DT as part of a scientific/technical workflow

I have DTPO and I’m still struggling to integrate it into my workflow. Part of the difficulty I have is because it seems that using DT involves a bit of a time overhead. For instance, if I create a database of PDFs, I might want to link to my Bookends references, or create tags…all of which require some time investment. Does anyone use DTPO with scientific papers (in my case seismology and geophysics)?

One area where I could see potentially using DT is to keep track of my research. A lot of what I do involves writing simple code (usually in Matlab) making figures or data files, and then examining the output. I could envision there might be a way to keep track of how tweaking a particular parameter affects output. I’m curious if there are any science-y folks out there that might be using DT for similar reasons and how you do it. Right now when I have the time I keep stuff in Curio, but more often than not I just keep stuff in the directory where it is created.

I feel like DT has real potential, but I just can’t grok how to use it effectively.

Take a look at Christopher’s blog here



organizingcreativity.com/201 … -workflow/

dellu.wordpress.com/2012/11/16/ … -mac-lion/

lucbeaulieu.com/2014/05/10/new- … think-pro/

All of the above have also participated in this forum in the past.



Thanks. I’ve read many of these posts at various times, but have struggled to adapt their processes to my work life. But maybe it is time to re-read them and think about it some more…

Could you share an anecdote or two about how the “processes” you’ve read about are not working out for you?

It’s never a good idea to try to mirror someone else’s working protocols, nor is it a good idea to dive in and get lost. Databases do require time to maintain – but if you experiment with a couple of simple routines at a time you might enjoy success.

Databases of PDFs linked to Bookends is a very common practice here. Yes, it requires a time investment. If the time involved is not worth the benefit, then struggling with the process is not worth the time spent.

You mention the need to track your Matlab work, record notes about results, and store the notes and code in the same location. That suggests a database for your projects. Groups for particular coding tasks. Perhaps subgroups for the code and other subgroups for the coding diary. Or a single group. Experiment and see what works best.

I’m still learning, so take this with a suitable grain of salt, but I might have an hour a day to read a paper or to work on my code. Most workflows that I’ve seen seem to be quite involved in how they read read, annotate, link, and connect papers (e.g. organizingcreativity.com/201 … -workflow/). I’m trying to figure out if the long-term investment of doing this is worth, say, a 20% tax on the time I am actually reading papers.

I’m sure I’ll come to a workflow that works for me, but I’m hoping to speed up this process by learning from the experiences of others. Right now I like to read and annotate PDFs stored in the cloud with GoodNotes on my iPad (this is where my Bookends attachments are stored). I am trying to create groups of PDFs (indexed from my cloud storage) for a given project or topic in DTPO, and I’m trying to type in my annotation in DTPO, and also separately add the annotation text in Bookends, which also contains a link to the paper in DT. I’m also trying to link citations in PDFs in DTPO to other PDFs that are also in the database. I haven’t done this enough to get a payoff yet, but I feel my approach is inefficient.

I suspect it may feel inefficient because of the massive number of possibilities that software like this offers. I speak from experience. With such a huge range of possibilities, you never quite get convinced that you’re ‘doing it right’. I’ve been a user for just over 10 years. In that time, my use of the environment has changed. In some cases, this has been a function of adapting to take advantage of new features, but in other instances it has been a consequence of a conscious effort to find a comfortable workflow (which is what it seems like you are seeking) and playing around with slightly different approaches. I’m now have several large and well-curated databases that provide me with unbelievable precision in ‘info mining’.

Here’s one example. I’ve accumulated almost 3GB of academic journal articles (not included the 10GB+ of other industry-specific documents) over the years. All of those files/articles are sorted by journal title and, ultimately, reside in a single folder called “Research - journal articles”. That folder is indexed in a separate DTPO database which is synced to Dropbox, thus making it readily accessible on (1) another laptop immediately to my right (I’m playing with a two laptop/screen setup these days) and (2) on my iPad (using DevonThink To Go, or DTTG). The benefit of this is that I can take any article and highlight it using DTPO, DTTG, PDF Expert (iPad) or Preview (Mac) and the highlights persist regardless of which app I use in the future to view that same article. More importantly, I can leverage the power of DTPO’s indexing capabilities and find other articles that relate to the one I’ve selected (the so-called “See Also” feature). It’s not unusual for me to find things I did not know I had.

