[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: GPLv3 comedy unfolding -- Eben: "Copyleft Capitalism: GPLv3 and the
Re: GPLv3 comedy unfolding -- Eben: "Copyleft Capitalism: GPLv3 and the Future of Software Innovation"
Sat, 03 Nov 2007 17:04:16 +0100
Today I heard the best single presentation on open-source and free
software I have ever personally attended, in a talk by Columbia
Professor of Law Eben Moglen, on the topic Copyleft Capitalism: GPLv3
and the Future of Software Innovation. He spoke for over 90 minutes. I
understand the session was recorded, and I have asked that a transcript
be prepared. I write the rest of this post based on the notes I took
during his extraordinary talk.
Eben was introduced by Joe Latone of IBM Research. He mentioned that
almost everyone knows one or more of the following facts about Eben:
He is a Professor at Columbia Law School;
He worked for the Free Software Foundation for over a decade, serving as
its counsel and was thus Richard Stallmans chief legal adviser in the
drafting of Version 3 of the GPL License, gplv3;
He is currently the Chairman of the Software Freedom Law Center,which,
in its own words,SFLC provides legal representation and other
law-related services to protect and advance Free and Open Source
Software. Founded in 2005, the Center now represents many of the most
important and well-established free software and open source projects.
Joe then remarked that few people know that:
Eben began his career as a systems programmer for IBM in 1979;
After receiving his Law degree, Eben worked as an IP (intellectual
property) attorney for IBM in the Yorktown Lab of IBM Research;
I know one additional fact about Eben that I think is worth knowing as
you read my notes below, as well as any of his other writing.
According to his resume, Ebens degrees include:
1993, Yale University, Ph.D. with Distinction in History
1985, Yale University, J.D.; M.Phil. in History (with honors) Articles
Editor, Yale Law Journal, Vol. 93.
1976-80, Swarthmore College. A.B. (High Honors), 1980. Phi Beta Kappa.
Philip Hicks Prize, Literary Criticism.
If you read his resume closely, you can see that for a year he attended
one of the best law schools in the history of this country. Admission
was very selective, for a single individual served as the Dean,
Admissions Office, and Faculty of this school. Its students were called
clerks, and he taught them personally by example each and every day
they had the great good fortune to work under his tutelage. He was one
of the greatest figures in American jurisprudence:
1986-87, Law Clerk to Justice Thurgood Marshall, United States Supreme
See for example, Ebens A Vigil For Thurgood Marshall, which begins as
Three days after his death, on January 27, 1993, Thurgood Marshall came
to the Supreme Court, up the marble steps, for the last time. Congress
had ordered Abraham Lincolns catafalque brought to the Court, and on it
the casket of Thurgood Marshall lay in state. His beloved Chief, Earl
Warren, had been so honored in the Great Hall of the Court, and no one
else. Congress was right about the bier, and spoke with the voice of the
people: no other American, of any age, so deserved to lie where Lincoln
I brought along a few members of my Linux Menagerie, a group I call the
Tuxers, and placed them on the podium before Eben began his talk.
Eben Moglen and LTC Tuxers prepare to speak.
Here then my notes:
Copyleft Capitalism: GPLv3 and the Future of Software Innovation
Eben Moglen, Professor, Columbia Law School
gplv2 and gplv3 as technologies
cannot separate law, technology and politics
how oranization affects making of software
(I noted about thirty people in audience)
Started for for IBM in 1979 in San Jose, CA, in what was then known as
STL, (Santa Teresa Laboratory), and is now known as SVL (Silicon Valley
Laboratory). It is one of the larger sites of IBMs Software Group.
Was one of 330 mainframe programmers, doing various major language
Used 2371 (number unclea) DASD [Note from Dave: This dates Eben and me
as this is an old IBMism; it stands for Direct Access Storage
Device, what we now call hard disks)] Several large rooms of this
large disks required to contain a the enormous amount of storage, 29
gigabytes. You can buy a small disk with that capacity today for under
[Note from Dave:To illustrate his point, when I worked at NYU/CIMS in
the 80s, I recall that we bought an external Davong hard disk for an
IBM PC in 1983. It cost $5000, and had a 5 megabyte disk, for a net cost
of about $1000/megabyte. This translates to about $1M/gigabyte, so
Ebens 29 gigabytes would then have cost about thirty million dollars.]
