Mark Bernstein is chief scientist of Eastgate Systems, Inc. He’s been writing hypertexts and developing hypertext authoring software since the late 1980s. Mark is the creator of Tinderbox and other tools for thinking that “harness the power of the link.” In this conversation, we discuss thinking through connected notes.

Show notes

Show notes include Amazon affiliate links. We get a small commission for purchases made through these links.

Transcript

Jorge: Mark, welcome to the show.

Mark: Thank you. Great to be here.

Jorge: I’m very excited to be able to talk with you on the show. I am a big fan and advocate of Tinderbox, which is a product that you designed and produced and have been doing so for a long time. But I suspect that a lot of folks listening in might not be familiar with your work. How do you introduce yourself?

Mark: Well, broadly speaking, I write tools for writers. And by writers, I really mean thinkers: people who work with lots of information, both to understand it and to explain it to other people. Tinderbox is a tool for thinking, but no one really knows what that would be. So I call it a tool for notes, a way to visualize and analyze all your notes, not just today, but over the months or years that a big project is going to take.

Thinking with notes

Jorge: I’m interested in the relationships between notes and thinking. How do notes support thinking?

Mark: The Japanese hypertext scholar Kumiyo Nakakoji talks about amplified representational talkback, which is a general design phenomenon. An architect, an artist, or a writer puts something on paper and then looks at it. You look at it, and then it seems different from what you had in mind, and you either correct what you’ve written, or you see that what you’ve written is right and correct your bad idea. That kind of representational talkback is fundamental to all sorts of all creative processes, from the sciences to the arts.

Jorge: The concept sounds familiar to me in that I think that that’s also part of the design process, where you make a prototype, and the act of making the prototype and seeing it take form in the world informs your thinking about what the thing should be. There’s a feedback loop, right? And I guess what I’m hearing there is that notes serve that purpose in articulating our thinking onto… I’ll use the generic phrase: “the page?” somehow, it changes how we think about it.

Mark: Yes. Of course, it’s nice to have the permanent record; that can be very useful and, at minimum, saves you lots of time. But being able to look at an explicit representation of what you had in mind and see if this is actually right or not is a tremendously useful component of everything creative; I would argue everything difficult, where difficulty here is that you don’t know exactly what it is that you need to do.

Jorge: So the act of putting your thinking down in these more permanent representations can help guide you in some way?

Mark: Well, for example, you set out to write a book, and a book is a big thing. It’s going to take you months — probably a couple of years — to write. You’re going to consult lots of people and read lots of other books and think about lots of things. And until you start to actually write, you don’t really know what it is that you have to say. Once you have begun to write, you can revise. You can see that what you’re saying needs to be expanded or corrected. Roger Ebert used to say that all of writing is revision, but until you’ve written the thing, you can’t revise.

Emergent structure

Jorge: Folks, upon hearing the phrase “a tool for notes,” the image that might come to their mind are the sorts of notes that they might capture with a tool like Apple Notes, basically a digital version of the sort of pocket notebook that has been around for a long time. My sense as a longtime Tinderbox user is that that image significantly undersells the capabilities of a system like Tinderbox. These are not just notes, right?

Mark: Well, something like all Apple reminders are great for a line or two and for a fairly small number of notes. They’ll get you through your grocery list. And grocery lists and things of that size are actually amazingly tricky. And they’re quite important; I don’t want to minimize them. But again, if we’re setting out to write a book, study a subject, even a little paper…

I just did a little book chapter on where the ideas for the world wide web came from. And in fact, this ties into all sorts of important ideas in the arts and the literary world of the 20th century and ultimately to the experience of the Great War and the Depression and World War II. Now, this requires looking at lots of sources and lots of… well, a couple of languages and spanning about a hundred years. There’s a lot to get down. And organizing that sort of mess isn’t something you want to do with a list of notes in the Notes tool. There’s just too much.

And when you’re reading a particular source or thinking about a particular subject, it’s interesting now, but you don’t know how it’s going to fit into the big picture. You don’t know what the big answer is going to be. And so you need to work around. You need to work to figure out how the information actually matters to you. We call this emergent information; emergent structure. You don’t know in advance how everything fits together. That’s, in fact, the most important thing a student or a researcher needs to discover in the course of their work.

Jorge: I’m hearing two things here, and I’m going to reflect them back to you. One is this idea of representational talkback, putting your thoughts down into a tangible form where just the very act of capturing them in that way establishes this feedback loop. And then, I’m hearing something that might be part of that — but might also be different — which is, once the ideas are down, you somehow have to organize and find patterns. Are those different, or is it part of the same operation?