As I write this, there are features which I still do not take full advantage of, and one day (perhaps this coming summer) I’ll take the time to see how they might fit. Case in point is the use of annotations (which I believe you mention). I’ve read lots about how others are doing it, but I’ve not yet taken the time to see if it works for me. I agree with korm: you can’t easily copy someone’s workflow and hope for similar results. Study it, try it and assess.

I guess the point is that it can be a long-run game. It’s taken me a bit of time to figure out my ideal workflow, to be sure, but the sunk cost of my time has, in my view, been worth it.

Hope it works out for you.

Ah, I can see why that feels like a tax. You’re annotating in GoodNotes, typing the annotations in to DEVONthink, and typing them in Bookends. I’d suggest not adding the annotations to Bookends, and considering doing your annotations in DEVONthink on the desktop, with either the built-in annotation template or one of the other custom approaches that forum readers have shared.

In other words, find a working mode that lets you annotate once and copy never.

This pushes your annotation work toward the desktop, which might be unattractive given your interest in using the iPad for this. My own preference, for technical and productivity reasons, is to do research on the desktop when serious reading + note taking are involved. I don’t and wouldn’t use DEVONthink to Go (or any iPad app) for serious work involving libraries of documents and complex research – it’s great for carrying around documents but for working through a pile of research the iPad is a very inefficient tool in my opinion.

(Case in point: over the weekend I had to reclassify / regroup a large number of documents I had accumulated in the Global Inbox. For jollies, I thought I’d try this task with DTTG. Whereas on the desktop I could just drag a document from the Inbox to a group and I was done, on the iPad I had to do from 3 to 5 (or more) presses and touches, across 3 or 4 menus, for each move. My fingers got stronger, but my patience got a lot weaker.)

If your annotations are in separate files that link back to the PDF (a common approach) then those annotation note files are available to the DEVONthink AI which can be used to suggest related materials – i.e., select a file containing notes and use See Also to suggest relations.

It takes a while and a bit of persistence to get to the point that you have accumulated a critical mass of documents and notes for the value of the AI to kick in.

Thanks. That’s essentially what I was wondering. I’ll give it a go, and with summer coming up research time becomes much more free.

Sometimes its hard to find the right tool for the job. Keeping track of those sort of results might be a better fit for a scientific notebook like Findings.


I’ve been trying to incorporate the iPad into my workflow a bit more now that DTTG is increasingly feature-rich and the iPad Pro is a very good-performing computing device. My goal is to have near-symmetry in my capabilities whether I’m on my Mac or my iPad (so I don’t have to worry about which one I take with me or have available, or so if one has downtime I’m not left with reduced functionality, etc. etc.)

Long-form academic writing, reading, and annotating are my primary work tasks. I have roughly (and badly) categorized PDFs in folders stored in Dropbox, indexed by DTPO and synced with DTTG.

I have an ongoing queue of articles or books to read tagged as “ReadingList” in DT, and sometimes I go more granular if the reading pertains to a specific manuscript or project.

When annotating, I typically do the following on either Mac or iPad:

  • Open PDF in PDF Expert (using the Document Provider on iOS or directly from DTPO on Mac)
  • Open a Markdown file with the same name as the journal article in DT(TG/PO) and make that a secondary (20%) window. I use a level 1 header of Author-Year-Title and use markdown to hyperlink back to the item in DTPO (I have this partially automated on macOS, full-manual on iOS at the moment) (If I’m feeling really smart I’ll copy the citation placeholder from my citation manager as well, and include that as a subtitle)
  • As I read in PDF Expert I take nots using Markdown formatting as I please.
  • I highlight relevant passages in PDF Expert as I see relevant.
  • When I’m done reading, I use PDF Expert to export Annotations and copy those direct quotes to the end of the markdown file. This file gets tagged “ReadingNotes” and is saved in the same DT group as the manuscript.

In the end I have a single file with my own notes and extracted quotes, with the title linking to the original document. The original document has my highlights in it.

