Although I keep all of my research papers and articles in the DTPO database, I prefer writing in Scrivener (for some reason). DT has become the “bucket” for capturing everything that is interesting to the topic - but this is far more than what I can put into the final article. An important step while writing a paper is the creation of an outline, which acts as the “skeleton” for the text. At first I write the chapter titles into scrivener, then “flesh out” the draft by filling the empty chapters with my main points and ideas.
In this stage I change a lot between DT and Scrivener (or whatever the writing app is): Command-Tab and search: oh, I can use this thought in chapter 3, command-c, command-tab, command-v. You could also create a document or a folder in DT, if you prefer not switching apps. (if you like shortcuts, there is also a shortcut for switching windows, remember?)
BUT
copy and past from a pdf is tricky, because even a converted pdf+text will produce ambiguous line breaks and incorrect citations, a cardinal error in academia.
Instead of re-typing every interesting passage from a pdf, taking screenshots is much faster. It is even possible just to take a screenshot of a particular area (e.g. line 1-5 of an article) and place it in the clipboard. In Scrivener or any other word processor, you can then “paste” and - voilà, there is your clipping.
(for keyboard people, the shortcut is, um, shift-command-ctrl-3. I actually reprogrammed my middle mouse button to this key combo, just to avoid remembering those keys)
For later use, it is really, really helpful if you do not forget to write a few comments (into the same line) about the article where you culled it from. As a Bookends user, I include citation delimeters (in case I want to cite it later - e.g., for p.49, {Author Articlename @49}.
The next step is usually re-shuffling the clippings and thoughts, so that they structure your argument.
With that very very rough draft, the actual writing is easy, because all of the arguments and quotations are already in place. I found it useful to see this rough draft in a window (actually, a scriv. pane) while writing the actual first draft.
I found this technique quite helpful when I was writing my first chapter. Usually I create a lot of excerpts, but I was in a hurry and didn’t want to produce unneccessary text.
Which tricks are you using to get past this “capture”-stage of your writing?
I’ve used that method before. The hotkey you listed doesn’t work for me (software specific? sorry if I missed that). Cmd+Shift+4 does a very nice “crosshairs” capture, though, and deposits a PNG on your Desktop, sequentially named.
Thanks for the tip, very helpful. One of my pet peeves with PDF is those weird line breaks that get dumped in when you try to copy text (though devon’s word services reformat tool has saved my much pain over the years!) . I have, however, noticed that the Skim PDF viewer http://skim-app.sourceforge.net/ takes the opposite approach and seems to strip all line breaks from copied text. This is very useful 99% of the time, though a royal pain if you’re trying to copy a bulleted list or copy text that spans a paragraph break.
Thanks, all, very useful suggestions! I’m forever trying to improve my workflow between capture/research (DTP) and writing (Scrivener). If I may ask, how do you folks deal with taking notes in DTP? I’m amassing all these PDFs in my DT database, but I’ve yet to find a good way to take notes on them that I’ll be able to find again. I like Skim’s third-panel note column, but it seems like a waste to put stuff in DT just to take it out again to read it and comment on in Skim. And unless I’ve missed something, I can’t read a PDF in one DT pane and open a new RTF note (to jot some comments re the PDF) in a second DT pane simultaneously. Net result so far: lots of PDFs in my database that I never read or digest! All suggestions and approaches warmly welcomed …
Tzero, I actually use the reverse of your process. I don’t store my PDFs in Devonthink, I only store my notes. Since my database is based on a specific project I am assuming that my notes (as in notes based on ideas as well as notes based on text that I highlight in PDFs) are “high relevance” material and that the rest is, well… less useful to me in that context.
