Thanks Markus. The current business as usual in publication could be dangerous, as all effort of reviewers and editors to fix important but minor errors could be jeopardized in the translation from TeX to the publisher's typesetting system.

As both TeX and XML are markup languages, there should be a way to create a standard and downgraded version of the XML-based typesetting system used by the publishers in order to migrate authors to the same typesetting environment in order to reduce the risk of alterations caused by the currently business as usual. At the same time, this will also reduce the costs associated with publication.

I don't think there is anything missing. TeX & LaTeX are amazing, and beautiful books can and are published using them---I own some. Math looks great, text looks great. In use, updating content is easily automated. Typesetting and style is consistent, regardless of content, since it is somewhat enforced by document class and used packages. Once a framework is established, new documents are a snap.

Commercial products like InDesign, Quarkxpress and FrameMaker can do that too, for well trained compositors. Marketing behind commercial packages, and the perceived safety of buying a them are probably key but there is also likely a lower level of initial satisfaction; particularly for those accustomed to fiddling with every page as content is added and those unaccustomed to writing code. Add to that the fact that trivial tasks, like using an arbitrary commercial font or creating a large table, is a bit of work under LaTeX, paying $1000/seat for FrameMaker may not seem bad to a commercial publisher. As for a barrier to xml, there is LaTeXML, though i have not used it.

Thanks Bjorn, I also think that TeX/LaTeX-based systems have the capacity to handle the end-to-end publishing process.

However, in my recent experiences, I have noticed that some publishers use a LaTeX system from submission to acceptance, and then suddenly migrate the work to another system during final stages of proof reading and publishing. From publishing point of view, it is not harmful, but from a scientific point of view, this migration can introduce unintended but severe errors and changes that could not be captured and corrected by the editors or reviewers at those final stages of publishing.

My question is on why such a migration step is considered by the publishers in the publishing process, and how we can avoid it.

I also don't think that TeX is missing anything. I guess, however, that there could be some issues with the conversion of TeX to HTML required to produce a version of the article that can be read directly in the web browser. The journals which do not provide this feature seem to be perfectly fine with TeX alone, as illustrated e.g. by the math journals EJDE (http://ejde.math.txstate.edu/) and SIGMA (http://www.imath.kiev.ua/%7Esigma/).

Dear Dr. Reza Farrahi Moghaddam, Publishers' production offices look for lower labor costs. In general all authors use Word at any of the stages of their write-ups. Majority of authors do not know the advantages of TeX. But the typesetting industry has developed around Word. Also, there are lots of plug-ins and tools (most popular is PTC Arbortext Advanced Print Publisher) to clean up text, structure it, and many other features which all comprises in Word. So, as most typesetters have a Word-centred workflow, with a lot of (very good) development work invested, when they get the odd TeX file they jam it into that workflow, by converting the TeX file into Word.

One my recent books was written in Lotus WordPro, then converted to Adobe PDF and MSWord. It was copyedited in Wales, typeset in India, printed in Singapore, and published in England. When I read the printed book against the copyedited manuscript, I was aghast. So many errors that I couldn't count them all. Spaces were eliminated, resulting in gibberish word combinations; uppercase letters were reduced to lower case, and vice versa; etc., etc., etc.

Digital handling of texts was supposed to improve the whole process, but in my experience, the opposite is true.

Dear Artur, Afaq, and James,

Sorry for being late. It seems that there is no technical challenge that has forced publishers to adapt (free LaTeX -> commercial XML) model. Also, it seems that there is no motivation to avoid the conversion step in the middle of this model. Under such a circumstances, I would suggest that the proof-read version of an article that is sent back by the author to the publisher is transferred back to the editor and reviewers in order to get their final green light for publishing. This could help to correct some errors that have unintentionally introduced in that stage.

I use LaTeX for more than 20 years now, I used it for publishing several mathematics books and for publishing magazines  and I was always satisfied with the result. But there is a condition : that you do all the work yourself. So you must provide 'ready to print' texts to your publisher. I add that I always use a commercial version of TeX, named VTex (made by MicroPress, New York). I consider this product as the best thet was available for it includes a postscript interpreter. Thus I realize the pictures in postscript and include them easily. There is a just a problem: VTex is no longer available! Now I think also that it is time to convert to "floating scientific texts" readable on a tablet. So the aim would be to produce a scientfic text in a format such as ePub. Does anyone have experience in that direction?

I agree Guy that the case of publishing a book is more straightforward and manageable. However, when someone tries to publish an article in a journal, there is almost no control after submitting the accepted version.

In terms of a good combination of tools for LaTeX, I would suggest Texstudio editor + Miktex + PDFLaTeX class. All are free and available. Using the PDFLatex and the graphicx package, you can directly include pictures in their original format (such as PNG) in your LaTeX file without converting them to postscript format..

Thanks to Reza for this answer. I have no problem with bitmap pictures. But the fact is that I realize geometric pictures directly in postscript (using a set of macros called PsGeo2d.ps that is downloadable on my researchgate page). So what is important for me is including pictures in their original format, that is postscript.

I doubt that LaTex lacks any tool, but the problem will be more to the staff who are required to use LaTex.

1.) A "Photoshop" style software may look labour intensive to use, but how labour intensive is it to teach and master? If you look into the staff turnover in these editing houses, the final degree of the workers, you may notice that LaTex may be slightly too difficult and time consuming. If it is in South East Asia, the staff may not even read, write or speak proper English. You are asking these people to be efficient in their task by reading and understanding LaTex help manuals.

2.) An editing house should not trust the source file (Word (.docx), LaTex(.tex), FrameMaker (.fm), InDesign (.indd), etc.). They should only trust the resulting PDF. Now the LaTex cmd of "latexdiff" has no meaning because it is not comparing the final PDF, but the source file. One way is using the "Compare" function in Adobe Acrobat Pro 8 or 9 (not X), the other is flipping the printout of each page. For this process too, how easy is it to compare 2 LaTex created PDFs with Acrobat (or alternative)?

It is not the problem of the tool (LaTex), but the workflow. If it was a loyal, highly paid, relaxed work environment, then there may be time to educate new staff, but that is not the reality. From experience in SEA, if they are bright enough to use LaTex as efficiently as they could use FrameMaker or InDesign, they will not stay in an editing house, they will find work else where.

ps: My assumption is that the software you saw was Adobe Illustrator. (Is's an easy (but dangerous) tool to use for a quick fix, by the printer or the editing house)

pps: Sorry it I missed the point.

Hey Guy, I noticed you produce your images with PostScript and add them to your document. If you don't want to use raster graphics images (JPG, PNG etc.) you could still use the EPS and PDF formats, which are vectorized. LaTeX accepts PDF images natively and is able to convert EPS to PDF with the addition of a few packages. Also, SVG images can be converted to PDF by using a free tool like Inkscape.

Now for the topic at hand, I do believe LaTeX can be a bit cumbersome to work when doing highly customizable work, like book publishers do. Every new book has to be different and LaTeX is really good for the exact opposite: creating standards. It's not easy to choose the exact placement of images and LaTeX has poor support for creating book covers, for example. In these cases, I can see why publishers might choose Word or InDesign. With their WYSIWYG approach, you just fiddle with your images, write text inside floating boxes and they will be right where you want them.