Reiser Says MS Settlement Reflects Deep Failure To Understand Implications of "Patching" Technology
Jan 30th, 16:43 UTC
Hans Reiser, author of ReiserFS, submitted his comments on the proposed settlement today, and encourages all readers to add their comments as well by emailing microsoft.atr@usdoj.gov.
Hans Reiser, author of ReiserFS, submitted his comments on the proposed settlement today, and encourages all readers to add their comments as well by emailing microsoft.atr@usdoj.gov. His words follow:
The positions of the DOJ, the States, and even Lawrence Lessig are
based on a failure to understand that something unique to the software
industry, which programmers call "patching" technology, makes software
products infinitely separable if an essential facility called "source
code" is provided. No disclosure of APIs, and no structuring of APIs,
can accomodate all potential products in the manner that disclosure of
source code plus use of patching does. Every line of source code is a
possible location for insertion of new code that forms a new product.
This new code can be distributed separately from the original source
code, and post-sale added by the consumer, via what programmers call a
"patch". Patching technology fundamentally changes product
separability, making separation dependent on the essential facility
called "source code". Non-programmers seem to not yet understand
this. Persons who work in the Linux industry know this from
experience, and I will try to convey this experience as someone who
has built a business from the sale of patches (for the ReiserFS
filesystem) in the only market where I had access to kernel source
code.
Software is unique in that "Compiler" technology allows consumers to
effectively reassemble software themselves.
A compiler is a computer program that takes a set of instructions
about how to build a program (called "source code"), and builds the
software. Almost all software is actually assembled by compilers not
humans, and the work of humans is almost entirely in creating the
source code.
You have probably never used a compiler to assemble software yourself
as a consumer because:
- you are not a Fortune 500 company with a staff of trained system
administrators
- you probably use Windows not Linux, and Windows does not give you
access to the essential facility known as "source code" that your
"compiler" needs to reassemble your software
- the new crusade by Linux to make the compilation process user
friendly has only just started
Because you have never done it yourself, your intuition may tell you
that it is not feasible, or that it is not feasible for a large
market. Beware this intuition, it is simply wrong. The Fortune 500
are a significant market for antitrust purposes, and Linux is rapidly
moving towards making asking compilers to perform reassembly a
friendly experience for average persons.
It is frequently efficient to post-sale integrate software for a large
part of the market, and it is getting more so with time. This is
deeply different from physical products such as cars, in that most
persons do not find it as effective to buy a collection of parts and
self-assemble because they would have to do the work of assembly.
With software, the computer does the work of post-sale assembly, and
the consumer simply tells the computer to do it, goes to make some
tea, comes back, and the job is done.
For instance, the business that I own (Namesys, see www.namesys.com)
made its money entirely from sales of a filesystem (ReiserFS) that was
sold separately from the operating system (Linux) for the first few
years of our business. The revenues from this were enough to support
us. Paying consumers such as MP3.com would take our source code, add
it to the Linux kernel source code, use a compiler, let their computer
do a few minutes of work to reassemble the kernel, and get a better
filesystem as a result of it. This allowed MP3.com to save $20
million dollars according to their estimate. Others in my industry
also sell filesystems separately from operating systems
(www.veritas.com got its start that way, and still makes simply
enormous amounts of money from doing so, there are others....).
Notice that I say filesystem. Your intuitive notion of what is an
operating system probably tells you that the filesystem is part of the
operating system. You may be tempted to think that what is part of
the operating system is not viable as a product sold separately from
the operating system. Lessig thought so, and this is because he lacks
experience selling operating system components in the Linux/Unix
programming industry.
Think of Jefferson Parish, and understand that software takes the fine
distinctions of Jefferson Parish to their extreme:
- Software can be integrated in its functioning, and yet separate in
its sale, and this means separate as a product for purposes of
anti-trust law. (Most software products are functionally integrated
with a separately sold operating system.)
- Software can be integrated in its physical distribution, yet separate
in its sale. (Purchase of a CDROM holding the software is often
separated from purchase of a license to use, and it is often
considered efficient by publishers to bundle physical distribution
without bundling licensing.)
- Software can be sold and transmitted over the Internet with no
physical product created at all.
There is only one characteristic that necessarily defines the
separation of a software product, and that is the license. A license
is a contract, and contractual tying is illegal under the Clayton and
Sherman acts.
Yet wait, if software products are so easily separable, why aren't
there far more OS components out there being sold? Control over an
essential facility is the answer.
