited to the table.  This means they aren't paying attention.  HP is a primary patent filing company like IBM on the hardware side.  (Hmmm -- so is Intel on chip technology.)  And then of course one might imagine Oracle and Microsoft care just a little about software patents.  Any move any of them make now (i.) looks like playing catch-up, (ii.) has to co-ordinate with the existing first public moves of IBM else they look divisive in the teeth of IBM's community and academically focused initiatives, (iii.) is already less influential.
• Last Spring (2005), Microsoft and Oracle announced their desire to see patent reform in the U.S. except they're trying to do it via the U.S. Congress, in the teeth of a system designed to protect the bigger interests of industries like the pharmaceutical industry.  Here, instead, IBM is avoiding the entire expensive (lobbying), time-consuming, messy legislative route and plugging straight into the USPTO to help improve the tools they can use and dealing (potentially) directly with the human bandwidth and expertise problem the USPTO has regardless of what legislative changes may be enacted at some future time.

I once heard a story (probably apocryphal) of how IBM convinced the USPTO to consider the IBM Systems Journal to be one of the sources of prior art that was searched.  This is breathtaking in its brilliance.  They could mainline directly into USPTO:  And it's genuinely helpful.  The question becomes: What would they want to publish or patent in the teeth of their competitors this issue?

While apocryphal, it's believable because there is a maturity to the intellectual asset strategy in it that goes well beyond more-patents-faster, and speaks to the same level of brilliance of their new initiatives.  This will be a fascinating set of initiatives to watch. 

Open Standards, Patents, and the Open Document Format Play

Bob Sutor (IBM, vice-president of standards and open source) recently posted a great discussion on defining an open standard.  This is important because we are still as an industry struggling to articulate the business relationship between standards and open source.  We also continue to try to articulate the business rationale that accounts for patents in the context of open source or standards.  One doesn't have to agree or disagree with software patents, but one does need to deal with the reality of their existence.

After laying out a set of criteria, and some discussion of the problems in definitions in this space, Bob offers the following perfect observation:

If you have read this and conclude "ha, we can't really get an open standard so we'll muddle through and do what we want", then you have misunderstood what I am talking about. Since most standards fall between closed and this "purely open" characterization, you need to decide what is important to you. For example, a government may decide that it really cares about Maintenance, Accession, and Implementation and will try to use standards that maximize those characteristics while being willing to concede a bit on Development and Modification by Others.

There isn't one "right" definition here, and the context is different for different developers and consumers of standards at different times. And yes, you might imagine Bob's example there was for the benefit of the Commonwealth of Massachusetts. 

Another recent post in this space that provides context based on the economics of participation is Andy Updegrove's blog discussion on standards and patent pools.  Andy is a standards lawyer (literally) at Gesmer-Updegrove and knows whereof he speaks.  Mid-post he delivers a great synopsis on the economic benefits of a patent pool.  One item left out of the list would be that the pool companies can also stand united against a lone litigious patent holder in the space. Remember: the point isn't to validate software patents, but rather to deal with their reality. 

Some good work has been done by a number of people attempting to define the term "open standard" with respect to the freedom to implement such standards in a royalty free unencumbered way.  Unfortunately, such definitions put the standards body in an indefensible position.  The open source definition applied by a community to its project rested within the control of that community.  The same can not be said for the open standards definitions that exist, in so far as an outsider to the group attempting to develop an open standard can arrive on the scene with a patent and put the definition of "open" at risk for that standards project. 

I've offered views on patents and standards before in the context of the OASIS IPR policy and around patents as no more than tickets to a negotiation.  The decision of the Commonwealth of Massachusetts to move to the Open Document Format, however, will bring this all back up for discussion. 

Microsoft is in a bind at this point.  They appear to have thought that by publishing the proprietary specifications for their Office formats, accompanied by a royalty free license for any essential patents, they would have a reasonable debating point that they were as "open" as any standard.  The fact that no one else has claimed to be implementing them doesn't help their case that they're trying to encourage multiple implementations, but then that probably wasn't their desire.  (And what about a license for other non-essential patents that might be infringed?  Are they to be royalty free as well, or is that the troll beneath the bridge to be brought out with the excuse that "it's our property".) 

This is not the first time Microsoft has been hit on royalty free patent licenses. In the context of the SenderID debate through the Summer 2004, Microsoft went out of its way to do the right thing.  They announced they had applied for patents that might read on the IETF SenderID anti-spam standard (which wasn't required of them), offered a royalty free license and published its text for all to see (which was pretty upfront considering they didn't have the patents awarded at the time), BUT the patent license wasn't sublicensable so certain free and open source implementations would be prevented if they were licensed under the GPL. 

They appear to have a legitimate legal concern that if software implementing the standard was licensed under the GPL, and subsequently someone used that software source code for a project unrelated to the standard and still licensed under the GPL (through the GPL's attachment), then the scope of the patent could creep and they might lose their ability to license the patent in other contexts under different (presumably royalty bearing) terms. 

My question for the patent lawyers "in the room": if there was a way to structure the royalty-free patent license to be strictly for implementing a particular standard (or group of standards), and allow it to be sublicensed in that context (so the GPL could be used), then doesn't the GPL become the guard on the patent?  The GPL requires a user to have all the rights to distribute the software.  The only patent rights granted in the sublicensable royalty-free license are to implement the standard.  Anyone trying to use that software for other purposes and distribute it would be prevented by the GPL from so doing.  Indeed, Microsoft could ask the Free Software Foundation for help enforcing the license.  Just a thought.  I'm clearly not a lawyer, so I would defer to legal minds on this one. 

