Zettelkasten Forum


DevonTHINK and nvALT integration?

Hi, all
I'm just wondering about the ways in which DevonTHINK and a plain-text Zettelkasten via nvALT/The Archive might work together. I'm thinking in terms of using the former for document storage and archiving, and the latter for notes. Christian has already raised the issue of DevonTHINK's 'smart' search features being a replacement for manual, 'hands on' searching and link-making, with all that this entails for automation dangers and so on (forgive me if I'm misrepresenting your view here, Christian). Does anyone else have any experience of using DevonTHINK with a Zettelkasten (the latter stored separately), or any views on how it might work (or not work)?

What I'd really love to hear, of course, is something along the lines of: 'No, Jeremy, you don't need DevonTHINK; just read the documents, take notes on them, put those notes, processed appropriately with citations, into a Zettelkasten and find links and add tags manually. Now go away and do some work...'

Thanks, all. Stay well.
Jeremy

Comments

  • Much as I want to support The Archive, I'm at this point leaning toward sticking with just DEVONthink, with a separate DT "database" as my Zettelkasten. I understand why DT's search and association features can't substitute for manual linking, but I don't see why they can't complement it, or how they'd get in the way of also building a web of manual links. Meanwhile it seems like having source materials and my Zettel notes in the same application (while still keeping the ZK as a distinct entity) yields a smoother workflow. Even apart from that, DT handles most of the overhead associated with ensuring notes have unique IDs and URLs for me. I guess that's the opposite of what you say you'd love to hear....

  • @Eurobubba said:
    Meanwhile it seems like having source materials and my Zettel notes in the same application (while still keeping the ZK as a distinct entity) yields a smoother workflow.

    Did you try it? Can you give us more insides into the actual work flow?

    I have two issues with this

    1. To me, it is mandatory to have only my thoughts and ideas in my archive.
    2. I stick with Luhmanns principle: Never touch a text twice. I cannot avoid it 100% of the time. But if I have to go to a book or article a second time, I consider the first time faulty. This is not true with text you have to go over and over again (The Iliad as a historian for example or core text of the field come to my mind).
  • Thank you, both.

    I've lost count of the amount of times I've started reading something, and then thought: 'Hang on - I've already read this and made notes. Now where are they...?' My implementation of the Zettelkasten method is precisely to try and avoid this issue: of ineffective reading. I worry that the DevonTHINK approach makes us too susceptible to the 'collectors' paradox', i.e. I'll put the text in, file it away in my memory as 'read and processed' and then promptly forget the whole thing... Having said that, I don't really know that much about DT's capabilities as I haven't yet given it a try. As Sascha says above, I'd love to hear me about your workflow, Eurobubba.

    Thanks again.

  • And yes, it's DEVONThink, not DevonTHINK, isn't it. Apologies. No-one likes a shouter...

  • Well, my workflow is still evolving, so I don't have a lot of insights to offer at the moment on that score. With regard to @sfast 's issues, I agree with 1), which is why I've set up a separate database called Zettelkasten within DEVONthink (the developer spells it with a lower-case 't', since we're being careful about that). As for 2), my counterpoint would be that at any given moment, not only is it not necessarily worth the time investment to thoroughly squeeze every conceivable bit of value from every source document, but I also can never be sure a given document won't take on new significance at some point in the future when I've acquired more information and/or further developed my own ideas. Plus I'm a bit of a data hoarder, so I like to file information away where it doesn't get in my way, but I can still retrieve it easily if I feel the need.

  • Well, @jemscott, No, Jeremy, you don't need DevonTHINK; just read the documents, take notes on them, put those notes, processed appropriately with citations, into a Zettelkasten and find links and add tags manually. Now go away and do some work...

    Seriously, though, I'm asking Marko (author of the DEVONthink review) to chime in with some info.

    Using DT as a more sophisticated literature and reference manager sounds like a beneficial use to find files in large collections again. If you need to search the full-text much later. Which, of course, is a bad sign in terms of "have I processed this PDF file fully?" But not all work is about processing files; I'm not a historian, but historians using DT seem to put tons of stuff into their databases at lightning speed to build a reference file.

    Back in the day, DT could index an external folder but then could not keep up with the changes automatically. I don't know if that limitation is still in effect. In any case, the semantic search suggestions were interesting inasmuch as they were different from the search results nvALT brought up, but I didn't have a project back then and couldn't evaluate if this difference was making a difference (pardon the Bateson pun :)), i.e. if this is actually producing better results.

