Embedded OS technonationalism

When I was beginning my doctoral studies in the mid-1990s,
the U.S. was still worrying about global competition between technological rivals like Japan and Europe. Among a certain crowd, everything was about economic competition within the “Triad,” Ken’ichi Omae’s term for the US, Japan and Europe. (Obviously before the rise of Korea’s per capita GDP and China’s trade surpluses).

A few American scholars even talked about “technonationalism,” normally in the context of Japanese industrial policy that was aimed at the US (and that we should emulate). (I was briefly part of this crowd, and published my first few papers in 1994-96 quoting some of this work, but never made any significant contributions.) This was both fueled by and fueled the Clinton Administration’s foreign policy doctrine that war was obsolete and economic competition was the wave of the future.

One of the biggest pre-Clinton proponents of such arguments was Laura Tyson, whose treatise Who’s Bashing Whom? earned her a ticket from Berkeley to chair President Clinton’s Council of Economics Advisors. After the White House gig, Tyson went back to Berkeley’s Haas School as dean and then moved on to become dean at London Business School before returning to Berkeley. (There’s no record of her covering any Sting songs or even Blondie songs after either the CEA or dean gigs). Meanwhile, the economist passed over for the White House job has become a bitter New York Times columnnist who no longer does serious research.

All of this being a long forward to an odd article from the SJ Mercury that I read Tuesday morning on the plane (out to Philadelphia) about Japanese industrial policy for automotive embedded operating systems. The Japanese government kicked in $8.4 million [check yen] to help fund a consortium of 10 Japanese firms to develop said OS. It includes the big three auto makers (Toyota, Nissan, Honda) and major electronics suppliers like Denso (which as Nippondenso began life as was a Toyota division) and Toshiba. The story of the Japan Automotive Software Platform (JasPar) was also written up Sunday in Yomiuri (one of the big Japanese dailies) and PC World.

The money is coming from the Ministry of Economy, Trade and Industry, née Ministry of International Trade and Industry, the famed tsûsanshô of Japan’s postwar economic miracle. Spending $8m on embedded car software seems like a far cry from MITI’s glory days.

In fact, Chalmers Johnson’s book on MITI and the Japanese Miracle was required reading for budding technonationalists (or for anyone else in comparitive political economy) in the early 1990s. Listening to Chal’s night school class in 1993-1994 (and the associated office hours) had more influence on my becoming an academic than anyone else.

The MITI of the 1950s and 1960s used trade restrictions to build up Japan’s infant industries. Mark Mason studied how MITi delayed the TI integrated circuit patent in Japan long enough for Japanese firms to develop their own semiconductor capabitilies without having to pay royalties. Marie Anchordoguy (the inspirtiaton for my earliest Japan research) did her Ph.D. thesis at Berkeley (before it became Haas) on how MITI built up Fujitsu and other Japanese mainframe makers, using liberal financing in the domestic market to enable Japanese firms to undercut the prices of the the superior IBM computers.

It’s not just the Japanese, since nowadays the French view of industrial policy (dirigisme) has become the norm at the EU: after one success in a perfect storm of market timing and technology opportunity (GSM), the EC has spent almost two decades in a futile attempt to recreate that storm. (Tuesday’s report implied the Europeans also have their own automobile software consortium but I couldn’t find it, only their trade association.) Presumably the Koreans and Chinese will follow with their own national consortia, and then the US Big 2½ automakers will scream to Washington that they need their own subsidies.

But I can’t figure out what the deal is with embedded operating systems — it seems such small potatoes. The Mercury newspaper article parroted the Japanese press release “A standardized OS across the industry could have some big economic benefits, making it easier to bring new automotive technology into multiple models and bring down costs.”

Standardization has such benefits, but that doesn’t mean the Japanese carmakers have to go build their own. They could standardize tomorrow on an off-the-shelf solution from Wind River or MontaVista and get the same benefits — unless of course “bring down costs” is a code phrase for “stop paying foreign royalties.”

