[p2p-research] Open source is a company; social media is a country

[Paul D. Fernhout]

 From being around some open source projects, and having read a lot of the 
emails about various other ones, I think there is a lot of variety.

I read somewhere Linus Torvalds had a lot of success with the Linux kernel 
because, while he wrote so-so implementations, he wrote great interfaces, 
which let other programers improve the implementations; and his personality 
style and humor fit in well with a certain culture.

The Apache project, with maybe twenty core contributors early on, who were 
heavily systems administrators, is a different story in collaboration.

See, for example:
"The Myth of the Benevolent Dictator"
http://it.toolbox.com/blogs/database-soup/the-myth-of-the-benevolent-dictator-18668
"Of course, like other simplistic comparisons it's nonsense. If you look at 
the most popular and well-funded OSS projects, they run the entire gamut of 
leadership structures: Apache was started by a triumvirate and now has a 
non-profit beaurocracy; Eclipse is a trade council; Debian is raw democracy 
bordering on anarchy; PostgreSQL is democratic; MySQL is a company with a 
half-dozen executives; Java is a huge mess of different organizations and 
companies; and Perl is nominally a dictatorship but I defy anyone to name 
the last time Larry Wall gave someone an order."

Also, projects can change in needs. Guido van Rossum, a brilliant guy and 
"benevolent dictator for life" for the Python project filled that role very 
well in the early years. (Although, remember, he was taking ideas from the 
academic ABC project he had been a programmer on and putting them in a new 
context, so he had stuff to draw from that may have been designed 
differently.) As Python has grown and diversified, his status as "bevelonent 
dictator" has grown more problematical.

Still, Python has done well because Guido made aspects of it modular. In 
general, software that is designed to be modular has less social issues. :-) 
There is less reason for there to be central bottlenecks and core arguments 
if software is modular.

Squeak is a good anti-pattern in that regard. :-( It has had endless 
problems and strife due to lack of modularity, because that put a lot of 
pressure on a small group of people who started it at Apple and then went to 
Disney. But that core group had other priorities so they neglected 
addressing that modularity issue and related ones, thinking the community 
might magically fix it, when they were not making a virtue out of lack of 
modularity. Squeak also had licensing issues that prevented fixing the 
modularity (both the original license itself not being FOSS except in 
spirit, and the lack of clarity of the licensing of contributions). If you 
started to fix the modularity or other issues, fixes got left behind by the 
core group because it was not a priority. But forking was problematical 
because you would be stuck with a non-standard license at the core, and so 
was it worth making the investment to fix the problems? The two issues 
(modularity and license) were interwoven (because you could not really 
easily fix one without the other), but people in the community did not seem 
to understand this.

Both were issues I wrote about at length many years ago, with some huge 
disagreements from some people in the community including one lawyer on the 
list who endlessly defended the originally Squeak license. One example of my 
comments from 2001:
   http://aspn.activestate.com/ASPN/Mail/Message/squeak/522309

An example of intimidating lawyer speak by a lawyer that, in the end, proved 
wrong as far as the effects on the community:
   http://aspn.activestate.com/ASPN/Mail/Message/squeak/523288
"""
I'll note that as between Paul and myself: one of us is a well-regarded,
published computer lawyer, and the other is not.  Disney, and their
lawyers seemed unafraid.  Apple and their lawyers seemed unafraid.
Exobox and their lawyers seemed unafraid.  Gentle readers are invited to
decide for themselves.  Clearly, each individual should consult their
own attorney whom they have hired to represent their interests.
   But the kind of comfort Paul seeks in an open source project is
unavailable anywhere.   There are huge legal risks inherent risks in ALL
open source development projects -- anything less than a dedication to
the public domain risks a unilateral revocation of each and every
contribution.  Paul seeks a degree of comfort at odds with every open
source ethic, and not likely to be found anywere.  Apple just released
(indeed, bet the farm) by committing to an OS relying on the
reasonableness of a vast amount of BSD and open source software and
their licenses, none of which meets the standards Paul insists are
necessary.
   The legal questions ARE interesting and worthy of discussion, however,
just as are the issues of modularity.  None of these issues require, let
alone suggest the virtues of, a fork, however.  None of them.
"""

What he did not know, and I did not want to make a point of then, was that I 
had had IBM's lawyers review the license, so I was speaking in part from 
that knowledge as well as my growing knowledge of software communities. And 
the result was that Squeak missed a chance to be something much greater, 
especially in the face of competition from Java. (Squeak still was pretty 
nifty, anyway.)

But those two issues (modularity and license) are now, almost a decade 
later, finally being addressed in the pharo fork. :-)
   http://www.pharo-project.org/home
"Pharo's goal is to deliver a clean, innovative, free open-source Smalltalk 
environment. By providing a stable and small core system, excellent 
developer tools, and maintained releases, Pharo is an attractive platform to 
build and deploy mission critical Smalltalk applications. Pharo is MIT 
licensed and is steered by a board of benevolent dictators. The board makes 
final decisions if no consensus can be reached within the community. Pharo 
fosters a healthy ecosystem of both private and commercial contributors who 
advance and maintain the core system and its external packages. ...
The license of Pharo is MIT. All contributors are required to sign our 
license agreement."

