L2/99-342


Subject:  AFII font for UCS:2000
From:     Asmus Freytag
Date:     October 22, 1999

Tex and others have raised a number of points worth addressing.
I'll first summarize the status quo and recent history, and then
I'll present my suggested course of action to UTC/L2 at the end.

AFII (Association for Font Information Interchange) used to be
the registration authority for ISO 10036. Among other things
the registration covers glyph identifiers. During the time the
Character/Glyph model was written, it was widely thought that
10646 and 10036 could play the role of the main standard in each
of these domains: Unicode and SC2 would standardize the characters
and AFII would register the glyphs.

Along the way it became clear that 10646 and Unicode had enthusiastic
backing from industry and user communities, and that 10036 did not.
As far as I can determine, the last three years, AFII got only a single
registration request (for the euro sign). Nobody stepped up to request
(let alone pay for) the registration of the missing glyphs needed to do all
the scripts in Unicode 2.1, let alone 3.0.

During the same time, the main customer of the registry, Adobe, switched
away and is now using non-ISO 10036 methods of naming glyphs. If SC34
was a commercial enterprise, it would have taken ISO 10036 offline as
an underperforming asset a long time ago. A corresponding recommendation
has been made by AFII to SC34, but it will take time for SC34 to deliberate
and take action. In the meantime, there is a need for a registration
authority.

Two years ago, Unicode agreed to be available as a registration authority.
Registration does not need the kind of technical deliberation that the
Unicode Technical Committee was created for, but is more of an
administrative activity, and given recent trends, a very minimal
activity, which would be carried out by part time staff as needed.
The decision to do this has been reviewed and ratified repeatedly by
Unicode's board of directors. In the meantime, the interest in the
registry declined further, so this may become a no-op in the end.

At this point, the matter is in the hands of JTC1 which reviews and approves
changes in registration authority.

AFII's other main activity has been printing the character charts for
ISO 10646. This work is largerly finished for the second edition of 10646,
except for some minor corrections. Since AFII has gone out of business,
Unicode (the corporation, not its technical committee) has agreed to
provide the necessary resources to guarantee the completion of this
activity. The fonts collected for the purpose of printing the standard were
donated by font suppliers with the stipulation that they be only be used for
printing both Unicode and 10646 and related documents (e.g. TRs').
However, they cannot be distributed to either member companies, or the
public.

Neither of the two activities that Unicode agreed to take on involve the
Unicode Technical Committee or it's work load, agenda or technical focus.
Nor are they now or expected to ever become huge resource drains on the
corporation.

What might reasonably come up at the next UTC and L2 meeting is the question
whether UTC and L2 endorse AFII's recommendation re: shutting down the
registry for lack of interest. I personally would be in favor of such a move,
having experienced first hand the truly low level of interest in
this registry.

In my analysis there are three main reasons why the registry does not work
as conceived.

o The first reason is that it failed to attract critical mass
among font suppliers and font users. Fonts using the glyph IDs and
software written to support them, would have to be created simultaneously
and to be widely available. This requires dedicated efforts to promote the
technology. In the face of a huge legacy installation with apparently
acceptable existing solutions to the problem, the chances for such
a concentrated effort to be undertaken, let alone succeed, are slim.

o The second reason lies in the fact that the glyph registry is badly
lagging behind character set standards. Currently there are between
10,000 and 20,000 characters in Unicode 3.0 for which not a single
glyph has been registered. If one were to estimate the size of the glyph
complement needed for some of the complex scripts the number of needed
glyphs to be registered would be much larger than that. With the the work
already under way for part two of ISO/IEC 10646, the number of required
glyphs doubles or triples.

o The third reason is the absence of a ready source of funds
to offset the cost of registration. Estimates of the actual cost of
registration range from $5.00 per glyph to $50.00 per glyph, depending
on the size and quality of the registration request. Under the ISO
registration model, registration requests are supposed to be paid for
by the requester. (In the past AFII has itself acted as requester,
essentially using its membership to subsidize the registration.
This is something Unicode has no intention of doing).

In the absence of a strong and widely supported organization speficically
devoted to championing the technology, securing funds for carrying out the
registration and having a creditable program in place that guarantees
the usability (i.e. completeness) of the registry, I can see no future
for the ISO 10036 registry. In fact, one can argue that it has
already been largely bypassed by other technologies.

During the last two years AFII sold precisely two (2) copies of the registry.
Compare this to over 6,000 copies of the Unicode Standard, Version 2.0, sold
to date.

A./

Former President
AFII