Don’t want royalties? Embedded Linux (or BSD if you dislike compusory sharing) is already available off the shelf for any firm. Why is the government involved? Is it that METI has to put up seed money to get Japanese firms to share and work together?

As industrial policy goes, it’s much less interesting than something major like TD-SCDMA. I suppose government industrial policy for an embedded OS is like the old line about the talking dog — what’s interesting is that he talks, not what he says.

Instead of responding with more technonationalism, perhaps MontaVista (or even Wind River) could put together an international open source coalition based on open source technologies. Given the popularity of open source in Scandanavia, they might able to peel off Saab and Volvo, which (this week at least) remain American-owned. If they did, the Chinese firms could choose to leverage the GPL technology while free-riding off the toothless enforcement of compulsory sharing in China.

Technorati Tags: compulsory sharing, consortia, embedded Linux, EU, industrial policy, Japan

Skype, the GPL scofflaw

For once, a story about Skype that has nothing to do with its business model, its impact on the telecom industry, pushback by powerful incumbents, or anything else having to with its real business.

This week, the blogs and Linux sites are abuzz because Skype supposedly lost a court case for violating the terms of the GPL (v2). The Inquirer (of Britain) started on Wednesday, then PC World on Thursday, then Linux Devices on Friday. Linux Devices notes that all claims about the case trace to an article in one German magazine, Golem, and can’t be independently verified, and it seems to be the only story with significant original reporting.

The purported case revolves not around Skype’s software, but its distribution of a Linux-based Wi-Fi phone, SMC’s WSKP100; a separate action is also said to be pending against SMC Networks, although Linux Devices could not verify that. This is a typical embedded GPL dispute, ala TiVo or anything else.

If we assume that everything reported is true, there were several interesting things about the case. First, it was about a marginal violation of how the code was distributed, and not something about the core compulsory sharing (some call “viral”) proposition of the GPL: once you put your code with my code, your code has to adhere to my rules. It doesn’t seem like Skype would ever be a good test case, since they already provide their source code under at least some conditions.

Second, that the case was brought by a German gadfly, GPL-Violations.org, and not the Free Software Foundation. Finally, it adds to the many cases we have in Europe, but AFAIK we still don’t have a relevant legal precedent in the U.S., or, even more importantly, in China.

So are there no GPL zealots in the US (unlikely) or China (probably not worth dying for)? Are there no comparable violations in the US (also unlikely) or China (darn near impossible)? Or are the relevant courts not as interested as German courts in enforcing compulsory sharing?

Technorati Tags: compulsory sharing, GPL, open source licenses, Skype

Non-compulsory sharing

One of the original goals of the GPLv3 was to make application services trigger the compulsory sharing clause. Now CNET reports that Google has shared its changes to MySQL to improve its reliability and manageability.

Google is famously secretive, treating most of its internal technical and human systems as trade secrets. Despite this, it’s decided that it’s worth sharing its MySQL changes — even though it doesn’t have to. Perhaps it wants its changes to become part of the main distribution, as I found Lawrence Livermore when shared its Linux clustering changes. Or maybe it’s hoping others will build upon and improve its code.

Either way, sharing happens without holding a gun to people’s heads. It happens in BSD. It happens in Apache. Sure, there’s a tension between common and private interest, but sharing already happens today when it’s economically rational.

Technorati Tags: compulsory sharing, Google, GPL

IBM and Joel vs. FSF and Matt

The latest update on the GPLv3 saga is that it has (apparently) been dissed by IBM’s top software honcho, Steve Mills. A lifetime IBM employee, Mills was the subject of a front page article (subscription required) in Monday’s Wall Street Journal:

Increasingly, the public face of International Business Machines Corp. is that of Steve Mills.

Mr. Mills heads IBM's rich and acquisition-hungry software unit, which has buoyed results in recent quarters. He frequently represents the company at investor conferences and software-customer gatherings, and his rising profile reflects a new reality at the technology giant.

