Publish or Perish – typography where are you going?

Publish or Perish is a well known phrase describing the life of us scientists at the university – publish as much and as quickly as possible, or you perish! Many things have been written about the insufficiencies of this concept, many of the most famous mathematicians would have dropped out (been thrown out?) of their scientific career. But today I want to write on the other side, the journal side. Publish here means publishing as quickly and as much as possible, understandably from the economical point of view. But perish is something different for me: The death of Tyopgraphy!

I recently got the final proofs of one of my articles, and I’m wondering how much worse it can get!

But before continuing I have to say that the publisher reacted promptly and fixed the spacing (with one small exception the layout is now fine). I still post this here since it should work as a warning for those submitting articles for publishing.

For those out of this business, here is the typical workflow when publishing an article, but with focal point on the actual publishing part instead of the scientific evaluation:

• scientists write an article, in mathematics often with LaTeX
• they decide on a journal, and send a pdf to the editors of the journal
• reviewing process
• after acceptance, the editors hand the paper (pdf and LaTeX source) over to the publisher
• the publisher, in most cases, sends the paper off (often to India) for rekeying
• the rekeyed document is sent to the authors for review and final proof
• after final corrections the article is published, normally first on the web and then in print

Now the key part here is rekeying: this is not putting a new key into a door, nor exchanging cryptographic keys in a protocol, but typing in the document in a different publishing system. I don’t know what “my” publisher uses, but it seems InDesign from their web page. And the conversion from pdf/LaTeX source into a different system needs mathematical (or at least typographical) knowledge.

Here are now some examples of how bad it can get. The article we submitted was as usual heavy on formulas. During proof reading both of the main authors had problems to parse our own formulas. We were wondering, what should new readers do?

Let us look at an example of what has happened. In the above image I superimposed excerpts from the proof and from our document, showing the same formula. The upper, narrow one, is the proof, the lower, wider one, is our document as submitted. It is obvious that the rekeying removed all the spacing necessary to easily, with one view, tokenize the formula. At the above size this is not so obvious, but at normal size in a paper it would come over like this, which is hard to parse:

Another point is inconsistent spacing. Like with every art, also math is approached with a certain visual expectation. Consistency is one of the things. Here a nice example how it should not be:

It makes me sad to see typographic quality going down that far. I understand the pressure to have a streamlined publishing process, that at the same time works for the web as well as print. But doing this on the costs of readability, which in the case of formulas means understandability, is not a good move. Not for the publishing industry, not for the authors, not for the readers, and not for science.

Knuth created TeX, Metafont, and the CM fonts out of his dissatisfaction with the decrease of printing quality when publishers switched to photo typesetting. He showed us that excellent typography is still possible, and made all tools available for free. How sad that only few publishers use these tools to provide high quality typesetting.

At the end I want to repeat that the publisher fixed the spacing after we contacted them. The only remaining issue is that when justifying a line, the to be distributed space is also added to inline formulas, which means that some of the formulas look spaced out. Anyway, the final article will (hopefully) be published soon. Thanks to the publisher for taking our suggestions and critics serious. But it shows that errors and insufficiencies will appear if we rely on typesetting systems that do not contain any knowledge about proper math typesetting. Humans, and especially often untrained rekeying workers, will mess up.