I was at a conference lunch with two members of my thesis committee, when someone else asked me when I would finish.  "Two to three years", I said casually.

"We expect you to finish in two," one of them said.  So I'd say it's time to focus and organize so that I can bring together and work on the things I said I would accomplish into a thesis.  This will help me to know what to work on, and will prevent the loss of valuable work (such as searching for articles and annotating them).

I need to do some reading now, so my next step is to find a good system to manage references.  I'm going to try Zotero and BibDesk to start (for mac).  I saw a few others around, and found that they call themselves "reference management software", which helps to search for them.  Try this article: en.wikipedia.org/wiki/Comparison_of_reference_management_software

I'll post an update when I've gained some experience with the software.

I expect to print articles and annotate them by hand - It's just too hard to make e-annotations with math.

I'll recount a curious sequence of events over the past two days, where an interesting product was released just as I was thinking it was needed.  I gave a talk to the Discrete Math Group at the University of Victoria yesterday and posed a question to my audience:

Do you use or know of any collaboration tools for writing mathematical papers?

The consensus was that they email the source files back and forth, agreeing to work on separate sections or having one author edit the paper at a time.  The next day I gave the same talk at Simon Fraser University, asked the same question and received the same answer.  This comes as no surprise, of course, since it is how I do it myself. It might be time, however, to consider doing things another way.

Make no mistake, there is no substitute for meeting face to face.  Nothing stimulates and speeds up collaboration more than meeting in person, as needed.  This is often not our situation, however, and I believe we are long overdue for an integrated collaborative development environment for mathematics.  Why would we benefit from this?  This is not my science, but

1. software programmers use it (they seem to have been thinking about it for a long time http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.95.1704;
2. students use it for collaborative note taking, project writing;
3. Google "realtime collaborative" and it will seem like everyone is doing it except us!

Real time document collaboration is not a new idea;  Wikipedia tells me that the first demo of such an application was in 1968 by Douglas Engelbart.  But it isn't until recently that these tools have become popular and much easier to install and use.  We now have Wave, Google Docs and EtherPad (http://mashable.com/2009/02/21/online-document-collaboration/) to name a few, none of these require an installation, or even registration, if you haven't been living under a rock for the past six years.  Why haven't mathematicians caught on to this?

Typesetting support on the web isn't worth a damn!

To create documents we need to be able to include other source files, define macros, use packages etc.  In this respect, web-based LaTeX, MathML support and such things are just toys.  But there is hope...

In reply to the question asked of this week's audiences, a friend points me to http://docs.latexlab.org/.  A very promising application which will integrate LaTeX support into Google Docs.  The interface suggests that we can expect to be able to choose between our local TeX or the server's and presumable we should be able to include other source files (I couldn't get this to work properly when I tried it).

There are two problems, however.

1. You will have to share your research with Google.  For example, this might breach of your research contract or discourage people from using it.
2. The application does not have inline preview.  (an amazing feature, see auctex for emacs, http://www.gnu.org/software/auctex/).

docs.latexlab.org has the right idea however, because they are addressing the number one issue.

Mathematicians have a LOT invested in LaTeX and they are not about to learn any other typesetting system.

Programmers love making productivity tools for themselves.  It is therefore no surprise that the internet is swarming with integrated development environments for software development and such.  It is time for a such a collaborative tool for mathematicians.  The following features would be ideal:

1. It is easy to install on Mac, Linux and Windows.
2. It is open source.
3. It does not require any new knowledge on the part of the user.  The obvious is that it be entirely LaTeX based.  Less obvious and more complicated is that it should have a TeXShop and, WinEdt (or whatever you people use) mode as well as be available as vim and emacs plugins.
4. It has inline preview
5. It uses a communication protocol that provably protects our privacy.
6. It has integrated voice chat, instant messaging (with LaTeX support) and maybe some kind of whiteboard-type extension.

update/continuation

I've discovered another thing or two.  Two more web-based tools for online LaTeX collaboration are https://www.verbosus.com/ and http://www.scribtex.com/.  More to come as I learn more about this topic.

Join Google Wave to participate in this public discussion