I use skim to highlight text and add notes of my own. I then use skim’s “export” feature to export my notes and the text that I have highlighted into a .txt file. This text file gets chopped up into smaller files (that are stamped with page-number and the citation number of the original document in case I need to go back to the source for some reason) based on aspect of my project and then dumped into Devonthink. Basically my database is a collection of quotes from hundreds of academic papers. I rarely need to go back and look at the original full text documents so I’ve simply left them out of Devonthink. Its extra work, I’m sure, but no file in my database is larger than the display window, so when I search something I don’t have to go digging through a document for it.
Basically, my reading takes place in Skim and my notes/searching/research/writing lives in Devonthink. The advantage, for me, anyways, is that because all my notes are in text files I am free to wiki-link the life out of them while creating references to other relevant notes. I don’t think I could do this if I relied on the notes embedded in a PDF.
Sure you can. You can’t open additional panes in, say, the three-pane view, but you can open as many windows as you like. I often have a PDF up in one window and RTF notes in another.
thanks! that’s an interesting workflow, almost like a (smart) heap of index cards; maybe i’ll try to incorporate. i’m in a similar situation, working on a nonfiction book project, with pdfs and book quotes and my own notes from all corners of the earth, and am trying to stay on top of it all …
Glad you liked it: You can change the keyboard shortcut in the system preferences. I changed it to an F Key because I travel without a mouse at the moment.
For reshuffling larger amounts of clippings, I start to re-discover the offline world. It might sound heretical, but sometimes it feels good to work with something real again, especially after a long day of cut and paste.
For smaller projects, the step of reshuffling the notes and clippings in Scrivener works great (cut-paste-cut-paste…), but I currently produce a heap of clippings, which are hard to organize electronically.
Another paper was due: panic! Suddenly, all of the small notes on the screen made no sense at all.
Plan B:I printed the whole notes and clippings document, cut the resulting heap (with a huge pair of scissors), built different piles on my desk (desk still larger than screen), reshuffle, sort, etc…
To be honest: I even taped some notes together. And yes, I know I could have achieved a similar effect with a decent mindmap application (which can handle pictures and has good print/export options - I tried Mindmanager and conceptdraw MM , without success).
After two hours of sheer kindergarten, writing turned out to be a lot easier. I have the impression that the process of cutting, touching (ouch!) and simply looking at every single f***ing note again (and again) might help. At least, under pressure…
What happened? I found that while it is helpful to have all of the notes stored in the computer, writing is a different cup of tea. The tricky part seems to be not the capture system, but how to inspire the grey stuff between my ears…
I hear that. I think it’s going to be a while before computers can create an environment with as much control and intuitive design as paper and scissors and markers. Not because it’s impossible, but because no one seems to be interested in doing it. In general, OS and application developers seem to be taking for granted that software can’t be tangible – while neglecting to develop the advantages software inherently possesses over things like pen and paper.
The software market has stagnated, as far as I’m concerned. I think we should have Core ZUI in OS X (and something similar in Windows and Linux) to complete the break with the desktop paradigm. I feel like even the most rudimentary mindmapping application, basically a glorified MS Paint, could be almost incomprehensibly cool if it just took advantage of the ideas of the zoomable user interface.
Just a comment about human-computer interaction. I’ve been using a ModBook (a custom Mac tablet based on a MacBook) since 30 January. The MacBook on which it is based has reasonably powerful specs – 2.2 GHz Core 2 Duo CPU, 4 GB RAM and a 200 GB 7200 HD. The conversion turns it into a digitizer-pen sensitive screen tablet or slate computer, with included GPS. The strong glass screen is scratch-proof and provides good paper-like feel under the pen.
As an old Newton user, I’ve always wanted to be able to enter handwritten notes (converted to text, or left as an image, or both), chemical or math equations (as images) and diagrams (as images) into an OS X Mac. Adding a Wacom tablet was clumsy and not very portable.
I learned to write with a pen or pencil in kindergarten, and I’ve been doing that all my life. I use a keyboard when I’m entering a lot of text. But there are times when I want to do things that would be inconvenient using a keyboard (scribbling down a phone number while on the phone, making notes during a meeting) or impossible, such as a diagram (perhaps my visualization of connections among ideas) or just plain free-form noodling with ideas. (The digital pen even has a functional eraser when used with sketching software.) When I’m replying to email while sitting in a comfortable chair in front of the fireplace, I prefer using pen entry with the ModBook instead of trying to type with my MacBook Pro balanced in my lap.