Hardware engineers achieved miracles.
[Note from Dave: The advances in disk technology have outpaced Moores
Law. That is, disks have gotten smaller in size, bigger in capacity, and
cheaper in cost faster than chips have gotten smaller in size , faster
in speed , and cheaper in cost.]
But there has been little improvement in the software development
process,especially in Microsoft software.
Eben then mentioned that he had once worked with Michelle Hack. [Note
from Dave: I had earlier eaten lunch in the Hawthorne cafeteria, and had
seen Michelle, though he was busy and as I was with someone, I hadnt
bothered to say hi. Ive known him for close to twenty years. He is one
of IBMs Masters of the Mainframe.]
There are fundamental limitations with respect to software.
Software engineers are as good as hardware engineers, as smart, as
The problem is that we havent allowed them to get better.
For there has been a structure that:
introduces increased friction
There are additional concerns related to interoperability.
Note from Dave: At this point my notes just say
eloquent for in a few
show sentences he gave a remarkably brilliant summary, but to fully
appreciate it you have to get each word, and each comma right.
Notes page with eloquent tag
Hardware has improved by nine orders of magnitude in the last quarter
century; while software has improved by less than one order of
Difficulty due to Redmond (i.e., Microsoft) but they are not unique,
just a particularly aggravating example of the syndrome.
Microsoft dominates software industry, yet produces no hardware except
keyboards, mice, and a table top supercomputer (by which Eben meant
Apple differentiates on expensive hardware, but not nearly as great
improvement in software.
There have historically been few patents in hardware since patents
expose your current state of knowledge, so that potential competitors
can work around them or separately exploit the paths suggested by the
However, the patent system has become pathological in handing softare
As example, Eben said that in 1984, while working at Cravath (Note from
Dave: one of the leading law firms; I think they helped IBM in both the
Antitrust case of the 60s and 70s, as well as the recent LAffair
SCO), Eben helped (I think on behalf of IBM) in suit against Hitachi
related to their having crossed the line (though he didnt use quite
these words) in gaining access to IBM technology. He said that 20 years
later, after he learned that IBM had sold the disk drive business to
Hitachi, he spoke with one of their people, and one of them remarked,
When someone has been trying to burgle your house for twenty years,
then you sell them the house.
Eben mentioned he worked on VS APL, which had 470KLOCS of assembly code.
[Note from Dave: VS was one of IBMs operating systems; APL is a now
almost-forgotten language that still has a following in Wall Street. I
know some of the old-time APL folks, and asked Eben if he knew Aden
Falkoff. Eben said Aden was his manager for this project.]
Eben then went on at length about how this was back in the days when IBM
freely shared source code, and that his group served as a key part of a
shared community, in which customers sent in patches, there were
meetings between IBM and customers, and so on.
Product developers were clearinghouses, part of community of shared
interests. They got customer patches, attended SHARE meetings. There was
a technical collaboration between customers and manufacturers.
[Note from Dave: I asked him if this was in the pre OCO days and if he
had attended a SHARE meeting. He said Yes, and, Yes. For those who
dont speak IBMese, OCO stands for Object Code Only and its introduction
in the early 80s marked the end of IBMs open sharing of its system
code. SHARE was the IBM user group, and it had frequent conferences. I
didnt work at IBM when OCO came in, but I then shared an office with
Jim Lipkis at NYU. He was quite active in SHARE,and remarked that he
thought this was a bad move by IBM. It is interesting to speculate what
might have happened if IBM had maintained open access to its source
code, including that for the PC; the computing landscape today would
certainly be much different
But that was back in the early 80s.
Today the situation is much different, as Microsofts only asset was its
source code. It was not vertically integrated. Though it was integrated
with respect to software, that was too narrow a base. Thus Microsoft
could not afford to share their code, and due to the dominant position
they managed to achieve, they have created a software industry in their
Microsoft has thus suppressed innovation, so their users are
This is not possible in general, as cannot disempower large companies
such as GM and GE.
