On November 21, a law firm called Wolf, Greenfield & Sacks, P.C saw fit
to issue a
on the evils of the GPL. By the reckoning of Steve
Henry, a "senior intellectual property lawyer" with the firm, the GPL is
This "time bomb" lurks because a popular license for open source,
the GNU General Public License, (GPL) is "viral." The license
attaches to any product with GPL-licensed code, including a
derivative work, he said. The entire software package becomes open
source and the company thus must distribute it freely and let
anyone copy it. A widely used open-source utility, for instance,
could "infect" hundreds of software products and destroy their
We found this reading of the GPL to be interesting, so we asked Mr. Henry
to explain his reasoning a bit. We thank him for getting back to us; for
the curious, we have put his full response
on a separate page. We'll
just look at the core of his claims here. What Mr. Henry tells us is:
Those who portray the GPL as an entirely innocent and voluntary
instrument take a simplistic view of the GPL itself as well as of
both copyright law and contract law. They often project onto
others the benevolent behaviors and actions they attribute to
themselves. The problem is that others are not always so benevolent
and if the GPL is an enforceable contract, then it may not only be
enforceable by the licensor, but also by third-party beneficiaries
(under at least some conditions).
So, if you see the GPL as a contract, those who have received GPL-licensed
software can enforce that contract's provisions against you. How could
that be a problem? According to Mr. Henry:
So, if a company downloads a GPL product, and incorporates it into
the company's product in such a way that the company's product is
considered a "derived" work or a work "containing" the downloaded
code, not only is the company obligated to use the GPL to
distribute its product, but also it is obligated not to charge. And
its licensees automatically receive a license under GPL terms for
the original code. If the company uses a different license (a) it
could be liable for copyright infringement, (b) it could be liable
for breach of contract, and (c) it could be subject to a court
order for "specific enforcement" of the GPL obligation to
distribute the derivative work under the GPL. The licensor of the
downloaded code could enforce the GPL, as might a licensee of the
company (as a third-party beneficiary).
Mr. Henry's point (a) is not controversial; if you use copyrighted work in
violation of the license that applies to that work, you are infringing the
copyright. There is nothing unique to the GPL there. Point (c) is the
crux of the matter: Mr. Henry claims that, if you distribute a product
containing GPL-licensed code, anybody receiving that code could sue to have
your proprietary code relicensed. The fact that nobody has ever
attempted to do this is irrelevant by this analysis; in the future somebody
could make a try at it.
One could argue that, even if this reasoning holds, there is no real
problem here. If a company does not wish to abide by the terms of the GPL,
it should simply avoid incorporating GPL-licensed code into its products.
Once again, the GPL does not differ from any other software license in this
regard: if you do not like the license, nobody forces you to use the code.
But the fact is that, by this argument, GPL-licensed code is more actively
dangerous than other code. If you get caught using somebody's proprietary
code, all you have to do is settle the copyright infringement claims and
get on with life. With GPL-licensed code, you still have the infringement
issue, but you could also be forced to give your proprietary products
away. That would be a heavy price for a company to pay just because one of
its employees slips some GPL-licensed code into its product.
But does this reasoning hold water? We dropped a note to FSF counsel Eben
Moglen to get his opinion on Mr. Henry's argument. His response was:
So far as "specific performance" is concerned, there is *no* legal
support for the claim. "Specific performance" is the name of a
contract remedy; the GPL is not a contract. In the event of
copyright infringement the relevant possible remedies are: (1)
damages, actual or statutory; and (2) an injunction to prohibit
If the GPL is not a contract, what is it? If you look at §106 of the
U.S. copyright code, it states:
Subject to sections 107 through 121, the owner of copyright under
this title has the exclusive rights to do and to authorize any of
the following: (1) to reproduce the copyrighted work in copies or
phonorecords; (2) to prepare derivative works based upon the
copyrighted work; ...
One of the rights given to copyright holders is to authorize others to
create copies and derivative works. The GPL is that authorization: you
have the right to create certain kinds of copies and derived products from
GPL-licensed code. You have not signed a contract with the copyright
holder, and you have not paid any sort of consideration, which is a
required part of any legal contract. So you, as the recipient of
GPL-licensed code, do not have any contract rights against those who
distributed that code to you. Even the copyright holder lacks such rights,
though the holder does have the right to claim infringement if the
provisions of the GPL are not followed.
Mr. Moglen concluded with: "This talk about 'incorporating' GPL'd
code in a product leading to forcing the rest of the product open is
scare-mongering." We are inclined to agree. Anybody who is truly
concerned about such issues, however, should discuss it with their own
lawyer rather than taking our word for it.
