An eternal question regarding DEVONthink integration: TinderBox vs. Curio

Greetings to the dynamic community of DEVONthink! This is my first post on this community – I have read a lot of what’s here, but please be patient if I am asking things that are already addressed elsewhere. Essentially, I am wondering how my use case (developing a dissertation on architectural history) would best mobilize DEVONthink and Bookends in conjunction with either (a) TinderBox, or (b) Curio.

(I would like to flag @beck and @kseggleton here, whose various materials and “Geekery” on their respective websites have been fascinating and, truly, helpful: thank you both.)

I am in that position right now of, using DEVONthink with the trial of both Tinderbox and Curio over the past month, and hesitating to commit to either of the two latter software. As I mentioned at the beginning, I am starting up a PhD project in architectural history. My project spreads across several periods and geographies, focussing on questions of connectivity. (Specifically, I am looking at the reception of a more syncretic [i.e. beyond Greek and Roman] antiquity at stake within nineteenth-century European architecture. Specifically looking at the reception of global archaeology, anthropology, and geology as forms of “deep history," which are (I am proposing) accounted for as a more “global” classicism within certain architectural texts and buildings produced in C19 France and Germany.) So it’s all about systems of connectivity, exchange, and portability.

TinderBox seems to have the long-term possibility of producing a synthetically networked repository of these inquiries into the movements of objects and patterns across geographies. But, at the moment, I do not understand TinderBox! However, what seems clear is that it has an extremely fine-grained ability to sort and tag and display timelines and mind maps and different views. I am a (hopelessly) visual thinker, so the potential to not only visual connections between objects and ideas, but to reframe the materials in terms of different relationships, seems extremely useful for my project.

This visual dimension is what attracts me, likewise, to Curio. (I saw in a Forum post that @kseggleton has chosen Curio over Tinderbox.) To me, Curio has the advantage of being more attractive to look at; more user-friendly, and, ultimately, more efficient to use, in the sense of not having to constantly deal with maintenance and tinkering in quite the same way as seems (??) to be necessary with Tinderbox. And yet: I sense that that Curio’s emphasis on individual pages (even with links) would perhaps not produce quite the same result of a fractal representation of exchanges, which is so appropriate to my project.

And BTW, I am posting this here because I have the sense that DEVONthink needs to be in the picture, and presumably users here have experience working across DEVONthink and either TinderBox or Curio. (And, admittedly, I wanted to address both @beck and @kseggleton, and they are both users of this particular forum, but they are not both members of either the TinderBox or Curio fora). If this seems to anyone more like a TinderBox question, then please say so and I will post over there instead.

Thank you so much for any thoughts and direction you can offer here! Especially links to reading or just any kind of advice would be great.

All my best,


These are my own personal views, and others will have their own.

The developer of Tinderbox has written about the concept of “incremental formalisation”. This basically means that you put material into Tinderbox and gradually add metadata, like tags, keywords, user attributes, and so forth, as you come to understand the patterns and associations between individual notes, or collections of notes. The program is extremely flexible, and any Tinderbox file is capable of growing greatly in complexity as you work with it. In my view, it is a tool for meaning-making or discovery. It is a tool for analysing and understanding your data. It is not a tool for merely visualising your data, though that may be a useful way of working with it. On the Tinderbox forums, James Fallows commented that one of the strengths of Tinderbox is in making associations, and that this works better than creating hierarchies in the program. I found this a valuable insight. I would contend that Tinderbox is not a mind-mapping tool (at least not in the classic Tony Buzan style), partly because it will do so much more than that.

I have not used Curio for a long time, though I like some aspects of the program, but to me it is more about presentation than analysis. It is also more limited in some ways, because you pretty much have to work with the tools you are given by the software. Tinderbox is much more difficult to get to grips with partly because it expects you to invent your own ways of interacting with your data. If you want to find every mention of Doric columns in your file, colour the background of the note blue, and arrange the notes in chronological order on a map, you can probably do that. But at the same time you could have another tab that displays mentions of Doric columns together with mentions of Palladian mansions.