The nightmare of any business if to use control of its destiny by having
their destiny in the control of their suppliers.
Companies want to maintain control over their destinies.
For examplek, one of the way computer companies do this is to acquire
patents, for this gives them freedom of action.
Microsoft, which needs to protects its model, has said their model is
the only model that can achieve innovation.
In effect, this is Microsofts Big Lie, as in order to limit
innovation it needs to argue that its model is the only one that allows
The Microsoft Tax on hardware that was needed to protect its profit
margins grew razor-think, thus reducing the air supply, in a way
causing an asphyxiation of innovation.
Microsofts success had a devastating impact on innovation:
added friction in running a business
programing became a discipline that was accessible only by using a
This is opposed to the free software, where:
You start from something that works
Build on that
Eben then started speaking of the gplv3 process. He began to say a key
step was the difference between gplv1 and gplv2.
As I was only familiar with gplv2, I eagerly waited to say what he would
But then someone from the audience asked about SaaS, and he never got
around to this again.
[Note from Dave: SaaS stands for Software As A Service, The software
world is currently divided into two groups: those who know what SaaS is
and believe it is the next Second Coming in the Era of Software, and
those who dont give a damn about it. Most of the people in the first
group work for large software companies such as IBM. I work for IBM, but
am in the latter group.
Though Eben was articulate in his comments, thus demonstrating that he
would be in a formidable foe in a courtroom, I tuned out on this part of
Eben then spoke or SugarCRM, saying it would evolve into three options:
copyleft capitalism, based on free software, which could be used for
Copyleft - by requiring sharing it has created commons in user
Decision between GPL and BSD is how businesses maintain control over
[Note from Dave: idea here is that GPL, by forcing disclosure of any
changes, lets businesses invest in free software knowing that no one can
hijack their contributions it must all be shared.]
IBM is the most forward-looking organization in that its licenses
require disclosing source code.
[Note from Dave: I played a role in drafting IBMs first open-source
license. My charge to the attorneys was to take Apache as a base and
add whatever is needed to keep IBMs patent attorneys happy. I never
requested a license that required that source changes be disclosed, and
noted with surprise that when the license finally was approved this
requirement had been added. Looking back, wiser heads most of them
with legal degrees made the right call.]
Hypothetical: What if FSF is under control of somebody we cant trust?
Eben said he had spent two and half years on the gplv3 process: one year
of planning, followed by sixteen months of open discussion and revision.
He felt that gplv3 addresses the sociological difficulty of the
commons in this decade, so they had a friend for solving the patents
[Note from Dave: I probably got the previous sentence wrong, my notes
are a bit murky here.]
All of the gplv3 revision was done under pressure of ideological
DRM (Digital Rights Management)
There is still significant diversity of priority over ideological and
pragmatic aspects of business.
The presence of enormous [illegible] between interests of ideology and
pragmatists is where Eben labors now.
Question from audience: Will this (gplv3?) raise the quality of
Whatever are the real problems, we can brute force them by creating a
billion more programmers.
Closed code made if hard to become a programmer.
Eben then spoke of some of his experiences working on APL almost thirty
There was open access to the source code, so one of his first projects
was to see if he could build it (it was about half a million lines of
code). Able to leverage someone elses work and learned how to build it
within a day or two. When he showed this work to his manager, they were
so impressed they gave him increased responsibility.
Eben then expanded on the billion programmers motif, saying that now
with free and open-source software there is a process of Brownian
motion for software, which he also called fractional distillation, in
which large crowds of programmers reach a consensus on what works best.
He then said we should, Let this engine run.
He said his estimate was that at most five percent and this was
probably too high of Microsofts developers were capable on building
We now have a system when deeply thoughtful people can observe, analyze,
and synthesize the work of others.
His next point was what happens where innovators are innovating of the
stuff we use. He said that he had given many talks at universities,
often at the request of a faculty member, but that when he showed up to
give his talk, he learned the faculty members were skipping the talk,
yet the room was packed with students eager to hear what he had to say.
He then said that much software innovation is now done on the Linux
kernel, using the gnu toolchain. [Note from Dave: I brought up the
toolchain part, knowing that FSFs greatest contribution to free
software is that toolchain, and one must give credit where credit is
He then said that this innovation is bringing rubber out of the rain
forest, so that innovation arrives as code that is working. It comes to
use as a proof of concept in the form of running code.
