[ox] Protocol of meeting at WOS
- From: Bettina Berendt <Bettina.Berendt educat.hu-berlin.de>
- Date: Tue, 3 Aug 1999 10:02:59 +0200 (MET DST)
here it is ... finally ... the protocol of our "inaugurating" meeting.
First, my attempt at summarising the issues raised in the discussion,
second, Holger's transcript of much of the discussion.
My apologies for this mail being rather late, and also for a
not-very-pretty layout. "My job is eating me", to brutally translate
a German complaint.
Best wishes -
PS: Replies to the questions marked as "afterthoughts" are more than
Bettina Berendt berendt educat.hu-berlin.de
Humboldt-Universitaet zu Berlin http://www.educat.hu-berlin.de/~berendt/
Abteilung Paedagogik und Informatik Fax [PHONE NUMBER REMOVED]
What do Open Source Providers get in exchange for the release
of Open Source? /
What kind of value is generated in the Open Source context?
(I say "Open Source Providers" because programmers are only one group.
For example, there are "Open Source text writers" like Medosch and Ghosh.)
"Value flow as a superset of exchange" (Ghosh) -> so what value flows?
Comments are the Currency
- - One reason for releasing software/opinions for free is to get comments.
For example, in mailing lists a small percentage of the participants
writes a large percentage of postings. Most of the remaining postings are
comments made by the other participants. These comments are valuable
feedback/input for the (more obviously) active participants.
(Ghosh mentions the Perl-porters newsgroup as a concrete example.)
"Economy of Attention"
- - Problem: this is not working very well (the "star system" - a handful
of people like Torvalds and Stallman get all the attention, and the money
for interviews, talks, etc., that goes with it; even if the proportion of
Linux written by Torvalds is now below 10%)
[? Raymond's paper also says something about the economy of reputation and
its problems ? I'm not sure that I interpret my notes correctly here. ?]
- - This is not too surprising, given that attention
(and human memory performance) are scarce resources.
- - Knowledge-based methods may help, but will not solve the problem.
Ghosh mentions a program that he is co-authoring that
filters newsgroup postings
by the "personal evaluation of the author of the posting".
This program operates on the principle that a positive evaluation of a
person and her/his opinion is transitive: If I have marked some person A's
name as "a person whose opinions are interesting/relevant to me",
and that person A has done the same
for a third person B, then I want my filtering program to
select A's as well as B's postings (Ghosh: "networks of reputation").
Open Source and Advertisement
- - Medosch, Ghosh report that texts/articles published as "Open Source" in the
Internet brings money to the author in a number of ways:
- publishers' attention is raised; they ask for further texts to be
published in "ordinary" ways (e.g., as a book which they publish)
- publishers re-publish texts, e.g., on mailing lists, and
pay the author for the
right to re-publish. The publishers recover
these costs by subscription fees from the readers. The value added (which
the end-readers pay for) is the service of filtering relevant information
from the Internet.
- Indirect monetary gains through the gain in prestige and attention.
But text authors are protected by copyright laws.
[*Afterthought*: Q: What is the current situation with software copyrights?
Can someone be taken to court because of infringement of a GPL licence?
- Holger Blasum thinks there are actually some precedents, but cannot
quote them. -
And who is going to pay the lawyer? ...]
Value created by filtering as a service
- - (cf. mailing lists)
Aspects of Economic models of Open Source
- - In contrast to the Fordian model, distributing software has very low
[? I found that in my notes, but I am not so sure any more that I got this
right. As far as I remember, production costs per item also fall in the
"standard" Economic models. I think the remark referred to Ghosh's emphasis
on the near-zero distribution costs of *every* copy, which does indeed make
a difference to the Fordian model.
- Remark by Holger Blasum:
I agree. Ghost stated the value of making the first copy is 100%, of all
other copies the value is 0% (if somebody else mirrors the site). I
he really spoke of value not of costs, but I do not quite understand
- - The programmer cannot recover the cost of the first item;
selling the last item would have higher transaction costs
than the selling price.
[? Again, I am not too sure how to integrate this remark.
Anyway, it seemed
interesting enough to include it in this protocol. ?]
Street performer protocol (Ghosh)
- - cf. http://www.firstmonday.dk/issues/issue4_6/kelsey/index.html
(article by John Kelsey and Bruce Schneider on Street Performance Protocol
and digital copyrights)
- - production costs do not rise with number of listeners
(=~= number of copies sold). This differs from, e.g., a musician playing
in a hall: A hall for an audience of 500 is more expensive than a hall
for an audience of 50.