Comments (55 posted)
Anybody following the SCO Group story is aware that, in the last couple of
weeks, the company has issued a new set of threats. Among other things,
SCO claims that it will, soon, file suit against at least one Linux user.
It is tempting to disregard these threats as just more bluster coming out
of the company. Threats against other Unix vendors have failed to come to
pass, the deadline for the company's "half-price Linux License" promotion
continues to recede, the flood of invoices they promised us never appeared,
etc. Why should things be different this time? When the weakness of SCO's
case and the fact that a copyright suit would require a rather more
straightforward unveiling of the company's evidence is considered, more
lawsuits may seem unlikely.
There is, however, a recent
Gartner Group pronouncement which is relevant here:
SCO has declared in filings with the U.S. Securities and Exchange
Commission that its competitive position could decline if the
company can't obtain additional financing. The latest share issue
will dilute shareholders' investments about 3.5 percent. It comes
on top of a previously announced arrangement giving Boies, Schiller
& Flexner a 20-percent share in SCO if the company were sold. SCO
also received an investment of $50 million from BayStar Capital in
return for 17.5 percent of outstanding shares. We believe that
these moves compromise SCO's mission as a software
company. Increasingly, the legal and financial aspects of the
intellectual property infringement cases will absorb the company's
attention, and a law firm will be in an increasingly powerful
position to set the overall agenda for its compensation. Therefore,
SCO will likely pursue claims against Linux users quickly.
Of course, one could rephrase the above more succinctly: the company has no
revenue stream and the lawyers are running the show. SCO has no real
alternatives to income from litigation at this point, and its lawyers have
nothing to lose from filing more lawsuits.
Gartner could be right: SCO might indeed try to open up more legal
fronts in the near future.
If the company chooses its
targets carefully, it might just succeed in finding one that will decide to
settle rather than get involved in a long intellectual property case.
Or so SCO management must hope.
At this point, however, there is enough
information about the company's claims out there that any SCO target which
takes the time to research the situation may well turn out to be less of a
pushover than SCO might wish. In fact, as SCO carries out its search for
the softest targets, chances are good it will pass over any company which
makes it clear that it will fight back. Potential recipients of SCO
licensing claims would do well to bear that in mind.
Comments (7 posted)
The U.S. House of Representatives passed a version of the "Controlling the
Assault of Non-Solicited Pornography and Marketing Act of 2003," on
Saturday. Commonly referred to as the "CAN-SPAM" bill, the House agreed on
a version of the bill very similar to the bill passed by the Senate in
October. This makes it likely (but not certain) that the U.S. will soon
have a national law governing unsolicited commercial e-mail (UCE) -- better
known as "spam" (or any number of less polite terms) by the rest of us.
Very few outside of the Washington beltway or the Direct Marketing
Association (DMA) seem convinced that the CAN-SPAM bill is going to put a
halt to spam. A number of people, including
several state Attorneys General have argued that CAN-SPAM will make
matters worse, rather than better. There is a fair amount of evidence to
support this opinion.
The CAN-SPAM bill actually has the effect of legitimizing spam so long as
it is non-fraudulent and provides the recipient with a means to "opt-out"
of future e-mails. This is a big win for the DMA, and a major loss for the
rest of us. Having to opt-out of receiving spam from each and every
"legitimate" source of spam is a burden that should not be placed on the
user. Given that there are thousands of legitimate businesses that will
seek to make use of e-mail marketing, users are going to be doing a lot of
What about a "do-not-spam" list? The CAN-SPAM Act does contain a provision
to create a national "do-not-spam" list. This can only be seen as a
tactical error of gargantuan proportions. While a "do-not-call" list may
succeed in reducing or eliminating unwanted telemarketing calls, spammers
operating beyond U.S. borders are unlikely to be deterred by the CAN-SPAM
provisions. Indeed, getting a copy of the the "do-not-spam" list will
likely be a high priority for offshore spammers looking for a
roster of known-good e-mail addresses. Users who place their e-mail
addresses on a "do-not-spam" list may avoid spam from legitimate
businesses, but will still find themselves subjected to unwanted e-mail
from offshore spammers. Happily, the CAN-SPAM bill does not require the
Federal Trade Commission to create a "do-not-spam" list, it only permits
the creation of such a list. Given that the FTC has objected to this
provision, implementation seems unlikely.