Then people hack on that, producing an incentive to further improve it,
starting a cycle that yields a much higher level of innovation.
The question then is how to maintain the sweet spot where people come
whistling to work as they are so excited they cant wait for the day to
[Note from Dave: Many years ago, when I was at CIMS, a fellow student
said that he often noticed that Kurt Friedrichs, then a famed Professor
of Mathematics well into his seventies, waited impatiently each morning
for the elevator to arrive and carry him up to his office, as he was so
eager to start his days work.]
Eben then spoke of Samba, mentioning he had worked with the group since
started as one-man project
evolved into team
Now EU says Samba is the competitor, and that this was the reason for
the interoperability requirements imposed on Microsoft by the EU
Substantial profit flows out of the non-profit supply chain that is
the available free software artifact.
The usual course of a company is to acquire a supplier, thus getting a
Model is that non-profit supply chain is a renewable natural resource,
and so need to preserve the commons.
Three major points re gplv3:
First, gplv3 was not just an attempt to improve the commons, but was
done to address elements that had gone unaddressed in 91-92 when gplv2
was drafted, due to patent issue.
Purpose of gplv3 is within the commons encourage businesses not to
use patents against one another. This was done not to address external
threat. Real goal is to ensure that members of the commons act in a way
that preserves and strengthens patents.
gplv3 is an anti-betrayal agreement.
Second, the anti-defection rules allows downstream protection.
He then speculated that there will be unexpected consequences as a
gplv3-like approach is used in hardware, as Sun has just started to do,
so will have need for new patent protection.
Third, gplv3 has established a trans-national approach to drafting a
license. For example, Creative Commons is moving to use a similar
[Note from Dave: I commented that I had learned from Randy Metcalfe last
October, when he worked for OSSWatch in the UK, that many international
organizations had adopted and encouraged gplv2, even though they knew it
contained language based on the US legal system that didnt apply in the
world at large. This sparked, I think, some of Ebens comments here.]
Eben then started comparing PERL and APL, though I dont know why,
though the net seems to be that PERL supports lots of ways of writing a
program. [Note from Dave: I held my tongue here. Having written programs
in both PERL and APL, I know they are similar in that it is virtually
impossible to read code in these language that was written by someone
Copyleft Capitalism is the 21st century apotheosis of Yankee Ingenuity
(here I think APL was given as an example of a toolkit).
He mentioned that two hundred years ago Thomas Jefferson had predicted
it would take a hundred years for the United States to cross the
Mississippi, yet it had happened much sooner, due to the American
[Note from Dave: I have read Stephen Ambroses magnificent recounting of
the Lewis and Clark expedition. Among the more telling parts is that
Jefferson provided personal instruction to Lewis on science in the White
House. Wouldnt you have loved to sit in on one of those lessons?]
Eben said the arrival of the web in parts of the world where the best
brains are waiting will enable those brains to come online.
There is a pursuit of the commons in software that will provide access
to technology, and access to knowledge, that will itself be used to
further expand the reach.
The talk then ended, just about ninety minutes after he had started, and
Eben opened the floor for questions.
I was greatly pleased to recognize the voice of the first questioner,
for it was the same Michelle Hack whom Eben had mentioned had been a
colleague almost thirty years earlier! I think the question was about
Eben then spoke of Microsoft:
Microsoft failed in that, as a monopoly, it has for the first time
failed to produce a new version that will get the old version out of the
Vista has doomed them, as Microsoft is now competing with itself, trying
to displace Windows with Vista.
He then said the real problem was to limit the damage when Microsoft
fails, how to get out from under Microsoft as they enter the end game,
perhaps by paroxisms of litigation. [Note to Dave: take a lesson on
spelling paroxisisysm, or whatever the damn hell it is, in the same way
I had to learn to spell indemnification over three and half years
He said Microsofts model was to get people to spend more, to get less.
That while some counties had tried to sell themselves to Microsoft,
taking 10 percent as a tax, it was now impossible to do this. [Note from
Dave: I probably got this wrong, but it was a good metaphor.]
