There’s one thing in Rick’s last comment I would reply to, though, and that is the comment that using bibliographic managers “it’s difficult to move your focus away from individual works and towards your own analysis of multiple works”.

It’s difficult, yes, because bibliographic managers don’t force you to think about the way multiple works relate to each other. But I think it can be done (to an extent at least) by thoughtful use of tagging. Especially if you have very detailed tags—e.g. “debate re definition of concept X”. Apply that to a bunch of works and it at least provides you with a solid basis for writing something later that brings them together.

I think DEVONthink will let you set up a file of clippings (and relatedly, a linked html notebook, or wiki, like Styley mentioned) similar to what you can do with CP. That’s something I’m going to tinker with b/c I can see it being very useful.

By the way, if either of you ever get the urge to write a guest post for this site I’d be thrilled. You’ve both shared some very helpful information that deserves to be brought up out of the comments section. Let me know.

The trouble with spending too long on bibliographic information, at least in conventional bibliographic software, is that it’s difficult to move your focus away from individual works and towards your own analysis of multiple works. For example, it isn’t possible in Bookends or EndNote to link to Item B from the Notes field of Item A.

Two other points I would make:

An idea I picked up from the Circus Ponies Notebook forums is to create a Notebook with one topic set up per page, each with a clipping service. Any time you come across a useful source you have to get it into Bookends, of course, but you don’t make notes there: you commit yourself there and then to assign it to the topic where it makes most sense in the context of your work. This means that even if you’re not writing you are continuously questioning yourself as to the meaning of things you find. I guess this is a kind of compromise between making source-centered notes in Bookends, and Styley’s method of actually mapping out the whole structure of the thesis.

Another idea, this one propagated by Steven Berlin Johnson, is that you want to organize your own writing and notes on sources (and actual quotes from the sources) so that you can quickly answer questions such as “What was my preferred definition of Web 2.0?” or “What are Nielsen’s 10 heuristics to evaluate usability?”. Having to rummage around at length to recover such details interrupts the flow of any extended bit of writing, unless you are satisfied with notes to yourself like [insert Nielsen's 10 heuristics here]. His solution is to store everything in fairly small chunks in DevonThink, and each chunk would address one of these problems.
I think all three of these (good reference/citation data; topic- or chapter-organised notes; and easily searchable/browsable notes) are important. I think that even with advances in software at present that means at least two pieces of software, for all but the highly tech-savvy: a dedicated reference manager, and a notes repository, and preferably some system for linking them together.

I guess that’s enough for the short version!

I would do less summarising. I spent a lot of time summarising what I was reading. When it came to writing a lit review, you don’t want summaries, you want critical analysis. You might think that summaries will help remind you of what you read, but they aren’t very helpful for that, either. For one thing, if you can’t remember the original text, you can’t be sure that your summary is accurate, or whether you misunderstood what you read back then, or even misrepresented it. So you still have to go back to the original text before referring to it in your dissertation. For another thing, the chances are that your summary will focus on points the author makes that aren’t necessarily the ones that are of interest to you when you have developed your own ideas more clearly.

More useful than summaries, I think, are accurate entries in your bibliography (i.e. Endnote or Bibtex) which are tagged with a good number of consistent keywords so that you can search and find them later. A few notes to yourself in the bib entry on what the author argues are maybe a good idea too, but you don’t want to waste time on writing this up into full sentences or anything well-structured. (Do, however, include notes on where the actual text is! (URL for electronic papers, library call number if a book, notes about where you have filed it if a photocopy).)

I would also spend next to no time editing or polishing anything I wrote in the first two years or so. You’ll just have to do it again later anyway to make the piece fit into your final draft, and you may not end up using that section at all anyway.

The sort of writing that I think WAS helpful was anything where I came up with an original idea and then tried to build it up into an argument with actual evidence. That sort of writing is useful because it gives you practise in arguing your point, which is what you’ll be doing in your dissertation, and it helps you clarify your own ideas and see whether you really can muster enough evidence to support them. Again, though, I wouldn’t bother with polishing these sorts of drafts until you are dead certain they will be included in the final version, and have worked out exactly where in the structure they will fit.

Mostly, though, as Jim suggests, it’s about organizing what you produce. I write in Latex and I am (now) a big fan of putting notes straight into the text, using the comment symbol (%) to “grey them out”. So if I were starting again, I would draw up a thesis outline, create a latex file for each chapter, add in section headings, and then always make my notes either in my bibliography, or directly in the thesis file. Tagging is also your friend. You can tag places in the text just by using a particular symbol and a tag phrase, and keeping a list of these (e.g. **frameworks**). Then when you want to find any section or note you have tagged as “frameworks” you just do a text search for “**frameworks**. I started doing this later on, and wished I had done it from the start.

Other people use html notebooks for this sort of thing and then you can link between notes too, which might work.

For actual physical items, e.g. photocopies, I eventually set up a system of binders, each labelled with the relevant chapter, and divided by subtopic covered within that chapter, and filed everything in there. That way when it came to pulling together a final draft, I could flip through the binder and make sure I had mentioned all the relevant stuff.

Those are some ideas that worked for me. But I think the main thing is to have a consistent system and stick to it. Right from the start.

The Wikipedia entry on s.k.f. includes this relevant anecdote: “By deliberately using the tactic of incurring sunk costs beyond the point of no return, economic actors may get ventures going that otherwise would not have. In his autobiography, film director Elia Kazan explains how he repeatedly used the tactic of sunk costs to get his films started: “My tactic was one familiar to directors …: to get the work rolling, involve actors contractually, build sets, collect props and costumes, expose negative, and so get the studio in deep. Once money in some significant amount had been spent, it would be difficult for [the studio head] to do anything except scream and holler. If he suspended a film that had been shooting for a few weeks, he’d be in for an irretrievable loss, not only of money but of ‘face.’ The thing to do was get the film going” (pp. 412-13). The same tactic has been used for the construction of bridges, tunnels, and other infrastructure. Very few partially built bridges exist, because once construction has started sunk costs are too high to revert the decision and stop again.”

Nevertheless, there are a lot of unfinished dissertations out there….