\title{Something is happening, but you don't know what it is}
\author[Peter J. Cameron]{Peter J. Cameron\\
School of Mathematical Sciences\\Queen Mary
and Westfield College\\Mile End Road\\London
E1 4NS}
\begin{Article}

This is intended as a worm's-eye view of what is happening to
\TeX\ in the mathematical community at present. It seems
to me that there are some problems.

I have earned my living as a mathematician for 24 years.
For half that time, the tools of my trade included a portable
typewriter and large quantities of Tipp-Ex, and all formulae
were written in by hand. For the next six years, I used various
word processors; no more Tipp-Ex, but still handwritten formulae.
Then I discovered \TeX, and took to it with the inevitability of
a love affair. Is it always so difficult for a love affair to
last? \TeX\ and I have started having problems, and perhaps the
fault isn't all mine. Maybe we should have some counselling.

\TeX\ fills two roles in the working life of a mathematician, and
early on I fell into the trap of confusing them. On one hand,
it is for producing masterpieces of the typesetter's art; this
was such a delight! For this reason, early on I rejected \LaTeX:
I was unable to make a \LaTeX\ document look good; and while
ordinary mortals can write or edit \TeX\ macros, only superheroes
can mess with \LaTeX\ style files. In plain \TeX, with the help of
invented or adapted macros and a variety of fonts, I made everything
from books, through class exercise sheets, to character sheets for
my son's role-playing games, all stamped with my own design
(for better or worse).

The other aspect of \TeX\ relevant to mathematics is its function
as a communication standard. Several things contributed to this,
for most of which we have Knuth to thank. Most important is its
free availability on all platforms, and the fact that the input is
ASCII text without control characters, so that it can be sent
by email without damage (even to and from the UK nowadays, although
it wasn't always so!). Gone are the days when international
collaboration involved exchange of letters taking weeks, with the
inevitable crossing of information. I can now exchange drafts of
papers with co-authors almost instantaneously (though, of course,
time differences mean that I usually have to wait a day for a reply
from Australia).

Another significant fact is that email and \TeX\
provide a mechanism for communicating formulae in an email letter.
Between \TeX-literate correspondents, such snippets of \TeX\ are
never compiled (except, arguably, in biological computers).

Besides the two mentioned above, a third factor has contributed to
establishing this standard. Knuth, as a mathematician, designed the
plain macros to be as close as possible to the way that formulae are
pronounced by mathematicians. Thus, \verb|x \over y| produces
$x\over y$. (One of the few exceptions to this is that we have to
say \verb|\bar x| rather than \verb|x \bar| to get $\bar x$.)

My first encounter with \TeX, before I knew what it was, came about
when an editor sent me a referee's report in uncompiled form. The
dollar signs were a bit mysterious, but if I ignored them, the
rest made quite good sense!

Regrettably, many popular macro packages have lost sight of this
point, and seem obsessed with the need for all operators to be
prefix. Mathematicians, brought up with the infix and postfix
arithmetic operations, are free from this artificial hang-up, and
never refer to $x\over y$ as \verb|\frac{x}{y}|. (Well, maybe not
quite --- but this is certainly true for formulae simple enough to
be put into a letter.)

Now there is a clear conflict between these two roles of \TeX.
It was borne in on me when I wanted to send my beautifully-crafted
preprints to other people. At best, I could send along several macro
files, and assume that my correspondents could follow the instructions
for naming and using them and cope with the organisational problem.
At worst, the recipient would lack a font I'd used, and would be
unable to print the document at all. So, inevitably, I was forced
into keeping two copies of each file, a fancy one for myself, and a
plain one for everyone else. This made updating the files a nightmare,
especially when one was at home and the other at work. The next stage
was to abandon the fancy files, and keep everything as plain as
possible!

These problems, if understandable, were at least self-inflicted. But
it seems to me that the academic/publishing community is now falling
into the same trap.

It is now very common for publishers to encourage electronic submission
of manuscripts. Among the specialist journals in my field, with one
rogue exception which specifies WordPerfect (stop laughing at the back!),
the system of choice is \LaTeX, with a proprietary style file to
reproduce the existing look of the journal. Some of these style files
are less than perfect. (One publisher, attempting a clever redefinition
of \verb|\emptyset|, ended up leaving this command undefined. Another
insists on printing the journal's copyright message on my preprints.)

Often, these style files tempt the author with added features, from
the trivial (an \verb|\email| command to print the author's email
address) to the valuable (a \verb|proof| environment for the proofs
of theorems). If you bite the apple, you can no longer compile your
paper in ordinary \LaTeX, and so you can no longer email it to your
collaborators. Yet some of the features are too good to miss, and the
journal makes others compulsory. So, once again, I have to maintain
two copies of my files.

Further problems are caused by the proliferation of \LaTeX\ versions
and font selection schemes. Rather than stick to the lowest common
denominator, some journals provide elaborate format-switching
mechanisms whose instructions are very difficult to decode.

Electronic journals pose still more problems. We are told that this
is the way of the future, and that traditional journals will quickly
die out. Yet I am sure that many academics, (and not only in the 
Third World), are unable to read or access these journals. We get
busier and busier as time goes on, and installing Mosaic and all the
necessary supporting software on your computer is a non-trivial job.
And on a more mundane level: an otherwise excellent electronic
journal in my field has, as virtually the only style specification,
the use of \verb|cmcsc8| for the headline. This font is not in the
em\TeX\ distribution. Fortunately, the \verb|.tfm| and
\verb|.pk| files for this font were available on our Unix machines,
so I was able to download them and correct the lack. How many
beginning netsurfers would be deterred, by choice or necessity,
by just such a small irritation?

If publishers do force us into using discordant versions of \TeX\
by such means, then the role of \TeX\ as the standard for mathematical
communication will be threatened. If this is lost, one of the major
advantages of \TeX\ over other systems will go with it. Can we
save the situation? I do not believe that standards can be established
by wishing for them, even by formalising the wishing into a committee.
The only thing that seems to work is the commercial success of
particular hardware or software. But what could we ask of \LaTeX3
(or whatever is to be the standard)?
Just two pleas come from the concerns I have raised:
\begin{itemize}
\item It must be possible to impose different styles with the
absolute minimum  of change to the input file. This means that
all publishers' requirements must be anticipated and default versions
included in the standard style. Sounds totally impractical? But we
know what happens if you don't!
\item Either all the plain \TeX\ mathematical commands should be
available, or the commands that are actually used should conform as closely
as possible to spoken mathematics.
\end{itemize}
 
\end{Article}