Secret source code can be an essential facility the equal of putting a
combination lock on every bolt in a car, and then declaring the
combination to be a trade secret.
You wouldn't allow this for a car, yet traditional industry practice
is that source code is kept a trade secret. The crisis our industry
is facing, in which monopoly control is the norm in all parts of it
not in infancy, is directly caused by this industry practice of secret
source code. It is not necessary that the text be kept secret for
copyright protection on books to be maintained, and it is also not
necessary for software that the source be kept secret to protect
ownership of it. Far from it, the underlying historical motivation of
copyright and patent laws is to bring more information out of trade
secret status.
We have a widespread well-entrenched industry practice that keeps an
essential facility (source code) under the control of monopolists (of
which Microsoft is merely the largest), and we have almost complete
monopolization of the software industry in each of its mature niches.
These are cause and effect.
I pray to you to not allow their continuance. Open up the operating
system source code, and go even further. Declare that software is per
se separable where source code is available. Declare source code to
be an essential facility. Return copyright and patent practices to
their historical roots, and require that information created be made
public if it is to be protected.
Please do not hesitate to ask me to comment in greater detail or
respond to your questions in this matter. I am available for in
person testimony if desired.
I have great respect for Reilly and Lawrence Lessig generally, and for
their arguments in most other matters, and I hope it is understood
that I merely have an advantage in possessing "patches" sales
experience.
As for my needs, please create the legal conditions which will allow
me to port ReiserFS to Windows and sell it separately from the
operating system, by giving me the access to source code that I need
to do the port, and to sell the patch separately from the OS.
Essential Facilites Related Citations
[U.S. vs. Trans-Missouri Freight Ass'n, 166 U.S. 290 (1897)] is the
original precedent.
[MCI Communications v. AT&T Corp. 708 F.2d 1081 (7th Cir.),
cert. denied, 464 U.S. 891 (1983)] describes a case more recent (it is
a persuasive rather than controlling authority). Note that the 4 part
test lacks any component referencing the need for a market to have
been active at some point prior to the refusal to deal, and is the
better for that lack.
Profit To The Monopolist From Tying
The Chicago School, to which the current DOJ administration adheres,
holds that there is no incentive to monopolists to engage in tying
because it believes they cannot extract more profit from forced sales
of the tied product than they would from raising the price of the
tying product, unless business efficiencies exist. For this reason,
they feel that there is no need for the Clayton prohibition against
tying, and feel there are civil liberty reasons to avoid government
intervention into markets. Their analysis assumes the tied product is
part of a fully competitive market, and for this reason it is deeply
flawed.
The profit to the monopolist from engaging in tying is the difference
between the market price and the marginal cost. For less than fully
competitive markets, which is to say most markets, this is a non-zero
amount. For software, especially software sold and distributed over
the Internet, the marginal cost is close to zero, and the motivation
for engaging in tying is extremely high.
Senators Sherman and Clayton were much more knowledgable about
economics than the Chicago School is paid to think (various
monopolists have given large funding sums to pro-trust law schools).
Some might like to think that, but for government, free choice
expressed in the market would free us, but in sad reality the
government is not the only means by which people organize to control
and plunder the public. Cartels and monopolies take away our freedoms
as well. The only thing worse than a government controlled economy is
a monopoly controlled economy.
The Settlement As A Whole
I am opposed to the settlement as a whole. President Bush owns stock
in Microsoft, and he appointed to head the antitrust division at the
DOJ someone who is widely known to be opposed to laws against tying.
When someone is opposed to a law that they are supposed to prosecute,
they should not be allowed to settle a case their predecessor started.
The proposed settlement is designed to be toothless, and to do
nothing. Do not allow President Bush to settle this case, and thereby
cripple the ability of the next administration to enforce the law.
The failure of Microsoft and the DOJ to adhere to the contact
disclosure provisions of the Tunney Act is one more reason to reject
the settlement.
Conclusion
If you have the courage to firmly reject this settlement, if you
declare software to be per se separable, and if you move aggressively
to enforce the claim of the States while we wait for a new
administration, you will have earned the admiration of the American
people. Some of them will even know this. More importantly, you
will.
Sincerely,
Hans Reiser
Owner/Operator of Namesys
Author of ReiserFS, a significant component of Linux
5918 Marden Lane
Oakland, CA 94611
phone: +1 510-339-1044 (USA)
+7 095 290 6405 (I am currently in Russia)
(Submitted by Hans Reiser of Namesys)
| 