The subtleties here, however, are not to be tossed aside.  The curtain has just raised on the document formats play.  While Microsoft players are still fumbling their lines in the early scenes, it's unclear the first act is over, and the rest is still being written. 

IBM, Gluecode, and Open Source Business as Usual

Update: You can find a podcast of this blog entry here.
IBM announced the acquisition of Gluecode this week. In their own words: "Now customers and Business Partners can tap into the low cost of entry of open source technology to quickly develop and deploy applications, and migrate to WebSphere software as business needs expand." There's some nice commentary around the benefits of open source and the strength of trusted enterprise level service and support, and they very clearly position the Gluecode Geronimo offering as entry level application serving on a diagram showing the evolution to scaled high transactions on Websphere. Here's the CNET article on the announcement with some good thinking around it.

Stephen O'Grady (RedMonk) has a great Q&A styled analysis of the acquisition which I would encourage you to read, so I'll provide (hopefull) supporting commentary.

• IBM seems to understand Geoff Moore's Technology Adoption Life Cycle really well. They have done a lot of work around Linux (with an incredible number of influence touch points) to manage the AIX curve. While selling Linux aggressively against Sun, they have told us that AIX will absolutely be replaced by Linux ... in 10 years. This is about customer account management and control and giving customers well crafted messages through the relationship.
• Recently, Gartner moved JBoss into the same upper right quadrant of the app server world with Microsoft, WebSphere, and WebLogic. Love such reports or hate them, a lot of enterprise executives read Gartner reports and attend Gartner conferences. So IBM needs to ensure they deliver the same well crafted message around Websphere as they did around AIX. While they can talk about entry level app serving around Gluecode, this isn't about small-medium business — it's about enterprise messaging.
• IBM is well versed in the game of Apache licensed code.  The licensing discussion is nicely encapsulated here: "The rationale for Geronimo over JBoss was that it would be more open, less vulnerable to the whims of its owner vendor. Instead, even though IBM haven't bought ownership of the Geronimo code, they do own the core developers. And every contribution made by third parties in the OSS codebase ends up benefiting the IBM distro. That is the price of the BSD license: you don't need to publish your additions, but everyone else has the same right. Which is precisely why (L)GPL makes so much sense for startups trying to retain control of their software. MIT/BSD/Apache licenses are good for universal adoption, but not retaining control of "strategic" technologies."
• Marc Fleury's commentary is to be expected, but I think he misses a few things:
  • IBM has been using open source as a business tool for 7 years at this point since joining the Apache community. They have a lot of experience here, building over time, from Apache involvement, to Linux, to their own Eclipse strategy.
  • Marc says IBM wants to kill JBoss.  IBM is far more interested in account control and the customer relationship than they are in the competition. The community may have strong opinions about the relative merits of Geronimo and JBoss, but the average CIO reads the news. They now have reason to pause when they look for a replacement for expensive Websphere licenses using scaled out multimachine solutions, instead of simply exploring JBoss, or Microsoft's embedded solution. While Gartner says JBoss is maturing, IBM is loudly doing something with Gluecode, so the careful CIO will look at both — and customers will ask them in to discuss things and the conversation can begin, if indeed it's not merely a continuation of an existing conversation. You sometimes need to perturb the message of your competitors to begin the conversation with your customers. Through the acquisition they can perturb the message of Sun, BEA, JBoss, and Microsoft all at once. It's a bold enough move, all the existing CIO customer base is probably calling their IBM account execs to explain it to them.
  • "While we are focused on improving developer productivity with our work on simplifying Enterprise Java at the EJB3 standards level and in our own JEMS product suite, IBM's announcement adds complexity for their customers and partners, not simplicity." Don't misunderstand the feet on the street power IBM applies around "industry standards" and the number of touch points for influence they have across the plethora of related consortia and standards efforts. Even if you understand the power of this, you won't have enough people in a company the size of JBoss to keep up. IBM will bury working groups in virtual paper. Don't forget your partners have their own agendas and when it comes to standards, IBM remains one of the few companies that genuinely uses its reach to participate rather than issue press releases about why some effort isn't important.  In the standards arena, participation equals influence.
  • "If evolution has anything to teach us, it is that hostile environments accelerate mutations and natural selection favors the fittest organisms. We are the leaner, faster party, with the true professional open source community base." Marc's confidense is admirable. He might forget, however, that lots of agile species die while figuring out the environment. IBM isn't a dinosaur, it's an elephant, and elephants and their evolutionary kin have been around a long time and learned to dance.

The risks of embedding a GPL licensed JBoss would be higher (though probably not insurmountable.) So Marc probably didn't get an offer.   

But Matt Asay's blogging of the acquisition takes us into a slightly different view of the space.  As he correctly points out, a driver for acquisitions is expertise. This is why one might buy rather than participate. IBM certainly has that level of the game figured out, as seen in their participation around Apache, Linux, and the development of Eclipse. Websphere integration (and Eclipse integration) can more easily be done when you employ the community leaders.  But it gets better: think Harmony.

Stephen O'Grady's summary on Harmony is a good summary pulling together the relevant posts across the discussion (including Miguel de Icaza's mono-coloured experiences).  But with Gier Magnusson as a central figure in Harmony now at IBM through the Gluecode acquisition, and the fact that IBM has a JVM that could be licensed as open source, and they have the testing resources to solve any compatibility concerns, it makes the discussion more interesting in the never ending influence games around Java.