He then predicted that the day would soon be upon us when people would
be fired for buying Microsoft, and gave as example the aftermath of the
2000 dot-com implosion. CIOs asked that expenses be cut, so many folks
looked out over their server farms filled with Sun boxes. They then
replaced those Sun boxes with white (ie, commodity) Intel boxes running
Linux, after which they went back to management and reported they had
cut costs fifteen percent.
Eben then related how he had finessed the famed coupons in the
Microsoft/Novell brouhaha, describing how he had hacked Novell.
[Note from Dave: He was now in full stride, much as was Richard Feynman
when I heard him tell of his days cracking safes in Los Alamos.]
But Ill let others relate that wonderful tale.
Perhaps you can invite Eben to come talk to your company, or your
university, to hear him speak.
[Note from Dave: Indeed, the Eben Moglen Coupon will undoubtedly
become more famous than the Jikes Coupon. We shall see.]
I left at this point, though the discussion continue.
END OF NOTES
I had asked Eben before he began if this talk was available in written
form. He said no, so I took the notes that I have used as the basis for
recounting his talk.
I understand the talk was recorded and that it will be made available
along with a transcript, so those unable to attend the talk are in for a
great treat when that record becomes available.
Not only was the talk great fun to hear, it also had an impact, at least
Though I have voiced objections to gplv3 in many posts in this blog, I
now take them back. The gplv3 process was much more purposeful and
nuanced than I had appreciated, and I now place full trust in any effort
in which Eben has played or will play a role.
I started the post in the afternoon and spent most of the evening
working on it, but I did find time to put our pumpkin in place, so the
commuters coming home would see a welcome sight. Perhaps this will
become to photos what ESR did for memos:
Note poodle Scout as well as Ken, our Rodent of Unusual Size, a reminder
that I still prefer the Apache License to gplv2 or gplv3.
This entry was written by daveshields, posted on October 29, 2007 at
23:59, filed under open-source. Bookmark the permalink. Follow any
comments here with the RSS feed for this post. Post a comment or leave a
trackback: Trackback URL.
Alexander Terekhov wrote:
> Eben Moglen Whacks Tim OReilly!
> Posted 24 July, 2007 by Chris in Open Source, GPL
> I was at the OSCON Execuitive Breifing today and was looking forward to
> hearing Eben Moglen speak about Open Source licensing issues. The GPLv3
> was recently released and a draft of the Affero license is open for
> comments. I think the GPLv3 is a good but flawed license, and the Affero
> license in its current form is a complete joke.
> It seemed like rich fodder for an interesting discussion.
> I really like Eben. Ive heard him speak several times and hes often
> the smartest man in the room.
> Today, however, I think it must have been Ebens doppleganger because he
> whacked Tim OReilly. In front of more than 100 people!
> It was embarrassing. It started the minute he sat down.
> At first I thought it was simply old friends giving each other a hard
> time but soon it was clear that Eben had an agenda and was using the
> venue to make his point. I dont have any problem with a speaker having
> a point of view and arguing it vigorously, but when it turns personal
> and vitriolic theyve clearly cross the line.
> I think this was clear to the audience as well. How else could you
> intrepret statements like:
> You spend to much time with your billionaire friends
> Youve wasted the past 10 years trying to make money while freedoms are
> under attack
> This Web 2.0 stuff is silly, thermal noise
> And lots of other zingers that I cant remember (comment or send me an
> email, please). They were said not in a casual, ironic sort of way, but
> in a thoughtful, pointed downright mean spirited way.
> I felt uncomfortable just listening to the conversation and it was clear
> that many in the room felt the same way. Tim was a gracious host and
> gave his invited guest his full attention and asked how he could help
> address the issues, at which time Eben replied Take down your name (the
> large OReilly sign behind the stage) and start promoting freedoms.
> Tim was speechless (or so it seemed).
> What bothered me most about Eben diatribe was the outrageous duplicity
> expressed in his rhetoric and the actions of the FSF regarding the ASP
> loophole in the GPLv3. I even asked him about this. I told him his
> rhetoric seems hollow given how the ASP loophole could have been a big
> step forward in addressing some of the concerns he has.