- - therefore producer can afford to demand only a fixed amount of money
(e.g., the production costs), need not care about whether it comes from
one or a thousand customers/copies sold, and need not care from whom it
- - An aside: The OS programmer needs to think about the "audience".
Hiller mentions the case of someone who wrote an AppleII word processor,
proclaimed he would sell 10,000 copies and then donate it for free.
He did that, quit development after selling the first free copy,
and the processor died.
- - But protocols need to be established to remunerate the street performers,
those outside the star system. Maybe buttons on the pages "Click here
to send me a digital dollar"?!
[*Afterthought*: An example for this could be Stefan Münz's excellent
SELFHTML teach-yourself-HTML manual. This is available
free online, but also comes as a book, and the author encourages donations
to a particular bank account. (http://www.teamone.de/selfhtml) However,
I don't know how much money he has already made with this system.]
Open Source software as a byproduct of commercial activities
This seemed to be one of the results of the discussion:
Many software writers release pieces of software as Open Source which they
developed in the course of developing a customised product, and which is
likely to be useful as a tool outside the context of this customised,
sold product. In development, they also use a lot of Open Source (and free)
They make their living as independent programmers, working
in a company, or as a consultant. Companies realise that if every single
utilised tool was to be paid, the costs would be astronomical.
This leads them to allow some of the value created to flow back into the
free/open Source market.
- -Ghosh's "cooking pot model": everybody is putting
in value and getting out value, and the sum (= the contents of the cooking
pot) has more value than the sum of its parts (= the individual pieces of
- - Some free riders are not a problem: end users who only
download Linux but don't
produce any OS software themselves do not increase the
costs of writing.
- - But: free riders may still be a danger, cf.
"The person who makes the most money out of OS is Bill Gates."
(because he uses the growth of the Internet) [Tim O'Reilly]
What is released as Open Source?
Differential "openness" as policy of software writers
- - may release some developments for free (e.g. bug fixes,
in the hope that they will be included
in the next official version of the product), may keep some
to him/herself (to maintain competitive edge)
The future of Open Source
Problems of a growing community
- - The Open Source system worked with small, closely knit
groups. There may
be many problems ahead as the community is growing (cf. also Jeannette
Hoffmann's talk on Open Standards and their development).
The differences between the workings of small, closely knit groups and
large, only loosely coupled groups are well-known in Economics. Peer pressure
etc. are much stronger in small groups, and the free rider problem
becomes more severe with the size of the group.
The conflicts are still ahead
- - People are only starting to make money with Linux now
- - Some individual people / firms make a lot of money from
the current bubble / over-valuation: e.g., Marc Andreessen, Yahoo!, RedHat
[? Can someone supply more details here? I only noted it
and don't know anything else about this story. ?]
- - What will happen when RedHat goes public?
- - Apple entered OS solely for marketing reasons.
[? Can someone supply details here? I only noted it and
don't know anything else about this story. ?]
original notes by Holger Blasum (please correct if you feel
Only the content has been recorded and shortened, the wording is not
B: the guy with a black T shirt, (male, black hair, in seated front of
R: the guy with the red T shirt
?: to be identified
---: here the discussion thread appears explicitly "broken" to me (taking
a too fast turn)
Ghosh: In the perl-porters newsgroup I observed about 600 people. About
10% of these contribute 60% of the postings. The others appear to be
mostly hearing. But I found out that they are appreciated because the
comments and advice prove to be a valuable guiding and correction for the
more active community. In these groups, there are also no freeriders,
because the content is far too technical.
Mertens: They are not doing this for fun.
Ghosh: Of course they are not doing this for fun.
?: They are not doing this for getting an exchange.
Ghosh: I disagree. This is generally done for an exchange (in terms of
value). It not so important *who* gives that value.
Redclay: Comments are the general currency.
B: Apparently we can see that the lists are working !
Ghosh: So what we have to find out why they are working. We have to find
how they have evolved.
Berendt: In the sciences, what counts for the scientists is their
citation index. A good reputation ensures a good income. Of course,
science is subsidized by the government.
B: The majority of papers is not useful.
R: Nor is the majority of software. But in the long run scientific paers
are meta-reviewed and so are newsgroup postings when they are forwarded.
