Subject: New ESR paper: The Magic Cauldron
From: "Stephen J. Turnbull" <>
Date: Fri, 25 Jun 1999 11:28:36 +0900 (JST)

>>>>> "Ian" == Ian Lance Taylor <> writes:

    Ian> ESR has a new paper out: The Magic Cauldron:

Disappointing; his talk at ACM-Japan was better on ideas, although
this is more detailed.

    Ian> The Magic Cauldron seems less interesting to me.

Eric is not an economist and The Magic Cauldron isn't really about
economics (except for the chapter about "business ecology"), it's
about business models.  And nobody really knows what makes one work,
or if it will.  Trying to be specific makes things harder for the
writer.  This is probably a consequence of becoming more pragmatic: if
you don't see your business there, you won't be as interested.

    Ian> In his list of free software models, he doesn't mention the
    Ian> insurance model.

I'm not sure what you mean by that; "insurance" is implicit in
"risk-spreading" which he discusses in the Cisco case.  If you mean
"selling insurance", that's either support or risk-spreading in the
Cisco sense; I don't see what else you mean?

    Ian> As a canonical example of an application, he says a word
    Ian> processor, but in the real world I think Microsoft Word has
    Ian> just about become infrastructure.

My bet with the Abisoft guys is still open ;-)  What Microsoft Word did
to WordPerfect can be done to Microsoft Word.  It's harder---you don't 
need to rehearse the reasons---but can be done.  Don't think Bill
Gates doesn't spend time thinking about it.  (He probably doesn't lose 
sleep, though :-(.)

    Ian> He spends time on the tragedy of the commons, but I can't see
    Ian> why anybody would think that would apply to free software.
    Ian> Do people really make that objection?

Eric (or his consulting economist) is being sloppy.  He means the free
rider problem, that's clear from other parts of the paper.  The
Tragedy of the Commons is a much more evocative phrase.  Eric admits
to being a showman; I think this is pushing it too far, though.

Evidently you [Ian] know the difference, but I think it's worth making 
the distinction precisely.

Technically, the Tragedy of the Commons occurs when a congestible
public good[1] is made into a commons, ie, unrestricted access to
anyone (more generally, to too many people---this is what happens when
a bridge toll is set too low, for example).

Information, on the other hand, is a pure public good, with no
characteristic of rivalry.[2]

The free rider problem is a coordination problem: it is difficult to
coordinate payment from external beneficiaries of a project, since
they are not party to the decision, and may not even be identifiable.
If it is possible to self-select (ie, lurk, and hope that somebody
else with build the project), then you may end up with the tragic
consequence that everybody lurks although the aggregate benefit is
many times the cost of the project.

Thus, the "tragedy of the commons" requires congestion, which is not a
feature of information in general, and the cost allocation (if
predetermined) is irrelevant.  The free rider problem is a fixed
(overhead) cost allocation problem, and occurs whether or not the
public good is congestible.

The "tragedy of the commons" is due to overuse; the free rider
problem, dually, is about underprovision.

To answer your question, no, as far as I've seen nobody worries about
the tragedy of the commons.  People who want marginal (or average)
cost pricing of software (Brian Bartholomew) should worry about it,
however.  Without demand-driven "all-the-traffic-will-bear" pricing[3] to
indicate which projects are important to users, the group of
charitable hackers who donate code becomes a common pool resource with
no sensible criterion (except "hack value") for allocating their
effort to different projects.  This is congestion, in a sense.

BTW, the point about instrumental value is important, and has already
been discussed in a thread here.  It is the abstract version of the
idea that Cygnus can charge $xxxx for a product which is free
software[4] and still the customer doesn't redistribute, because it
would be subsidizing the customer's rivals.

    Ian> He adopts some ideas I first saw clearly expressed in Neal
    Ian> Stephenson's In the Beginning paper about software following
    Ian> a natural evolution from application to infrastructure.


[1]  My consumption of the good is not prevented by yours, but our
consumptions interfere with each other to some degree.

[2]  None in itself; Eric's point that the value of the information
may be instrumental to some other activity, and that other activity is
rival is correct, but not relevant here.  We are talking about
physical (or hedonic) characteristics, not economic characteristics

[3]  The use of prices to gather this information can be avoided by
use of user polls and the like.  However, such polls tend to be
extremely unreliable.  And of course people will "do necessary jobs
for a project" when solicited by acquaintances, but that begs the
question of why the project is important.

[4]  "You have the right to modify.  You have the right to
redistribute."  <Detective DJ handcuffs customer to FTP server>

University of Tsukuba                Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences       Tel/fax: +81 (298) 53-5091
What are those two straight lines for?  "Free software rules."