Once More unto the Breach

Shai Agassi, SAP, and Open Source Software

In mid-November, Shai Agassi, (SAP, President of the Product and Technology Group) caused a bit of a stir with comments about "IP socialism" and open source in a public presentation.  The original news article is here, Shai's blog commentary defending and clarifying his statements is here, the vnunet follow-ups are here and here.  Mårten Mickos, MySQL CEO, follows up with an excellent response (here) being ever pragmatic and observing that actions speak stronger than words and pointing out some of the positive actions SAP has done around open source software.  This is where I want to go in this post. One of these actions demonstrates that as poor as Mr. Agassi's rhetoric is (and we'll tackle that in a moment), SAP actually "gets" open source business tactics as well as Sun and IBM. 

In September 2002, at the SAP developer's event, SAP announced the release of SAPDB.  SAPDB was free software in every sense of the word, including the General Public License (GPL).  Essentially, they acquired the assets of Adabase (yet another relational database that tanked against the Oracle juggernaut) from Software AG.  They then spent 100 people times two years modernizing the old code base, adding objects and XML to the engine, and so forth.  A conservative back-of-the-envelop estimate would put the engineering cost at about US$20M.  And then they released binaries (for Solaris, AIX, HP/UX, Windows2000 server, and Linux) and source under the GPL. The history of sapdb.org is captured here (courtesy of the Wayback machine). 

Certainly the German government must have loved this move.  A preeminent German software company paying its taxes and hiring lots of other tax payers also does free software, something the German government certainly seemed interested in at the time.   This was a great PR play.  (Indeed, I wonder what the research and development tax implications of this move were in Germany at the time?)

Initially, they even played the game very well as a large conservative corporation around patent/copyright taint issues.  While SAPDB was free software, and they built a support and user community around it, they weren't accepting changes back from the community.  They claimed at the time they were still "figuring out how community development worked" (and indeed their lawyers were probably sending loud fear signals around potential litigation as all large firm conservative lawyers are culturally and genetically required to do.) 

The following Spring, SAP passed the entire community problem off in a partnership with MySQL AB, a company with deep experience in managing community development and with a different legal risk profile.  SAPDB was rebranded MaxDB.  The timing was probably perfect for MySQL AB as well, because it would have been just prior to their venture funding round and I'm sure a partnership with SAP worked well in the valuation discussion.  Everyone wins — SAP, MySQL AB, and most importantly for both companies, customers win. 

Geoff Moore pointed out in the original Crossing the Chasm (1991) that companies win that put forward the best "whole product" image to customers, i.e. a core product and all its complements.  SAP presumably already dominates the Fortune 1000.  How would they grow their R3 business?  Well, they could go after the middle-tier businesses, but those customers may not want to pay the Oracle/Microsoft/IBM tax for the relational database to run R3.  So SAP gave them one for free.  R3 is core.  A database is a complement.  Of course, SAP didn't want to go into the database business (free binary only product with a large margin-eating support business), nor did they want to simply give away a 20 million dollar investment by publishing the source in such a way their competitors might be able to use it.  So they put it out under the most business conservative free and open source software license available, and made a good PR statement at the same time. 

MySQL was the perfect company to pick up the community development issues as they would be able to (hopefully) turn those MaxDB users into paying customers of MySQL over time as MySQL evolved in functionality, and they themselves were comfortable building a business around free software and the GPL.  Indeed, this would be introducing them to a whole new class of small enterprise customers. 

Neither is this IP hostile in the sense of the (mistaken) Microsoft FUD or Agassi's own naive statements in November.  This appears to be the mature IP strategy of a large IP savvy organization similar to publishing ahead or setting up patent pickets around one's competitors.  This is a company that can clearly see that it will spend its IP strategy dollars on assets it wishes to turn into "legal property" in their core business areas, and not in the complement spaces.  This is a strategy worthy of IBM in its moves around standards and open source, understanding that not every asset need be turned into legal property (with those attendant costs, timing, and risks) when they can be used as complements to drive core business revenue streams with customers.