There have been many times as I used my MacBook Pro when I grabbed a pen and paper, often using the back of an envelope if that’s handy. But those notes and drawings on paper aren’t in my computer with most of my other information, unless scanned or typed in. And scanning offers only limited manipulation of the handwritten information.
Apple’s Ink for OS X does reasonably good handwriting recognition for printed handwriting (not cursive), but isn’t as good as handwriting recognition on my Newton 2100, which had gotten quite accurate and allowed options such as using collapsible outlines for notes. Nor is Ink as good as the handwriting recognition in Windows XP or Vista. But there will be much better third-part handwriting recognition for OS X in a few months that does a good job of accurate text conversion of scrawls in cursive, as well as of neatly printed handwriting.
Now I’ve got software and a computer that will let me fill in with text blank fields on a PDF document, enter handwritten initials agreeing to a modification, e.g., on a contract, and enter my signature on the contract – all without having to print the document, make the additions and scan it for transmission as an email attachment. Or enter handwritten or text marginalia in the margins of a PDF.
Keyboard and mouse interfaces with my computer had prevented me from doing things that I had been accustomed to doing for many years.
And of course the pen, used to point, click and drag, also feels intuitive for navigation. A floating and usually invisible onscreen keyboard provides access to keyboard shortcuts, page up/down, function keys and so forth, if I’m not using my wireless keyboard and mouse at the time.
So a digitizer pen and screen extends my interface with the computer in useful ways.
We’re going to see exciting enhancements to the human/computer interface in the future. Steve Jobs is innovating many right now, although I don’t agree with his dislike of pen input; perhaps that’s a residue of his dislike of the Newton. The touch screen of the iPhone and iPod touch works well for those devices. But I’ve never learned to write well with a finger.
I bet it does, and a mod book must be a great tool for capturing information. For scientific writing projects however, the capture stage is only one third of the actual work. The second part is about restructuring the collected information (quotes, ideas etc…) then it is all writing, applying changes on the go. Is the mod book also useful in the transition in the second stage?
Hi, Maak. Sure. It’s a Mac. When I’m using my wireless keyboard and Mac I’m interacting with it as I would with any other Mac.
But in the process of writing and exploring ideas, there are times when I want to noodle around with ideas. If I’m looking at possible quantitative relationships in a set of data I’ll grab a note pad or envelope and try to formulate some rough, preliminary expressions that might be useful. I don’t even try to do that with a keyboard.
Or I might play with a diagram to explore relationships. What things are contingent upon A but not B? What other factors might play a role? Quickly sketching possible networks and interplays is better done with rough graphics that can’t be done with a keyboard but can be done immediately with a pencil and paper. That’s part of the writing process, analyzing and structuring the information with which I’m working. I don’t use mind-mapping software. But I often do make notes that are diagrammatic.
What I like about the ModBook is that it’s a working environment that lets me directly write and draw into it as well as type into it. I do most of my drafting inside my DT Pro Office databases. Now my exploratory notes and diagrams can be entered directly into my computer, rather than done in pencil and paper on the back of a coffee-stained envelope. And they are inside my database, where I can find them.
Software could do more than it does, certainly, but I can’t see it ever replacing a deck of index cards scattered over a really big table. At least until we get to the thought-controlled room-covering display. (See Asimov’s Foundation books.)
I regard with serene curiosity anyone who can say that with a straight face. Index cards and butcher paper are acceptable only because they are really all that we have right now.
This is all coming from the perspective of a mythopoetic writer, so feel free to disregard. It’s probably not applicable to scholarly articles or most of the other common uses for DTP.