Forwarding a newsgroup is analogous to a scientific citation.
?: There is not always an exchange.
Ghosh: Actually this doesn't matter. Value flow is a superset of exchange
(not only the exchange between two parties). It is a kind of
B: So we are back to your cooking pot model - everybody is putting in
value and getting out value.
?: The person making most money of this is Bill Gates.
Redclay: Every transaction incurs transaction costs. The value of the
software is it's reproducability.
Ghosh: However the creation cost is the same for 1 copy of 1,000 copies.
This is different from other economic goods.
Hiller: This reminds me of the guy who wrote an apple II word processor,
proclaimed he would sell 10,000 copies and then donate it for free, who
actually sold his 10,000 copies and then quitted development, so that the
Ghosh: In the June issue of First Monday an article on the Street
Performance Protocol and digital copyrights
(http://www.firstmonday.dk/issues/issue4_6/kelsey/index.html) by John
Kelsey and Bruce Schneier.
We have different kinds of economics of giving something away: when in
musics a performer plays in an auditory this is more expensive for him
when he does this for 500 people than for 50 people (he will have two
rent a larger auditory and do more organization). On the other hand when
a street player performs in London subway and all he needs is 50 pounds a
day, it does not matter for him how he earns this (whether it's one
person or 1,000
B: How to deal with people who do not write programs.
?: On the internet a lot of customers download without being able to give
input, although they make downloads and copies.
Ghosh: Note that this does not increase the absolute costs of writing,
just the marginal costs.
(ISPs not being counted). This is not new economics, copies of linux are
Medosch: I often author to mailing lists, and as a writer this pays back
directly for me, bacause I often get asked to publish articles on ideas
expressed in newsgroups. For example for an article not yet previously
posted on a newslist that US media analyst pays me $200/page, for an
article previously published in a newsgroup I get 150$.
B: Of course, this due to the fact you hold the copyright on the articles
even when they are posted onto the mailing lists.
R: So we have an economy of the scarce resource reputation. However it
might be distorted by imperfect flow of information.
Ghosh: Information flow is distorted in every economic model, it does not
deprecate the economic model iteself. One of the project we are currently
working on is automatically generating networks of reputation. If I
esteem you and you esteem somebody else this reputation might be
shareable. Thus generated reputation network may help to circumvent the
Berendt: I doubt this undertaking this is feasible. Furthermore, many
people are tied by other things than economic gain in reputation, e.g.
familiy ties, for freedom ..
Redclay: nor can I buy bread with one pound of reputation...
Ghosh: in the case of very famous persons the model to nourish musicians
by donations even might directly work, of course strife for reputation
may be mixed with other incentives,
Medosch: as Professor Dr. Norbert Szyperski (who is absent) states
"everything is economic" ...
Ghosh: That is not true. We are using economic models, and ideology is
not observable by them.
R: I think the current OS bubble at Wall Street is a danger. Yahoo! and
RedHat being so much overvalued.
Ghosh: Well if the bubble crashes we will have realistic evaluations, so
this is no real probelm, the value of RedHat and Yahoo! won't change by
B: Some people, like Marc Andreessen took much money out it.
Ghosh: Which is not more immoral than winning in the lottery.
Berendt: The community of users is growing...
B: Some people are getting frustrated like the and go commercialize like
the founder of ssendmail...
Ghosh: But we do not have a real problem as far as we get enough
self-exploitng people who program for reputation and are possible then
B: I only release patches for bug fixes if that has a direct economic
benefit, e.g. when I want to see them included in the next distribution
of the program together with other useful features other poeple have
brought in. I do not release the patch when I see a competitive advantage
over my competitors in not doing so.
Ghosh: .... But still open source seems to work even if not all patches
are given in. So why does it work. Economically this is explainable by
reputation which translates into value. In the economic model, we cannot
include other motivation, e.g. ideology or lover
Remember: in my talk this afternoon I stated the internet is either 1)
run be a new type of humans more altruistic than normal non-net people
and thus not explicable
by economics at all. 2) Other motivations, such as everybody is doing
this in their spare-time to laern and play with the intrnet, but in the
long run this incentive will become unstable.
3) Human netizens do not differ from other human beings.
Except the few stars of OS programming, there is a large medium level of
e.g. company employees who use a lot of OS programs (company benefits)
and then doing the projcts adds value to the tools which is returned to
the public (company looses).
The same pattern applies to individual or on-project programmers.