IBM still gets most of its reputation from its computers and most of its revenue from services, but most of its profit growth comes from software. … On a stand-alone basis, IBM would have had the second-highest revenue of any software company after Microsoft Corp. Software revenue grew 14.4% in the fourth quarter and reached $18.2 billion last year. Software accounts for only 20% of total revenue -- but 40% of earnings. Fast-growing IBM brands include WebSphere, a variety of Internet tools for business, which gained 23% last year; Tivoli, which manages computer systems, up 26%; and Lotus, which makes email software, up 12%.

Last week, the trade journal CRN tried to drag Mills into the ongoing GPLv3 controversy.

"At some point you become so shrill and beyond what's required that you lose the audience and the audience moves on to something else," he said.

"We'll have to see what finally evolves through the [GPL] process, it's going through an update and the Free Software Foundation has a particular view of free software. Free software is a wonderful thing but there's also a business model."

"We think there are other licensing techniques, the Apache license and others are somewhat less onerous. We use them ourselves. We don't use the GPL for reasons of its restrictions," Mills said.

On Tuesday, my friend Matt Asay (co-founder of OSBC and former Novell OSS strategist) slammed Mills:

But IBM's fetish for all things Apache has kept it from seeing open source as a tool that it can monetize directly…

It's not clear what audience Mills is worried about the FSF/GPL losing. After all, the GPL governs over 72% of the projects on Sourceforge. He may well wish that Linux, Alfresco, Jasper Reports, Xen, etc. etc. were Apache-licensed so that he could drop them into his proprietary products and keep to his 20th Century business model. But just because it's comfortable for him doesn't mean that the open source world should capitulate to his whims.

So IBM hasn't figured out what the rest of us know with ever-increasing certitude: it's possible to monetize open source directly. Ironically, it becomes easier the more freedom that imbues the software. Even more ironically, this is so because companies like IBM don't want to touch software that is free - it threatens their proprietary software.

I think highly of IBM, but find its antipathy to the GPL to be silly.

I think highly of IBM too, because its early support for open source made it legitimate for IT buyers around the world. Unlike the one-trick ponies of open source startups (that depend on such things like dual licensing), IBM has the broadest range of open source participation and greatest diversity of open source business models of any company in the world. It creates open source and gives it away (like Jikes), it takes its software to create a new community (Eclipse), it installs and supports (the GPL-licensed) Linux and pays to support the Linux Foundation (née OSDL), and they were the first major corporation to back the Apache Foundation. The list goes on: there is no “IBM” strategy for open source, because no company with 330,000 employees can think with one mind or speak with one voice.

As for Matt’s criticisms, the easiest to knock down is license popularity on SourceForge. The vast majority of software projects on SourceForge are vanity projects, irrelevant to business, consumers or the economy. Comparing SourceForge license choices to those of real software is like comparing the IT choices of bloggers to those of major news organizations.

It’s also silly (as Matt knows) to suggest that IBM has an Apache fetish. IBM, after all, brought us the IBM Public License (which became the Common Public License which became the Eclipse Public License), one of the first licenses derived from the seminal Mozilla Public License. Using the CPL/EPL, IBM created the Eclipse project — the first open source project to really integrate vendor sponsors, a non-profit foundation and the community from day one.

Instead, I think Matt has a GPL fetish. In its most narrow (some would say precise) construction, the term “open source” means any license approved by the OSI that conforms to its Open Source Definition. The Free Software Foundation notwithstanding, there is nothing in the OSD (or the OSI policies) that say that the GPL is any better than the MPL, EPL or even the Apache or BSD licenses.

I think highly of Matt Asay, but his love affair with the GPL is silly.

Graphic credit: Chris Onsted’s Achewood cartoon, via Joey deVilla’s blog

Technorati Tags: compulsory sharing, GPL, open source licenses, IBM