Which is probably why Agassi's current IP-related comments smack of bad sophistry, or at least demonstrate the need of a mature PR policy to handle the growing problems SAP will face as open source alternatives begin to mature in some of their core business areas.  (They first need to learn it's a network and not a stack, and open source isn't eating it's way up the stack.)  SAP certainly has challenges to face.    But SAP would appear to be presently behaving as Microsoft and Oracle did three and four years ago, using tired naive rhetoric trying to frame a discussion around "property" and "innovation" rather than appreciating that all free and open source software licensing depends first and foremost on strong copyright laws, and innovation happens where you encourage it with little regard for how you license it.  Agassi's "we're open source because we share the source" sounds as awkward as early Microsoft Shared Source messaging before Jason Matusow evolved the message over time to the "move to the middle" language. 

Maybe that's what SAP needs at this juncture — either an Open Source Officer similar to Simon Phipps (if they're serious about an open source strategy) or at the very least a consummate diplomat similar to Jason Matusow (while they figure out what their strategy needs to be).  The current rhetoric seems at odds with the past behaviour. 

OSBC Wrap-up

The second Open Source Business Conference took place last week.  First,  an apology.  I really had every intention of blogging live as I did at the OSDL Enterprise Linux Summit in January and the conference simply got ahead of me and then I was in catch-up mode.  So instead I've pulled together a summary of all the brilliant links with some commentary along the way.  If you missed the conference this will give you the flavour of it all without the bad hotel coffee at break time. 

The conference kicked off with an opening keynote from Sun's Jonathan Schwartz.  His blog entry covers his keynote on the Participation Age.  Overall a mostly entertaining keynote, (I've not seen him speak before) but you've probably also seen it reported as Schwartz "attacking the GPL."  The Groklaw coverage is here although it's a bit extreme in lambasting Jonathan. Of course he is essentially being a senior vendor executive delivering the company message so one has to take the message liberally salted.

Here's the ZDNet coverage of Jonathon's keynote as well.  As reported, he also did the tired dance around not opening the source of Java because it might diverge from the "standard".  I had the pleasure of having that same discussion in a panel at OSBC last year with SunSoft CTO John Fowler.  At the time when Mr. Fowler tried that line on the audience I pointed out:

• Apache hasn't diverged off the HTTP spec since it was created.
• sendmail hasn't diverged off the SMTP spec since the software implemented it.
• The flip side IS true: at least one key OEM vendor in the mid-nineties shipped a C compiler that broke conformance with the ISO/ANSI C specification and didn't discover it until post-release.  They were very quiet about it until the next release cycle when they could re-align with the standard. 

Sorry Sun, you'll have to do better than that excuse.  And since Java isn't a de jure standard and you control the community process and the branding program, I still can't see how you'll loose control of the name and the spec and the implementation, etc. etc.

Jonathan's discussions around other topics were insightful, but you certainly had the feeling throughout that you were only seeing the large-vendor-approved-message without some of the context or concerns.  It would have been nice at something like the Open Source Business Conference to see the complete discussion that went on inside Sun. A little bit of opening the kimono would have gone a long way. 

Jonathan was followed by Larry Augustin in his new role as CEO at Medsphere.  Here's a good blog summary of his presentation.  The most fun was when he pointed out that if you take a traditional software company's SEC filing, and reduce R&D expense, dramatically drop sales and marketing costs (because customer's self select when they can try for free), and drop licensing revenues, you can still apply the typical MBA ratios and the numbers don't look bad. 

Day one was also the day we got another r0ml presentation.  (Please remember I have the extreme pleasure of working with r0ml at Optaros.)  As always, he forced us to think about things differently with a discussion around the Paradox of Choice, starting with some thinking from the book of the same title.  The finest bit of brain manipulation he did was to begin the conversation around organizing your IT software portfolio like a financial portfolio (i.e. diversity is good), which means you can actually apply some risk and option analysis to it.  Then start thinking about your enterprise support agreements from your large vendors and bond laddering to manage the future risk.  There could be a whole new set of financial tools for the corp procurement staff to use in their next discussions with vendors.  r0ml's slides are here (938.8K). 

Another classic keynote was Geoff Moore of Crossing the Chasm fame.  Ross Mayfield took exceptional notes and has posted them on his blog.  They capture the presentation perfectly.  One fun thing done at the start of the keynote (with a hat tip to James Surowiecki's The Wisdom of Crowds) was Mr. Moore put several OSS projects up on his modified bell curve representing the tech adoption life cycle and asked for the audience to confirm where the projects sit today.  Read Ross's blog to see the results. 