Tinderbox is pretty open-ended, which I think is one of the reasons why people find it mystifying, because they are used to software presenting them with ready-made solutions, rather like an Ikea cabinet. Instead, Tinderbox says “I see you have a pile of wood, well here is a set of tools like hammers, nails, chisels, screwdrivers, etc. What would you like to make?” Tinderbox will probably allow you to do things with your data that no other program will, but you have to be prepared to work to learn what is in the toolbox, what it can do, and how to use it. If you want a quick solution, it is probably not the way to go. But if you want something that will adapt as you come to understand your data, it is a good option.


I use all three apps (DT3, Tinderbox, and Curio 12). In addition to the apt analogy from @mbbntu, I would add that “data lock-in” is something to be considered.

I started using Tinderbox when it was introduced on classic Mac OS. All of my Tinderbox documents from that long ago are still accessible. (And yes, I have some that I have been using regularly since 2002.) If I ever did want to leave Tinderbox, the underlying xml of the files can be migrated (albeit with some effort).

I have also used Curio for a few projects. To me, those projects are trapped inside that app. Don’t get me wrong, I like Curio, and I found it valuable for presenting ideas to others thanks to its visual style and flexible data types. But as a long-term repository of ongoing research data, it doesn’t work for me.

1 Like

Devonthink is the ultimate zettlekasten, if you’re familiar with that term.

Curio has become my go-to outliner because I like the idea of an outline in the Organizer (the thing in Curio that looks a little like Scrivener’s Binder) with what amounts to a corkboard for each topic.

I made a few styles and stencils (templates) a while back. My idea spaces (pages, in OneNote lingo) look like they were written by somebody who cares, and it doesn’t take extra time to create them that way.

If I had a large project with a lot of supporting info, I’d use Devonthink. For outlining, I’d probably still use Curio, and you can hyperlink between the two for references.


There is a very long thread here about using Tinderbox and DEVONthink together Might be useful to you.


The general thrust of what you’ve said (“The program is extremely flexible, and any Tinderbox file is capable of growing greatly in complexity as you work with it.”) makes a lot of sense to me. But would you mind explaining what you mean by associations being better than hierarchies in the program? I tried to find the original comment by James Fallows, but wasn’t able to track it down.

Otherwise, you’ve made a very compelling argument for the “sand-box” quality of Curio vs. the more expansive “English garden” quality of Tinderbox. My sense at this point is that, the tinkering itself is part of the thinking process – perhaps better name for the software would be Tinkerbox. It seems that the people for whom the tool works are those for whom tinkering is a form of thinking. that is certainly the case for me. What it seems to give me is the capacity to externalize relationships which would otherwise remain ambiguous mentally - and through that formalization, which takes place in the tinkering, new relationships emerge. I realize I’m describing mind-mapping, but I do understand how eventually the other layers and “interactions” you describe will emerge. but even now in a kind of “Flatland” Tinderbox that I am currently operating in, the tinkering process is giving me something.

That’s my takeaway from what you’ve said - your response was enough to compel me to close the door on Curio, for better or for worse – and I’m pretty sure better. (But don’t worry: I’ll be sure to DM you some vitriol in a year if I find myself down a rabbit hole and wanting to get out! JK)

1 Like

I think the thing is for me, that I was perhaps exaggerating a bit the extent to which I am already using DevonThink. For me, being very visual, I think Curio/Tinderbox would need to be the primary residence, rather than the vacation home. DevonThink to me feels powerful for the integration of files, and for its inboxing, and honestly even just the OCR capability is amazing, but I find it hard to imagine it becoming my primary interface, just because I’m so dang visual… Given that distinction (which I neglected to explain in my initial post) I think that your consideration of Curio as mainly good for outlining smaller documents as an annex to DevonThink, also compels me (like mbbntu) to lean toward TinderBox, since I prefer to have that system be the primary interface.
(let me know if I’ve misunderstood)

Yeah that is a super red flag for me. I was wondering about migrating, in relation to those two: thank you for that. It does seem clear from the responses here that Curio is not really going to be viable for my use cases. Thank you!!

It seems clear from the comments that the post there makes sense, but I cannot understand it at all. (I’m that much of a TB neophyte.) However, I have bookmarked it, and hopefully one day in the future a bell will go off in my memory and I will find the bookmark and implement the TB --> DTP link that’s being discussed in the thread. Thank you!