    The thing with any artificial intelligence is that it's sufficiently complex to be opaque to humans (at first), and thus becomes an interesting puzzle. Even if you have a knack for backwards-engineering an algorithm to find out which are its triggers, "semantic search" will always be ahead of your ability to foresee the results. It's the Zettelkasten's effect of surprise, but sooner, because the algorithm does its job in its own way automatically, all the time, while the manual approach produces more surprise if you put in more work. -- In short, I think the feature of semantic search is always interesting because it's surprising and non-deterministic (for us). This satisfactory emotion may not be the best of guides, though.

    Author at Zettelkasten.de • http://christiantietze.de/

  • edited October 2017

    @Eurobubba Do you have a specific example of a text you had to go back to? I am interested in those use cases because in my experience they are interesting edge cases.

    I have it sometimes, too (rarely though). But then I don't need any sophisticated thing. I know that I have to go back to it. Then I just pull it from my library or download it again (pubmed and google are to this date always sufficient enough).

    @Unsureness: You can be never sure. My easy solution is: Hakuna Matata. :smile: You can be more productive if I don't give a shit.

  • Thanks for all the comments, people. Hopefully Marko will have something to contribute on this too, but this idea of 'if you have to go back then you didn't read it properly' appeals to my ascetic side. I'd love to know what the so-called 'semantic searches' can contribute to the ZK method that 'manual' searching, linking and tagging can't. Just don't yet know enough about the former to be able to comment.

  • @sfast There are a number of books that I've read multiple times and will probably read again. Zen and the Art of Motorcycle Maintenance and George Leonard's Mastery (not to be confused with Robert Greene's book of the same title) come to mind. If I'd been taking Zettel notes while reading them, I'm pretty sure I could have kept generating fresh ones each time.

  • What do you think is the reason?

    And how would DT help you?

  • @sfast
    Are you asking me why I don't think I can always exhaustively analyze a text with no need ever to return to it? I suppose because both my own mind and my mental environment are constantly evolving, so it's a different "me" encountering the text each time.

    As for DT, I want to be able to use its automated features as well as its ability to handle document types other than plain text. And since I'll be using it anyway and it also seems to be perfectly capable of providing the features and workflow I need to cultivate a ZK, I'm not really seeing any payoff to the overhead of adding another application to the mix.

  • Ah, ok. I think that should be true for all of us. On the other hand, I have the feeling that my time is spend more effective on new books than rereading. The exceptions are of course books that you want to know really, really well.

    By "feeling" I mean I measured my output.

    DT

    I still don't see the benefit of DT. :smiley: But I am curious how you piece it all together. (Hopefully, you will share your workflow with us)

  • edited November 2017

    @sfast said:

    I still don't see the benefit of DT. :smiley: But I am curious how you piece it all together. (Hopefully, you will share your workflow with us)

    You know, until now I hadn't really been thinking in terms of developing a "workflow" worth formalizing, describing, and sharing. But now that you've asked, doing that does sort of fit with some of the other ideas we've been thrashing around here. I'll take that on as part of the process of developing my personal ZK itself. No promises that I'll have anything worth sharing any time soon, though!

    And I think I've already said it, but just to be clear, none of my comments about DT are meant to discourage either development or use of The Archive!

  • Looking forward to it -- and don't worry! This is a honest place to discuss knowledge work, not a cheap advertising space for The Archive :) Even if somebody tore down the app with good arguments and feedback, that'd be a great opportunity to improve it, after all!

    Author at Zettelkasten.de • http://christiantietze.de/

  • Just to drop a link to an article of Gabe Weatherhead about DEVONthink and, above all, the organization of textual notes... some tips are remarkable:

    http://www.macdrifter.com/2017/03/my-text-corpus-in-2017.html

    • Marco.
  • @mrcmrc Thanks for this. Very interesting stuff.

  • I've used a number of systems over the years, but DT is probably the easiest. By that I mean to use and to replicate to other computers you need the data on.

    DT Setup
    1. Create a separate database.
    2. Use Rich text
    3. Create a template for your zettels so you maintain consistency
    4. Use tags
    5. Done

    Let me explain some of these. DT functionality is actually focused on rich text. If you use rich text you can set DT to automatically link to notes of the same title as your text. This is great when you write a title and realise you already have a note with that name because it turns into a link. It also speeds up linking.