I’m still working on the best method to sort and organize my literature (this has been an ongoing project for 6 years…) but overall I’m happy with the note taking and annotating system I have in place.

Great discussion, thank you all.

Korm, thank you especially for all the input over these forums for some time. I have listened to you carefully, with good results.

I have used Papers 2 for some time, importing the pdfs into DTPO, annotating in Papers. But now I am considering a break from that method, based on Korm’s last post (I agree that most of my serious work is currently on the desktop), and on Frederiko’s excellent Annotation script.

So, I was going to go with Papers 3, partly because I have familiarity with it, have a library (which I know won’t translate to DTPO), and because I like the Magic Manuscript function.

My plan was to download PDFs into Papers 3, but to ALSO import the same PDFs into DTPO. Then, I would read in Papers via iPad when I am portable (otherwise, would just save the citations in Papers without PDFs), but do my annotation and note-taking solely in DTPO following Frederiko’s method.

Does anybody foresee an issue with duplicating the PDFs in this way, outside of the extra disk space usage?


@sawxray, that’s an interesting idea. But if it were me I would soon be unable to avoid the temptation to annotate “just this one” on the iPad, and then “maybe just this one too”, and soon I’d have two separate libraries going with annotations in the one that was supposed to be read-only, or maybe both. :laughing:

I have often asked myself the same question as I spent quite a bit of time on figuring out my research workflow during the first few years of my dissertation. You can indeed spend an awful lot of time with tagging, categorizing, scanning/OCR, note-taking,… and it is difficult clearly to see the pay off.

In my experience, the time that I spent on scanning articles and sometimes even books, and on entering their metadata in a reference manager (I use Papers 3 for this) has been well spent. I have moved a few times to other universities and it is enormously helpful to have you entire library with you at all times and not to have to think about the logistics. Highlights and short notes attached to the document have been helpful, too. As you progress in your research, you often find yourself looking back at something you once studied in depth and, unlike during the dissertation, you have less time to look at in detail again. So for most articles my notes would be in the reference manager, not in DTPO.

In the beginning I would make separate folders for each source that I read in DTPO, with my reading notes and some quotes from the source. Using the houthakker script I would connect the DTPO notes to the reference in Papers. This has been impossible to keep up with – too much work. Also, when looking through articles in my reference manager, it would be impractical always to look up the notes in DevonThink.

I had hoped that new thoughts would really take shape inside DevonThink as I would write notes and used the ‘See Also’ feature. But I only got this to work sporadically. I would use the software mostly to collect all the longer ruminations and thoughts that I had (also on things less directly connected to my dissertation) in longer .rtf notes. I would classify those in thematic folders (quite hierarchical) and use tags (though not fastidiously). This has been helpful in the long term, when I would revisit an issue that I had abandoned or more elaborate ideas about an article or book that was key to my research.

For both the PDF’s (articles, etc) and my notes I have noticed that there is a drawback to neatly storing them away in a database, which is that you forget about what you have. I sometimes think that I might serendipitously stumble upon things more when I would keep all of it in paper form. But this probably doesn’t weigh up against all of the times you retrieve papers you’ve already read (because you’ve stored them, tagged them, etc.) which you would otherwise have forgotten about.

When I did really incorporate DevonThink in the research process itself, it was in one of these ways:

  1. I would remember a specific quote but couldn’t find it in my reference database – would then do a deep search through the PDFs with DTPO. Often, I would just export all the PDF’s that I thought the quote could be in and put them in a separate Devonthink database.
  2. I would be ‘stuck’ in starting a new project and would go through all my notes, either manually or by using tags (smart folders) and searches. This inevitably inspired me, especially when combined with the ‘See Also’ or ‘See Related Text’ functions. Sometimes I would start putting the relevant notes into new folders in DevonThink, and so a structure for a paper would emerge.
  3. Sometimes, when pressed for time, I would just be reading legal cases online, jumping from one to the other and forgetting to write down what I was learning in the process. Here, it was helpful to be able to select key passages and import them into DevonThink without leaving the webpage. After a few hours, I would often find connections with things that I had previously captured or written by using ‘See Related Text’