When you read a paper about "current research" or "recent findings" of a project, are you sure this project is still ongoing or finished only recently (considering the publication date and the date of reading)? Are you really reading about current research, about something that the authors are currently working on?
If the paper appears as part of proceedings of a computer science or computational linguistics conference, it is fair to draw this conclusion. A paper published in fall 2005 most probably describes research from late 2004 or early 2005. The proceedings appear at the date of the conference at the very latest. Of course authors have to submit their final papers a few weeks in advance—publishing with Springer, as we do for the Workshop on Systems and Frameworks for Computational Morphology (SFCM), requires editors to submit everything to eight weeks before the conference date, so authors have to submit their final paper roughly three months before the conference (and thus before the presentation of their work). That's pretty fast for an actually printed publication. It could be even faster for electronic publication only, reducing the time span to maybe one month.
However, if you look into other disciplines, publication speed is much slower. The last project at the University of Basel was somewhat interdisciplinary, involving linguistics and computational linguistics. So we went to conferences/workshops in both fields and we also published in both fields. As there are rarely proceedings for linguistic conferences that actually appear at the date of the conference, we usually submitted a paper to a call after the conference. Mostly, those papers appeared in edited volumes as part of a book series.
Due to different publication speed, the very first article we wrote (corresponding to the very first talk we gave during the project—there had been talks on the topic before the project started) appeared after the project was finished. So everything we said about how to tackle various challenges and what we would like to achieve was published only when we already had those results. Which in general isn't that big of an issue if we would have published all other papers in a similar way: articles on single aspects of the project or on the outcome would appear later.
However, some of the more technical or NLP-related aspects we published at NLP-related conferences. So we now have the strange situation, that the somewhat "starting" paper presented at a conference in April 2011 is published much later (mid 2013) than papers on the infrastructure we developed and used (fall 2012). Someone trying to follow the project thus has a hard timer figuring out what to conclude from which publication.
Late publications are often due to slow processes during submission (extension of deadlines on request of other authors), during review, during revision and resubmission, during editing, and then during actual printing or putting it online. Together with Robert Dale I wrote a handbook chapter which finally appeared now, in February 2014. The whole book project started in early 2011 (probably even earlier as there probably had been negotiations with the publisher first). We submitted our chapter on time, received two reviews and submitted a revised version (i.e., the final version!) in October 2012. And now, one and a half years later, we finally got the printed book. In the meantime I changed universities twice (from Basel to Konstanz to Stuttgart). So I had to report a change in affiliation, author bio, e-mail and postal address twice. If I wouldn't have reported those changes, the editors wouldn't have been able to reach me to ask me look at the galley proofs. (In the end, the publisher send the book to a totally strange address I never reported, anyway ...)
For most of the delays in various processes, explanations can be found. Sometimes someone gets sick, but most of the time it is due to poor production processes. Submitting articles in MS word format with graphics and tables as separate files forces the editor/copy editor to spend a lot of time actually producing appropriately running text including figures and tables. Marking keywords manually on paper slows down indexing extremely. And so it goes on and on.
Given the electronic tools we have today in document processing and document engineering, there is no real reason for slow publication speed. Apart from the discussion on open access and how authors can make an impact by preferring open access publishers, authors can influence the speed of publication by choosing publishers or publishing methods with reasonable processing speed.