One of the best Zoomable User Interfaces in the world is Google Maps. It bests a traditional atlas in several different ways – you can see a satellite view, a transportation view, or a hybrid view at the click of a button. Where it becomes genius, however, is that you can zoom in to see more detail or zoom out to see a larger picture. The closest you can come to this with index cards is to stand on a ladder (everything becomes illegible) or you stick tiny index cards with microscopic messages on top of the standard 3"x5" cards or whatever you’re using. Or, worse yet, you have ten or twelve tables, each at a different “scale,” scattered throughout your dancehall-sized office that you race between, steno pad at the ready.
There is no undo, no saving, no backup, and no easy duplication of elements with index cards. Relational networks are annoying; if you have five main characters (traditionally considered about the maximum that a reader can track), you have ten strings to run between their index cards (and do characters really deserve only an index card? Would you be happy with an index card describing your psychology, your history, your passions, your aspirations, your strengths and weaknesses?), each with an index card of its own describing the relationship between these two characters. Most stories will have supporting characters, who can admittedly in most cases be summed up with a good index card, but it still makes sense to define the relationships between these supporting characters and the main characters they encounter.
If you’re into the whole plot thing, it becomes even more complex. You have your overall plot structure – and probably more than one, unless you’re writing a fable – and within this plot structure you have acts, generally three or more, in which there are multiple situations (each with a beginning and a middle and an end, possibly new characters or settings or other information, dialogue and setting and description and so forth).
That’s plot and character, the two primary components of just about any novel. With elevated ambition or certain genres (historical fiction, high fantasy, mythopoetic stuff) you need copious amounts of information tacked to each index card or strewn about the table – what did a Roman slave traditionally wear? What god does this person pray to, and what is this god like? What were the economic, political, and ecological effects of the discovery of uranium near Mt. Taylor in the mid 1950s? Granted, you can ignore most of these things if you’re writing an inspirational-coming-of-age-in-the-ghetto novel or a slice-of-life-tale-about-a-cynical-but-tender-coffeeshop-and-bookstore-waitress-slash-clerk-at-Grounds-for-Thought, but even then you need to go on Wikiquote to find lines from Nietzsche and Sartre to plagiarize in order to give your younger self some depth and existential angst.
I assume that no one using DTP as a fictionwriting tool is coming from the slice of life or 21st-century bildungsroman sects (but if you are, I hope you’re deeply offended by my characterization). And really, DTP is a fairly good replacement for index cards in all areas but one – the spatial organization facet. In terms of spatial organization, all DTP really needs to become an incredible writing tool is to have a mindmap interface where you can drag and drop representations of documents and draw relationships between them. To become a dream come true, it would need a mindmap interface where the documents would not be linked icons so much as the actual contents of the document scaled to a user-specified degree – in essence, a ZUI. This would take thousands of print atlases and turn them into Google Maps.
If you subscribe to the nauseating metaphor that writing is a journey, I believe you must follow the idea to its logical conclusion: if you have a destination of any appreciable distance from your origin, you need a map, and the better the map, the more elegant the journey. The map becomes the territory in any creative project, so a writing tool’s utility is defined by its ability to remain useful throughout that process.
Examples:
Jef Raskin, who invented Burrell Smith, one of the designers of the original 128K Mac, founded a research organization that did some research into ZUIs. The following is an interactive demo that neatly presents the concept, albeit in Flash: rchi.raskincenter.org/demos/zoomdemo.swf
Even Microsoft digs the concept, although their technology is the result of an acquisition: ted.com/index.php/talks/view/id/129
The problem is that a zoomable interface is still limited by the size of the screen. When the magnification is low enough to see the complexity of the structure, the type is too small to read.
Now, I am a happy DTP user. I wouldn’t dream of trying to store everything on index cards, for all the reasons you give. But at the stage where I need to tie the details into the big picture, there’s nothing better.