Yes, that link is more “for later” rather than now. But it does illustrate that the two programs can talk to each other, which is useful to some people, and might influence choices.

I would suggest that you get to know the Tinderbox forum. The people there are very helpful. Some of them happily write code for others to achieve things you didn’t know were possible.

To try and describe how I see things (personal view again) a Tinderbox note has basically three elements: its name, the text of the note, and attributes (meta-data). The key thing, to me, is that you can make your own User Attributes. So you could, for example, make a User Attribute that is “Type_of_Building” and put in information like “domestic, public, religious, railway station” or whatever. That allows you to “surface” notes that fill certain criteria that you have decided are important in your work. And you can add about as many User Attributes as you like. I don’t think of this as tinkering, but building. You start out small and add things as you begin to see their importance. And one important aspect of this is that it is additive. If you need another attribute you can put it in, but the other ones are still there. If you put something in a folder because that is where you think it belongs, moving it to another place destroys the information that you had before (if you see what I mean). Moving items from one folder to another is tinkering in my view, and not always that helpful. (Though of course DEVONthink allows you to replicate files to other folders, which achieves something of the same.)

The comment by James Fallows is in another rather long thread: You may not understand much of it now, but it could be useful later.

Best of luck!


I originally used Tinderbox as my visual organiser of notes, but it became clear to me that I was expending a lot of energy in ‘fiddling’ and trying to understand the complexities of the program. I use Curio as the visual frontend of DEVONthink. All of my notes live in DEVONthink and I will pull notes from it into Curio as I create the visual linkages between notes. I find that the wiki linking feature of DEVONthink works in Curio. For example if I create a wiki link to another note, in Markdown, and then copy the HTML preview of that note into Curio, the link is active and I can navigate this deeper layer within Curio. I create reference links in Curio to documents and resources stored in DEVONthink. The other thing that I use Curio for is my project Dashboard with icons (with embedded URLs) linking to my task manager, or the Curio idea space containing outlines of my writing, or the associated DEVONthink group. I used Curio for my PhD, for each chapter I created idea spaces that outlined the chapter contents. For my data analysis (qualitative - using a grounded theory approach) I coded my transcripts in Curio and then created mindmaps that abstracted the codes into themes. I guess everyone has a different way of working with lots of data and for me Curio added a deep visual way of processing data.


This thread is a bit like watching a tennis match. I love reading the depth at which people understand and use these programs. They are, all three, exemplars in my book.

As for Tinderbox, @mbbntu’s assessment resonates with my experience. And as for @kseggleton’s Curio use case, it seems a brilliant way to push Curio to its most useful edge. I used Curio while writing my dissertation proposal and found it to be just the kind of space I needed to hold, and in someways develop, my writing strategy. Noteworthy is that I tried to use TBX first and found it too fiddly and Curio was just right for that particular use, especially given the integration between DT3 and Curio. For me, Curio is a sort of magical, virtual white-boarding space where I can chart out concepts, especially involving visual media. Seeing them all in one place, moving them around, and annotating them creates a gestalt and helps me make sense of the concepts.

I see Tinderbox as a workshop, in @mbbntu’s words, to build understanding. For example, if I’m reading a tough article, no better tool than a .tbx file to pull annotations from DT3 as I read and visually map them, link, recombine, and iterate on my evolving understanding. Another use case: I’m prototyping a qualitative data analysis tool in TBX, so that I can explode transcripts, zip link important passages into their own notes, pass those through smart adornments to give them various metadata, and sort them in agents later on, all the while retaining linkages to the original transcript. When I started using Tinderbox in a sort of “flatland” way (nice phrase), I never thought I’d know enough about the tool’s capabilities much less have enough programmatic knowledge to pull something like this off, but after a year or more of use, turns out I do.

So, ultimately, they are distinct tools, with some superficial overlaps, and I think they both warrant use in the academic’s arsenal. The depth at which one can use TBX is somewhat in opposition to the inertia you must overcome to actually use it, but it’s entirely worth it in my view.