Even worse, the bill overrides state legislation that may be more stringent
than the CAN-SPAM bill. This is presented as a solution to the difficulty
for "law-abiding businesses" to comply with anti-spam laws, but complying
with multiple state laws is a cost of doing business. This should not be an
excuse to shift the burden to users and organizations rather than
businesses seeking to advertise their goods or services. By overriding
state laws that require "opt-in" rather than "opt-out," the CAN-SPAM Act is
giving merchants free reign to send unwanted spam, at least until the user
asks to be left alone. While one may argue that any laws against spam are
unlikely to be effective, at least laws like those passed in California are
stacked in favor of the user rather than the spammers.
Some have claimed that the CAN-SPAM Act may make anonymous e-mails illegal
altogether. John Gilmore argues
that the bill would make it a crime "to use any false or misleading
information in a domain name or email account application, and then send an
email." However, this is a somewhat liberal interpretation of the bill,
which actually says:
Whoever, in or affecting interstate or foreign commerce, knowingly...
(3) materially falsifies header information in multiple commercial electronic messages and intentionally initiates the transmission of such messages,
(4) registers, using information that materially falsifies the identity of the actual registrant, for 5 or more electronic mail accounts or online user accounts or 2 or more domain names, and intentionally initiates the transmission of multiple commercial electronic mail messages...
A close reading of this language indicates that merely sending an anonymous
e-mail or e-mail with a falsified header would not automatically be a
crime. The provisions only apply to those "in or affecting" commerce, which
would seem to exclude a user who sends an anonymous e-mail for
non-commercial purposes. It might be that the language could be abused to
include someone who has sent an anonymous e-mail that may have some impact
on a business, perhaps a whistleblower or disgruntled customer sending out
negative commentary about a company, but then the user would have to send a
relatively large number of e-mails. Further on, the bill classifies
"multiple" as "more than 100 electronic mail messages during a 24-hour
period," up to 10,000 during a 1-year period.
Unlike some of the state laws, which allow users to sue spammers directly,
the CAN-SPAM Act seems to put users at the mercy of others to take action
against spammers who do not comply. The Act explicitly addresses the
ability of state and federal agencies to prosecute spammers under the
provisions of the Act, and provides authorization for ISPs to bring action
There are a few good things about the CAN-SPAM Act. The bill specifically
states that nothing in the bill requires an ISP to carry or deliver
spam. This prevents spammers from claiming that an ISP is in any way
required to deliver spam, even if it is explicitly legal. The bill also
contains a provision that allows the court to force a spammer to pay legal
fees for the party that initiates proceedings. This may make it more likely
that prosecutors will take on spammers who violate provisions of the bill.
CAN-SPAM also makes it illegal to for spammers to use open relays or other
methods of hijacking computers to send spam, and requires a working method
to opt-out of e-mail. Again, these provisions are unlikely to deter
offshore spammers, but the provisions are welcome nonetheless.
Finally, the bill provides for vendor liability. This means that if a
vendor contracts with a third party to send e-mail on their behalf, the
vendor can be held liable for failure to comply with the CAN-SPAM
provisions. This prevents companies from contracting with offshore
spammers to escape legal liability.
In all, however, the CAN-SPAM Act is disappointing legislation. It fails to
affirm users' rights to consent to e-mail marketing, and instead burdens
them with the responsibility of opting out of unwanted marketing. The bill
will negate tougher state laws against spam that have the backing of the
general populace in favor of weakened provisions that are backed by
lobbyists. After more than six years of Congressional foot-dragging, we
will likely be stuck with a law that does little good, and may even serve
to exacerbate the problem. It may well be that the spam problem is not
solvable by legislation, but, even if it is, the CAN-SPAM act is not the
law we need.
(For those who are interested, the full text of the proposed law is
available in PDF
Comments (21 posted)
Page editor: Jonathan Corbet
Inside this week's LWN.net Weekly Edition
- Security: Attacks on free software infrastructure; New vulnerabilities in iproute, pan, ...
- Kernel: BSD security levels for Linux; Review: Linux Kernel Development; kmsgdump.
- Distributions: Interview with Andreas Typaldos, Xandros CEO; new - Firenet mini linux
- Development: The MlView XML editor for GNOME,
new versions of Ogg Vorbis, omniORB, PostGIS, ntfsprogs, Barbecue,
Net-SNMP, Red Carpet, CUPS, ecasound, Rhythmbox,
Passepartout, GIMP, Gaim, Wine, RTMix, gputils, Bluefish, Mozedit.
- Press: 2004 predictions, Linus speaks from the Cruise, KDE at Comdex,
lots of interviews.
- Announcements: Lindows deploys 30K systems in Canada, MS Voucher refund,
CELF publishes Baseline Source Tree, Desktop Integration Bounty Hunt.