That was done more or less as I suggested back in 2001 (and a few times 
later), finally with a significant group of core people pushing it along 
(not me though, even though that is what I wanted to do back then). I'd 
moved onto other things myself (Jython and Java, mostly) after years of 
agita from all that (though I was trying out the latest verion of pharo the 
other day, and it was pretty nice).

Though, even then, it's not clear to me the actual status of contributions 
done while the core group was at Disney (any other company than Disney and 
it would be easier to take some statements about that on faith). Even Python 
had a problem with that at one point for work done at a not-profit that 
later claimed Guido had no rights to release his work; it got patched over 
but that's one reason the Python license is so long with various sections. 
Related:
http://python-history.blogspot.com/2009/01/personal-history-part-2-cnri-and-beyond.html
"In anticipation of the transition to BeOpen.com, a difficult question was 
the future ownership of Python. CNRI insisted on changing the license and 
requested that we release Python 1.6 with this new license. The old license 
used while I was still at CWI had been a version of the MIT license. The 
releases previously made at CNRI used a slightly modified version of that 
license, with basically one sentence added where CNRI disclaimed most 
responsibilities. The 1.6 license however was a long wordy piece of 
lawyerese crafted by CNRI's lawyers."

Even now, years later, I wonder if Disney can spring a similar things. 
(Publicity issues might hurt Disney too much now though.) A license is like 
a foundation for cooperation. If the foundation is problematical, it's hard 
to know if the rest of a structure is worth fixing. The nice thing about the 
GPL is that it provided a straightforward foundation for collaboration, 
essentially, it became like a de-factor peer-to-peer community constitution.

Anyway, that's all part of why I was so picky about licenses. Thankfully, 
these issues seem better understood a decade later, and the large culture 
has changed, so people seem to be releasing code under standard licenses 
more and more these days (even as there remain issues in people being clear 
about what they are doing sometimes). That is one reason Squeak eventually 
changed; the rest of the world (e.g. Debian) rejected it in part due to a 
quirky license.

So, even in software, one can see how infrastructure and social dynamics 
(including licensing) affect one another in positive and negative ways.

--Paul Fernhout
http://www.pdfernhout.net/


Michel Bauwens wrote:
> My take on this is that free software communities are
> emergent/self-organized and that leadership is crucial, both are true ... I
> don't think that it is true at all that open source leaders can appoint
> people to do specific tasks ... they are dependent on voluntary
> contributions (unless in the case of people paid for corporate open source)
> ...; the leadership is important for the nurturing, for the vision, for many
> aspects, but the real aspects of hierarchy are much more a posteriori
> through arbitrage, than a priori through command and control ...
> 
> Benevolent dictatorship seems to me largely a myth to protect an old
> authoritarian paradigm, just as market thinkers want to call it a market,
> even though nothing is bought and sold at a price ... I would rather analyse
> it as some kind of constitutional monarchs that have obtain a role of
> arbitrage through the accumulated respect they've generated ...
> 
> Any people from real free software projects who disagree with this and
> indeed think that open source projects are dictatorships, albeit benevolent
> ones?
> 
> Michel
> 
> On Mon, Sep 28, 2009 at 6:25 AM, magius <gmagius at gmail.com> wrote:
> 
>> by Chis Anderson (Wired)
>>
>> image At the Sourceforge breakfast this morning we got some questions
>> on what the organizational differences are between open source and
>> social media. Here’s my answer:
>>
>> One of the paradoxes of early 20th Century management was the
>> observation that companies are best run as dictatorships, while
>> countries are best run as democracies. Why was this? Management
>> theorist Charles Barnard, in his theory of the firm, proposed that it
>> was because organizations existed for a common “shared purpose”.
>> Countries, on the other hand, existed only to serve their people.
>>
>> Shared purpose required singular vision, leadership and top-down
>> control. Serving the people, on the other hand, benefits from
>> bottoms-up recognition of needs and collective decision-making
>> (voting).
>>
>> Many people mistakenly think that open source projects are emergent,
>> self-organized and democratic. The truth is just the opposite: most
>> are run by a benevolent dictator or two. What makes successful open
>> source projects is leadership, plain and simple. One or two people
>> articulate a vision, start building towards it and bring others on
>> board with specific tasks and permissions. The best projects are the
>> ones with the best leaders.
>>
>> Social media, on the other hands, doesn’t exist for a shared purpose.
>> It exists to serve the individual. We don’t tweet to built Twitter, we
>> tweet to suit ourselves. We blog because we can, not because we have
>> signed on to a blogging project.
>>
>> Seen this way, open source projects are like companies. Social media
>> is like a country. Benevolent dictatorships rule the first; democracy
>> the second.
>>
>> The point: the nature of participation is very different between open
>> source and social media, even though people tend to lump them together
>> into "peer production". Open source is hierarchical by design, while
>> social media structure is simply ruled by popularity.