I should note that I know Tinderbox much better than I do Curio, so I am likely shortsighted in this comparison.


I will only say that my view of Tinderbox is not positive. It was (very) expensive, it crashes (very) frequently and it demands from the user a lot of time to make it useful. It can be very powerful when you put in the time to master it, I don’t deny that, but I just can’t bring myself to keep spending time with it. (Something about the XML/RTF and HTML mixture just doesn’t feel right. Seems to be the reason why it is so riddled with bugs).


Hmm, I’ve been using Tinderbox for many years and I can probably count the total number of times it has crashed on one hand. But if you have a repeatable case, please report it to Eastgate. But I do agree that it requires you to invest time to reap its true rewards. (As does DevonTHINK.)

I have both Tinderbox and Curio. Tinderbox is like a Swiss army knife – it’s amazing what you can do with it. Curio is easy to use and can be useful – if you try the trial version you will get a good idea if it is what you need. On the other hand, the more you use Tinderbox, the more you understand its capabilities. You might take a look at and try the Tinderbox Zettelkasten program that has been uploaded there. It’s pretty impressive. Note that you can paste entries from BookEnds into that program.

Since 2018, I have exchanged at least two dozen emails with Mark. He was always very responsive and polite (so no complaints in this regard). At some point, however, my willingness to keep reporting stuff just ended. Crashes, bugs and glitches were, and still are with each new version, just way too frequent. It dawned on me that it is simply an unreliable software that mixes what should be a hypertext environment with loads of RTF text. So, in my view, the concept is really good, but the implementation is not.

Odd, I don’t have Tinderbox crashing issues at all and think the implementation is beyond amazing. All three of these apps have tremendously good coders.

1 Like

I use all three of these apps, Tinderbox, Curio, and DEVONthink every day, nearly continuously. I don’t have time at this moment to well-describe any of my working scenarios, because I’m procrastinating on grading as it is, but I’ll give a broad overview.

Any new project that comes up or new courses that need to be designed, the first thing I pull up is Tinderbox. It allows me to start creating notes about every little thing and begin to see the relationships between them. I can also toss in excerpts from peer-reviewed articles, reports, etc., that are all stored in DEVONthink, including the link in TBX to get to them in DT. I’ve done this for several things, including a risk assessment for Lakes Huron and Michigan, the building of an online course last fall semester, and a jump-off point for COVID-19 data analysis I’ve been doing in Datagraph.

I use Curio when I really want to organize what I’ve done in projects into visual representations and to manage a lot of it post-project, although it could also be used up front for planning using its mind-mapping, etc. For example, I created an online version of ENV105, Environment and Society, by planning everything out in TBX. I stored modified assignments, syllabi, schedules, etc., in DT and linked TBX to them as needed. My colleague and I would often have emails and phone calls about the project. For the emails, I sent the important ones into DT from Airmail and had a TBX folder “watch” them. That allowed me to look at the relationships in TB and make sure I didn’t overlook any changes that we discussed. When the course was ‘complete’ (are they ever?), I began managing the various sections in Curio, including taking notes about what was and wasn’t working from the original plan. I keep the enrollment lists there, drops/adds, my comments on discussion boards, etc., etc.

I also use Drafts (where I’m typing this), Mindnode, Agenda (actually extract from Blackboard for input into Curio using this; amazing sharing extension), and many others, including Hook. I’ve not really managed Hook into my workflow yet, but I’m working on it.