    DT's search is phenomenal and this alone is great for a ZettelKasten. You can however, follow note links and just move from note to note. Entering notes is dead easy and so is getting you information out should you decide to move it elsewhere. DT OCRs pictures which means that notes scanned in are also searchable. Finally DT's ability to encrypt your data and share it with another DT installation on another is invaluable for having all your zettels on all your machines, even iOS.

    As an aside, I'm not on the markdown bandwagon. Markdown was born as a text-to-html tool. It's point was to give you rich text. If I'm in an environment that already has rich text why do I need Markdown? Computers are more powerful, disk space larger, I see little value in Markdown except for its specified purpose. I'd agree with Sascha about smartphones. I use a macair for mobile and imac for my main machine. I use a Galaxy Note 8 so I can handwrite my notes on the screen at any moment I need to. Rich text is what it says, "Rich". Plain text is, well, "plain". When I write a note I can change fonts, sizes of words, word colour, highlight colour, and so forth. This conveys far more information than any plain text note.

    I would also disagree with Sascha'a point 2. (Or Luhmann). Knowledge is not static, even though it might be written on a page. Knowledge is dependent on our current level of understanding and frame of mind. If I read a book on Quantum Physics, I would understand very little. If I studied Quantum Physics and then returned and went back to the book I would learn more. As I grow in knowledge, going back to a book yields more fruit. I'm a great advocate of marking up a book correctly and creating an index in the book of key themes/topics/ideas/questions. This can be repeated with each reading of the book, yielding more and more fruit. A note may well not be able to hold the full context or I'd be rewriting the book. Hence we return to the book periodically. I also think that saying, "never return to a text twice", is saying, I've read it once and now understand everything. This is simply a false premise as, "The Student, the Fish, and Agassiz" would bear out.

  • I used DTP for a while. It's powerful and easy to get stuff into. In the end, I didn't like having my stuff only accessible from DTP, and the Mobile experience (4 years ago) was so bad, I couldn't do it. I wanted to access my stuff anywhere. So now I follow the plain text approach, as @ctietze and @sfast advocate. Still, if you want the powerful search and "See also", you can index folders that live outside DTP, AFAIK. So I index my ZK with DTP, and if I really want to get those other features, they're there.

    As for reference material (PDFS, books, other docs) - those need to live somewhere, yes. One could make an argument for DTP here, but I like having my stuff on the filesystem. I just expend a little more effort now in organizing. For scholarly articles, I let Papers3.app do the work. For everything else, I make shallow folders and use Spotlight/Alfred to get around. Again, if you wanted somehow to use DTP's features, you could index a folder with PDFs or something in them.

    This article influenced my approach to DTP a while back:

    You get the benefits of DTP without having to commit things to living inside of it.

  • I have a similar system that I have been using for 10 years.

    20180120-category-tag-description.extension

    I can find most things in macOS. The reason for DT is the advanced search inside files.

    DT to Go is nothing like the old app was. It is every bit as good as DT although it doesn't carry every feature and it's being actively developed.

  • @svsmailus said:

    DT to Go is nothing like the old app was. It is every bit as good as DT

    Agree!

  • @svsmailus said:
    I have a similar system that I have been using for 10 years.

    20180120-category-tag-description.extension

    I can find most things in macOS. The reason for DT is the advanced search inside files.

    DT to Go is nothing like the old app was. It is every bit as good as DT although it doesn't carry every feature and it's being actively developed.

    Glad to here DT to Go is better now.

    The search tools of DTP are advantageous and useful when you really need them. I just don't see a reason to have to create or keep stuff in DTP when you can get its benefits with things that live outside. The Indexing function does that for me.

  • @svsmailus said:
    I would also disagree with Sascha'a point 2. (Or Luhmann). Knowledge is not static, even though it might be written on a page. Knowledge is dependent on our current level of understanding and frame of mind. If I read a book on Quantum Physics, I would understand very little. If I studied Quantum Physics and then returned and went back to the book I would learn more. As I grow in knowledge, going back to a book yields more fruit. I'm a great advocate of marking up a book correctly and creating an index in the book of key themes/topics/ideas/questions. This can be repeated with each reading of the book, yielding more and more fruit. A note may well not be able to hold the full context or I'd be rewriting the book. Hence we return to the book periodically. I also think that saying, "never return to a text twice", is saying, I've read it once and now understand everything. This is simply a false premise as, "The Student, the Fish, and Agassiz" would bear out.