Ross Mayfield's blog has a couple of different OSBC entries covering the areas in which he was interested or directly involved.  You'll find some great reads:

• Meet the Community at OSBC
• Open Source Innovation

Larry Lessig gave a keynote again this year.  This is the fourth time I've seen him speak.  Every time is a genuine pleasure.  He covered familiar ground on the coming battle where large vendors will attempt to use patents and a patent system that's under siege to control innovation and creativity to protect their businesses.  In the questions at the end, when asked what we can do about it all, he reminded us (again) to consider how much we give to the cable company on a monthly basis versus funding to non-profits like the EFF that are fighting the battle for us in the courts. 

Mårten Mickos (MySQL) gave a keynote during the second afternoon, and it was as thought provoking and entertaining as always.  He remains one of the few CEOs that seems to give a real presentation on his business (as he did at LinuxWorld) and how they think about open source at MySQL.  You get the feeling he really is presenting the thinking and not corporate speak.  The topic at hand was Lagom, the Swedish concept of finding balance — neither doing too much or too little of something.  The MySQL Users Conference is coming up next week and he'll be presenting there as well.

I caught Jason Matusow's Shared Source talk.  Jason's talk was (another) very carefully positioned talk from a Microsoft perspective, essentially presenting material you can read in his blog on defining "open", yet another positioning of most-customers-don't-want-source must mean that source access isn't actually important, and sometimes Shared Source might mean Open Source.  Don't get me wrong  I worked with Jason at Microsoft and he's a mensch, a consummate professional, an entertaining presenter, and very good at what he does.  He even presented a little bit of what Microsoft learned from the WiX project which is a year old now.  But it would be so nice if Microsoft's presenters actually presented all the learning and all the thinking (even the bad), and quit the shades of grey and the paid "studies".  Really, it's not that scary.

The idea that large companies can present a real and balanced discussion without messaging to us still seems lost on them.  Why do the big vendors continue to think this is appropriate?  I missed the opening keynotes the second morning.   (I attended the OSI board meeting and met the new board).  There was some infomercial work apparently going on in the keynotes based on the hallway comments later.   Large vendors still don't get it.  Even when they blog, an audience can still smell messaging.  Give me company CEOs with real discussions like Augustin and Mickos every time, even if they don't have time to blog. 

Here are some excellent wrap up links:

• Matt Asay (conference creator and organizer) summarized his experiences here.
• Kevin Schockey offers up mostly real-time coverage on the O'Reilly Network with a Kick-off, some SpikeSource coverage, and some Groundwork coverage, and Day 1 and Day 2 summaries.
• Steve Mallet's wrap on OSDir.com.
  
• The Flikr file is growing and there's the promise of audio to come at IT Conversations.

I didn't follow the Legal or Venture Capital tracks.  If someone has good pointers or summaries, please email them to me and I'll update this entry.

Updated (2005-04-15, 15:39 PDT): Included an updated link to r0ml's slides. 

LinuxWorld: Fear & Loathing in Linux World: Commoditization, Commercialization and the Customer

I caught Mårten Mickos's keynote yesterday at LinuxWorld, before a last run through the trade show floor and out.  He's always a trend setter.  First, it wasn't a 45 minute infomercial for MySQL AB, but rather a talk about how to think about an open source based business.  Second, he said some brilliant things.  Here's a smattering:

The first community that arrives in an open source project are those that are willing to spend time to save money.  The second community are those that are willing to spend money to save time. 

He observed that the license is a balancing point between your community and customers.  It isn't just finding the compelling reason to buy something, but you have to ensure you don't create a compelling reason for the community to abandon the project.   

You can protect your open source business with:

• copyrights
• trademarks
• trade secret
• superior service
• product stewardship

While he didn't think there were trade secrets in MySQL (i.e. the code base), I would ask the lawyers amongst us to let us know if his customer list, and things the MySQL engineering team knows won't work (what they left out of the code) don't constitute things still protected as trade secrets.

He observed that the open source world is helping standardize licenses.  Approximately 70% of the 88,000 projects on SourceForge.net use the GPL.  If you had 88,000 proprietary products, you would likely have 88,000 different EULA. 

A vendor associated with an open source project provides stewardship, and a road map. 
He finished up on the forward looking note that he believes (as so many of us do) that open source is not eating it's way up the stack to commoditize all value out of software, but rather only a small fraction of the world's software needs have been met so far.  A wonderful keynote.