Like kaibo, this is also my first post. I am also working on qualitative academic research and have only recently become a user of DevonThink for this.
I’m posting my question to this thread because I think it follows on from the integration question and the visual thinking problem (also, I would love to have the links for the @beck and @kseggleton geekery you refer to up top.)
I do ethnographic style research, which results in a lot of pictures, sound and video, plus fieldnotes that are in very poor handwriting with sketches & other material that doesn’t OCR well. For me, rather than searchable text-based associations, I need to integrate metadata (especially the date/geolocation of a photo taken) and tags that I make as codes which ideally would be intra-file tags/links, so I could mark a region on a photo or PDF. For example, I have handwritten fieldnotes in PDF that correspond to a photo taken at the same time: it would be amazing to be able to link to the photo as a thumbnail embedded in the fieldnotes pdf; or to mark a region on a photo, a section of video, or a region in a fieldnotes PDF to tag/code, so I could review a tag collection later as its own document. It would be magical if that document could order chronologically based on metadata.
This is along the lines of what some CAQDAS software does (e.g. Atlas.TI or Trophy with tagging photo regions) but there are always caveats. As far as I can tell, I can’t tag internally on PDFs (in DT or in Atlas.TI) and, what DT offers especially is tags that are readable elsewhere (not so with Atlas.TI, without some manipulation as I understand). In any case, I think DT has more potential.
I’ve been looking through various posts here (especially here) and trying to experiment with DT to see how this might work. This thread makes me wonder if Tinderbox or Curio might be the shell layer through which I can work with the raw files in DT. Initially to me, it sounds like Curio might be a possibility, but I’m having some new-app-fatigue - I’ve tried Tagspaces, Eagle filer, and others before deciding to go with DT. Also, my ‘fiddling’ abilities are limited: I’ve gotten as far as trying out Markdown as a way to try to build this layer as a DT document, but didn’t get very far before semi-abandoning that route. I’m not afraid of coding but I have a limited amount of time to put into learning a system.
If anyone has suggestions or insight related to this, i would greatly appreciate it.

1 Like

hi Laru,
Based on what you’ve described, it sounds definitely analogous to what I have in mind! It sounds like you are doing something in anthropology, or anthropology-adjacent? I think you are actually a fair bit further ahead of me in developing a system for what you want to do with your information. So I am sorry to say that I cannot offer a more experienced perspective on your specific queries…*

Personally, I have not yet reached a decision about Tinderbox vs. Curio. I think there is definitely something to be said for all three, as @ChemBob says: he argued that Tinderbox and Curio are maybe too dissimilar to compare directly. Still figuring that out for myself though. Luckily I have some time over the summer to do so.

I do think new app fatigue and fiddling problems are a real thing: it feels like a barrier to Tinderbox for me, and that’s more or less the experience that @kseggleton described above. It feels like the app is designed for people with more fluency than “interface-only” folks like myself. From what I have read online, a lot of what people find valuable about the software is precisely it’s malleability, which doesn’t mean only the interface, but also the ability to “program” (?) various things, both within the software and in relation to other software. (In fact, the interface is sort of “thick,” in the sense that the “front and back” of the software somehow meet in the middle in a semi-user-friendly-programmable-jello space. It seems to be the blessing and the curse of this very particular software. I totally see the appeal, and aspire to gain that kind of fluency, but I’m not there yet. I wonder if it is common for new users to similarly “aspire” to gain fluency, but not manage to, or find it takes too long, and so on. I wonder if the highly active users are Tinderbox tend to be people who already had coding experience before encountering the software, since their learning curve would be much smoother. That’s all to say that, yeah, I think the fiddling and new software thing is 100% not a negligible factor in engaging with Tinderbox: seems like a huge learning curve.
As I said, I haven’t decided yet on which software to move forward with (or perhaps both), but I am leaning toward Tinderbox because it’s depth allows for me to use it primarily, with occasional recourse to what’s in DT. Meanwhile, with Curio, it seems that one has to work more actively between DT and Curio because the Curio is a kind of “skin” over DT. Extending that analogy, TB seems more “fleshy,” which works for me. (Again, as I mentioned above, I am super visual, and can think better in TB than in DT – if that weren’t the case, then I might find DT and Curio offer a tight and more-straightforward system.)

Beck Tench has uploaded some fabulous (and seemingly canonical) videos on ways to use TB for academic purposes:

Kyle Eggleton keeps a blog on “Geekery” which includes fantastic posts about this stuff:

Hope that helps a bit!

** In terms of the particulars of your post, I would be speculating, which is sort of useful when there are so many knowledgeable people around in the fora for DEVONthink, Tinderbox, Curio. I haven’t looked at the Bookends forum, but I believe that software integrates with all three, so folks there might have a perspective on your PDF stuff (even if your handwritten notes wouldn’t necessarily [?] belong in a reference manager).