    This is the weak spot in your argument. I totally agree with you with that example. I explicitely wrote that:

    This is not true with text you have to go over and over again (The Iliad as a historian for example or core text of the field come to my mind)

    Core textbooks in your field or things you have to work your way through are other examples of this. However, most of the books and texts in general do not fall in this category for most people.

    The main point is not that you cannot learn anything more. The main point is the opportunity cost of milking a book you already read once (or even more).

    For example: I have read a book called Digitale Demenz by Manfred Spitzer. (tranlated to: Digital Dementia -- a book on the dangers and pitfalls in the field of media). It was a very fine read and I went through most recources, made a ton of notes and learned a lot. I am pretty sure that I could learn more if I read this book again. But when I read it again, I have the opportunity cost of not reading another book.

    I am not in the business of being an expert on Manfred Spitzer. :smile:

    Also: I don't expect to stand on the false premise. Books are not fishes. Books (and other texts) are a look but not the thing to look at. Or as the great philosopher Bruce Lee would say:

    Its like a finger pointing away to the moon. Dont concentrate on the finger or you will miss all that heavenly glory.

    Therefore, I humbly restate my ideal:

    But if I have to go to a book or article a second time, I consider the first time faulty.

    It is an ongoing learning loop with (hopefuly) less mistake frequency.


    Summary:

    1. There are opportunity costs to re-reading a book.
    2. Books (and other texts) are not important as individuals. Important is your object of interest.
  • @sfast said:
    (Hopefully, you will share your workflow with us)

    I still haven't really developed anything like a coherent workflow worth sharing, but it's worth noting that I'm finding it convenient to keep
    * an inbox of materials that haven't been processed into Zettels yet,
    * bibliographic references (I'm not an academic, and the bells and whistles of a formal citation management system would just get in my way), and
    * the Zettelkasten archive proper
    all in separate folders in DEVONthink rather than in separate systems.

    I do take the point that Sascha and Christian have made about the longevity of plain text compared to proprietary app formats, so I do use plain text format within DT for my Zettels. Someday soon I'll get around to writing an AppleScript to grab a link to the underlying filesystem file to supplement the internal x-devonthink-item:// link so I don't lose all my interconnections if DT is ever discontinued.

  • I cannot recommend the underlying file system file link approach: DT optimizes its internal note database by splitting stuff up into UUID'ed sub-directories like 9B4D1D09-6542-429A-9FE8-EDD7444D387D (last time I looked), which may not even be permanent, although I'm not sure about this last point. A regularly run plain text export would be more useful than tying yourself closer to this arbitrary, computer-optimized structure.

    Author at Zettelkasten.de • http://christiantietze.de/

  • Current state of my ZK "workflow": revisiting The Archive as Zettel repository, but also indexing the folder in DT. I've always been a both-and kind of guy! :)

    You've increasingly convinced me that a UID within the note text is the safest way to keep the note permanently retrievable and linkable, regardless of the application I'm using the read it.

  • It is only a matter of time until you fully accept the dark side. :smile:

  • Well, I'm starting to see potential uses for UIDs hiding everywhere in plain sight, anyway....

  • edited March 27

    @Eurobubba said:
    Well, I'm starting to see potential uses for UIDs hiding everywhere in plain sight, anyway....

    Dumb question: Is there any special method to generate a UID? I just create my own, and then embed it as a text string in the document and in all documents that refer to it. Typically I use "ID=jones2016-ioet" random code.
    I guess I should also put this at the end of the file name?
    BTW, I use EagleFiler to hold my 10,000 documents, but am moving toward DTP for analysis.
    And for tags in programs that don't have quite what I want, I use @##keyword. For example, I have a giant outline (started many years ago; now I would use something else). On a relevant section I would add @##WW2 .

  • edited March 27

    @Reegor said:
    Dumb question: Is there any special method to generate a UID?

    Outside The Archive, if I need one, I have a Keyboard Maestro macro or TextExpander snippet set up to type or paste the result of uuidgen.

Sign In or Register to comment.