BLOGDIAL : We are the best

Date: Wed, 8 May 2002 12:59:19 -0500
From: Matt Deatherage
Subject: Font copyright and embedding problems
To: declan@well.com

Declan:

Agfa-Monotype's response is disingenuous on a number of levels, and I find
it hard to believe that a fine type foundry would not know that.

First, you'll notice that Agfa-Monotype consistently refers to a font as
"font software." That's because in the US, fonts _cannot_ be copyrighted.
Specific typefaces may be trademarked or even patented in some situations,
but the 1976 Copyright Revision Act explicitly removed typefaces as
"artistic works" that may be protected by copyright.

It is an odd law -- a drawing of a typeface can be copyrighted as an
illustration, but the typeface itself cannot. Typefaces may be protected
by copyright in almost every other copyright-friendly country, but not in
the US. The "TypeRight" Web site exists to explain why fonts should be
protected by copyright, including explanation of current laws and why
they're inadequate.

If DMCA or other copyright revisions since 1998 have changed this, I am
unaware of it. That's why digital typeface companies consistently refer
to fonts as "font software." Courts have ruled that while the typeface
design itself can't be copyrighted, the computer program that displays the
font can be protected by copyright, just like any other computer program.
Adobe, Agfa, and others have argued that the Type 1 PostScript code or
TrueType "hinting" instructions are a computer program, and that their
full expression in "font software" is a work protectable by copyright.

I'm not saying typefaces themselves _shouldn't_ be protected by copyright.
In most countries they are, and I don't believe in font piracy. I'm just
noting that in the US, a typeface is _not_ protected by copyright.

> We have become aware of the distribution of a program that has only one
> function, that is, to alter the embedding bits on a TrueType font to
> make the font fully embeddable. In other words, even though the creator
> of a particular TrueType font may have carefully limited the
> distribution of the font by setting the embedding bits to level 1 or
> level 2, this software alters these bits to a level 4, installable
> embedding.

As the creator of a font program, "Tom 7" is allowed to set those bits
however he wants. Unfortunately, one of the leading font creation
programs, Macromedia Fontographer, does not have that capability. In
fact, Fontographer creates _all_ TrueType fonts with improper embedding
information.

The embedding permission bits are stored in the "fsType" field of the
"OS/2" table of any TrueType or OpenType font.

Bit zero _must_ be reserved under this specification. Unfortunately,
Macromedia Fontographer was written in a day before these bits were as
important as they are now. It writes the value 0x0001 into the fsType
field of every TrueType font it creates. That's an illegal value that
should say nothing about permissions.

The problem didn't come up much until Adobe released Acrobat Distiller 4
in 1999. Distiller had previously more or less ignored embedding
restrictions on fonts. With version 4, it started obeying such
restrictions, either in fonts embedded in the PostScript stream or in the
local font on your system. Finding an illegal value of 0x0001 in the
fsType field, Distiller takes the conservative route and does not embed
the file.

This came as a shock to hundreds of independent typeface designers, whose
customers were suddenly telling them their fonts could no longer be
embedded in PDF documents. The designers usually had no intention of
restricting embedding. Rich Sprague published a detailed analysis of the
problem at PDFZone in 1999.

Since Fontographer has not been updated since this started (Macromedia is
now pretty much a Web multimedia software company), no one using it to
design fonts has _any_ way to control the fsType field or the embedding
bits. Fontographer's own control only allows values of 0x0000, 0x0001,
and 0x0002 -- two of those are legal, but they don't provide full control.

Macintosh developers that really want to bit-twiddle can download some of
Apple's bit-level font tools and play with them, but only TrueEdit offers
table-level access to TrueType fonts, and even it's flaky in days of Mac
OS X (it hasn't been updated since 1999 either).

If "Tom 7" wants to exercise his DMCA rights, he may _have_ to use a tool
like his own program, and he's certainly within his rights to do so
because he is setting embedding bits on his own fonts. He also makes the
tool available for other people to use on their own fonts. To argue that
he cannot is like arguing that Agfa-Monotype cannot create or distribute
its own tools that set the fsType field in TrueType or OpenType fonts,
because someone might use those tools to remove embedding restrictions.

Surely Agfa-Monotype is aware that _some_ software has to set the bits to
the desired value. Their own in-house font creation software does it, but
the company would seem to want to deny other font developers that same
privilege.

> It is difficult for us to understand why the author of a font would
> need software to alter embedding bits in his own font, since the tools
> widely used to create TrueType fonts allow the creator to specify or
> later change the level of desired embedding.

Fontographer does not. This is also widely known.

> The Digital Millennium Copyright Act ("DMCA") makes it a violation of
> federal law to distribute a product that circumvents a "technological
> measure" that controls access to a copyrighted work. Embedding bits are
> obviously such a technological measure, the program at question clearly
> circumvents the embedding bits, and the fonts containing the embedding
> bits are copyrighted.

By this reasoning, as noted, Agfa's own software that sets the fsType bits
is also illegal under DMCA unless it never allows making fonts less
restrictive than whatever value is already in the fsType field.

But as another Politech reader pointed out, embedding bits are _not_ a
technological measure. Embedding bits tell _other_ programs what is and
is not allowed with the font. It would be a DMCA violation to hack Adobe
Acrobat Distiller to ignore embedding bits, and it may be a personal
ethical violation (and a copyright violation outside the US, at least) to
change embedding bits in fonts you did not design.

Changing embedding bits is like changing a copyright string embedded in a
font -- it's wrong on many levels, but it's not a defeat of copy
protection and therefore not under DMCA's purview.

> Embedding bits are "technological measures" which cannot be
> circumvented without running afoul of the DMCA. Embedding bits are also
> an important method by which the creator of font software can protect
> the results of many hours of his or her labor and years of expertise.

Repeating it doesn't make it so. Software that refuses to embed based on
the bits is a "technological measure." The bits themselves are not, and
any font designer is free to alter them on his own fonts however he sees
fit.

Easy solutions:

If Tom 7 revises his tool to set fsType fields to *any* of the permissible
values and adds a splash screen noting that it's only for use on fonts you
created, the tool would avoid DMCA scrutiny because then it could *add*
restrictions to fonts instead of just removing them. Agfa couldn't very
easily complain that it's allowed to set embedding bits on its fonts
however it wants, but the competition is not.

If Agfa wants to stop people from clearing embedding bits on its own
fonts, it needs to send letters to the people doing it -- not to people
who use tools ranging from Tom 7's tool to TrueEdit to (for pete's sake)
ResEdit to change the bits. They are not a "technological measure," they
do not in themselves protect fonts (non-embedded fonts are neither
encrypted nor restricted in local use), and Fontographer sets them to bad
values. Resetting them by any means on fonts you created is not what DMCA
tries to prevent.

(Obligatory notice: We covered such issues in MWJ 1999.05.12 and
1999.10.09, from where some of this information originates.)

--Matt

--
Matt Deatherage
GCSF, Incorporated

-------------------------------------------------------------------------
POLITECH -- Declan McCullagh's politics and technology mailing list
You may redistribute this message freely if you include this notice.
To subscribe to Politech: http://www.politechbot.com/info/subscribe.html
This message is archived at http://www.politechbot.com/
Declan McCullagh's photographs are at http://www.mccullagh.org/
-------------------------------------------------------------------------
Sign this pro-therapeutic cloning petition: http://www.franklinsociety.org
-------------------------------------------------------------------------

My summer holliday

Attention NAI PGP users

First synthetic virus created

"Trailing edge tehology"

Aids vaccine 'within five years'