> I know better than to argue with Eben about any of this, but his
> response seem myopic and self serving. He basically said that closing
> the loophole would infringe on certain peoples rights and he didnt see
> any way to preserve everyones rights, so they left that problem for
> another day.
> Which is a total cop out.
> Rights are in conflict everywhere we look, but that doesnt mean we
> dont have laws that favor some rights over others. Look at our tax
> code! Heck, thats what our judicial system is for; to resolve these
> kinds of conflicts. The GPL is a friggin license, not a policy
> statement. They could put whatever they wanted in there. If Ebens fears
> loom as large as he says they do, he should have acted upon them using
> the single biggest tool he had at his disposal, but no, he chose not to.
> Rather, he chose to ambush Tim OReilly.
> In his later posting Chris elaborated:
> (More on Eben and Open Source Licenses)
> Lost in the discussion (here <http://blog.snaplogic.org/?p=65>, here
> here <http://www.linux.com/feature/118201> and here
> <http://www.theregister.co.uk/2007/07/26/oreilly_moglen_oscon/>) on Eben
> Moglens session with Tim OReilly on Licensing in the Web 2.0 Era was
> what was actually discussed.
> Tim believes that since the world is rapidly headed toward a more
> centralized computing model with power and control maintained by the
> on-line services like Google, eBay, Amazon, Salesforce, Facebook,
> MySpace, Flickr, etc. (i.e. Web 2.0), Open Source licenses dont matter
> much anymore.
> Eben, on the other hand believes that the pendulum is swinging the other
> way toward individuals controlling more and more of their own on-line
> activity and therefore licenses matter very much.
> He further believes the consolidation we see happening today is merely
> an aberration that will correct itself once we all realize the control
> these services hold is dangerous and threatening to our freedoms and we
> systematically reject them.
> Fundamental to Ebens argument is that we have the power to reject them
> 1) Todays laptop computers have capabilities of the most powerful
> supercomputer of 20 years ago and,
> 2) Free software will provide the means by which this will be possible.
> Therefore, since the compute problem is being pushed out to the
> endpoints (i.e. users), and not gravitating toward the center (i.e.
> services), licenses do matter. Matter a lot.
> Implicit in this argument, I believe, is that since we have the power,
> we also have the desire and skill and will also take action. I say
> implicit because we never really did get down to this level of detail in
> the session.
> Hes more eloquent and convincing on these points than my simple
> description, but independent of these other factors, Moores law
> trajectory is one of the major tenets of his argument.
> I am not convinced.
> I strongly believe that in the future (1, 3, 5, 10 yrs??) there will be
> more centralization of services/data/info, not less. The trend we are
> seeing today is not an aberration, but a secular shift in computing.
> None of Ebens arguments sway me on this. In particular, his Moores Law
> argument is completely specious. The premise that the growth in compute
> supply (i.e. CPU performance) shifts the equilibrium toward the edge
> would be more convincing if there was some mention of the compute
> demand. Unfortunately, I have yet to hear Eben mention even once the
> effect of compute demand.
> Empirical and anecdotal evidence suggests that demand has outstripped
> supply and has indeed shifted the equilibrium, but toward a more
> centralized compute model, undermining a major element of Ebens thesis.
> Yes, my laptop is a 2GHz, 2GB dual core device thats 1,000 times more
> powerful than early super computers, but today Ive also got a Terra
> Byte of data that Ive got to manage. Thats 1,000,000 times the data
> that I had to manage on my computer from 20 years ago. Im getting
> buried under three orders of magnitude more data.
> Advances in software might make it feel like only two orders of
> magnitude, but it still hurts. A lot.
> "Mathematics is primarily a language for ensuring reliable results
> in human social activity. "
> -- Columbia Professor Eben Anarcho-Dot Communist Moglen
Re: GPLv3 comedy unfolding -- Eben: "Copyleft Capitalism: GPLv3 and the Future of Software Innovation", Alexander Terekhov, 2007/11/17
- Re: GPLv3 comedy unfolding -- Eben: "Copyleft Capitalism: GPLv3 and the Future of Software Innovation",
Alexander Terekhov <=