I have too, and I agree – but I don’t think it’s a flaw in concept so much as a flaw in execution. I don’t know of any zoomable mindmaps. Really, the best mind-mapping software I’ve used is Graphviz, and it can’t even handle a real mindmap. Novamind or whatever is supposedly the best mind mapping software (from what I’ve read), and I think it sucks.
I think that’s a problem of too few levels of scale. Not saying “UR DOING IT RONG.” I should really do some mock-ups or something in Photoshop to test what I’m saying here. I’ve never tried this, so I might be talking out of my butt.
But I believe that if we have an infinite sheet of paper and can change the size and “frame” of elements arbitrarily (not just choosing between 100%, 50%, 200%, etc), I think the problem you mention would be nonexistent. Clutter might be a problem, but in that case you simply hide everything under, say, 10% of its real size. Or 25% of that size. Or you see the user-definable metadata or an icon preview when it’s at a less-than-legible size.
And I wouldn’t suggest trying to get real work done in a zoomable UI. It’d be a poor (at best) primary interface or even a secondary interface. I do think that with some ingenuity and tweaking, though, it could be a staggeringly useful visualization system. The content, unlike with index cards, could be automatically derived from the already-existing structure of your DEVONthink database. All top-level groups are 200% scale, all second-level groups are 100% scale, and so forth down to the minutiae. That might seem of limited usefulness, but with smart groups and rudimentary input from the user (on, say, the level of that offered to people making smart playlists in iTunes) you might see that this could head in a very neat direction.
I think this is heading in a very vague direction. That’s my fault. Anyway, I might try to come back here with a couple simple mockups and real-world examples of how I think it’d be useful. If I can’t do that, well, it means that I realized I shoved my big foot into my throat again (and I realize that even if I can make a crappy mockup in Photoshop, it doesn’t mean it’s even remotely possible to program)
the discussion takes an interesting turn. I like kalisphoenix’ vision of a zoomable interface a lot, though not for something like my current task. The real problem in idea generation has a lot to do with encoding and remembering data, the basis for the creative jump to new relationships. Now if we look a bit into cognitive psychology, it seems that the more senses are active in a given activity, the better the chance of remembering it.
When I learned latin or japanese, I wrote every character on index cards, with the translation on the back. I would flip through the cards and quiz myself. A lot of applications exist to assist you in this task. Especially for the japanese characters, these programs are not very helpful: Drawing the character with a real pen on real cardboard helps the information “get through”. Looking on a 12’’ (or even 30’’) screen is just another bad-posture-activity with only slightly different keyboard strokes than with any other activity. The information is only broadcasted on the visual “channel”.
When it comes to writing and storing text or other information, the benefits of the computer outweigh the shortcomings. In the stage of idea generation, however, involving other senses might be more valuable (for the task “at hand”) than the shortcomings of the method, as you mentioned: no backups, no save.
As Merlin Mann recently noted, paper is useful as a transitory stage: Either it goes into your head (index cards for learning) or the final product goes into the computer (conceptualizing/ pre-writing), but the paper scraps, remainders of the creative process, can be put into the trash without worries. Index cards for reshuffling and restructuring ideas are only valuable as a transitory stage. Touching and the activity of manually reshuffling scraps of paper is in itself a contribution to the process, so I doubt that anything that happens on a screen can ever replace this.
On the other hand, applications can be improved. Many technological breakthroughs are based on the observation of processes in nature, the lotus effect and how it helps glass manufactureres is only one example. The further development of The App We All Love is another issue. Reverting to paper is only an option for me, whenever the project becomes too difficult to solve on a screen. For most other cases, a (scalable) mindmap-module would be an invaluable addition to dTs existing toolset, because it would fulfill the promise that DT helps the user make sense of large amounts of data. Making sense of something could be translated in: see connections in large amounts of data. At the moment, DT is great in finding needles in haystacks, but much less in structuring the haystack into hay bales, so to say.
(and I realize that even if I can make a crappy mockup in Photoshop, it doesn’t mean it’s even remotely possible to program)