Representational talkback and finding patterns

Mark: Oh, well, they are different. Though, they’re different because they’re different levels of thinking; they’re different ways of thinking. They’re different problems, but you need to solve both of them. They’re similar in that at both levels; you want to reflect on what you’re doing. You say, “I see a way to structure the work I’m doing.” Or you say as a student, “I understand how what we’re studying now fits with what we were studying two months ago.” You’re a student; you may be wrong; you may have a partial answer; you may have a mistaken answer. So, you’re going to need to restructure. You can’t even really think about the structure until you have stated it in some way, but you also need to be able to restate it.

Jorge: So, there are at least two parts to this. One is putting down the idea somehow, and then the other is exploring the potential relationships and distinctions between the ideas.

Mark: Right. One of the first things that was discovered about building complicated technical hypertext is that you don’t know what the structure will be in advance. And as you’re adding information, you know you want to keep the information, but you frequently don’t know what the information you’re adding is. You can’t describe its type or its nature or its importance in advance. You just suspect that it’s going to be pertinent somehow. Or you see a terrific quotation that you know will be great to use, but you don’t know when that quotation will fit or even if it’ll fit in this book, or if you’ll have to save it for something else.

Finding ways to say, “I think these two things are related somehow, but I don’t want to commit myself yet as to exactly how,” turns out to be quite an interesting design problem. Hypertext people started out, in fact, by inventing the outliner very early — 1968. And outliners are terrific if you already know the structure of your information space. But hierarchies are not good if you’re just guessing about how things fit together because you tend to build great elaborate structures that turn out to be wrong, and you have to unbuild them, and then you’ve got a terrible pile on your desk.

Jorge: So this seems like an invitation to explore ambiguous spaces, is what it sounds like, yeah?

Mark: Yeah. I mean, idea space is always full of ambiguity, or at least the idea space that’s interesting to explore. There’s maybe less ambiguity in information spaces that you’ve mastered and know well, but you don’t need tools to work in those spaces.

What is hypertext?

Jorge: You brought up the word hypertext, and I understand what that means and why it’s coming up in this context. But I suspect that some folks listening in might not have heard the word before. How would you define hypertext, and how does it relate to a tool like Tinderbox?

Mark: Sure. Hypertext is writing that’s linked or connected. We are used to text that is either apprehendable at a glance, like our memo notes or our calendar entries, or that’s continuous; its sequence is fixed at the time that it comes out of the printer. That’s not always the best way to represent information. It is, for example, difficult to associate both a sequence of events and all the sources of those events the way you know those events happened.

So, people have been — for centuries — writing history as a dual narrative of a main story in the text and the story of how we found out and why we believe this is actually what happened though we weren’t there ourselves, in the footnotes. Footnotes, however, are clumsy. Editors hate them; readers hate them. They’re too limited. You start asking for footnotes in your footnotes, and the editors go ballistic. If you’re permitted, the readers start needing three hands to find the places, hold their places — multiple places in the book. So hypertext, now that we’re free from having to read on chopped trees, lets us think about textual structures that keep things connected in more complex ways than the scroll.

Jorge: I think that there’s a perception that… you used “scroll,” and I think a lot of people associate the written word with things like books — the codex format — which we have had for a while. Things like indices and tables of contents and page numbers do have some kind of non-linearity to them. But for the most part, we experience them linearly, right?

Mark: Right. And in fact, that non-linearity is good and important. Especially by the mid-20th century, people were finding that the information finding and cataloging of infrastructure that have been built starting in about the 11th century to handle this sort of meta information, to keep things connected, to connect things within books and between books and amongst books and libraries, was unsatisfactorily cumbersome.

Yes, you have indexes, but they’re really difficult and expensive to make. They often don’t list the very thing that you want, or they list the thing you want and give you so many hits that you can’t figure out which ones you actually want to consult. You end up running all over libraries, chasing down cryptic references to books that often aren’t there. And so, people started dreaming originally of tools that would make some of this footwork easier to do and also make it possible to connect things more abundantly since you aren’t asking people to go spend half an hour hunting through the library stacks.

The Tinderbox Way

Jorge: We didn’t cover this in the introduction, but besides being the designer and author of Tinderbox and other such tools, you’ve also written what I consider to be important academic papers in the space of hypertext. And you’re also the author of a book called The Tinderbox Way. And in The Tinderbox Way, you say at one point that Tinderbox, or the vision for the tool, emerged from “exasperation with the limitations of the first generation of PC writing tools.” And I’m wondering, what tools, what limitations, and whether those limitations are somehow related to this — I’m going to use the word “traditional” — way of authoring texts?

Mark: Well, what I had in mind, of course, was Microsoft Word and the various programs that it had triumphed over. A wonderful, huge piece of software for formatting — for making books — but not a particularly good piece of software for discovering structure and what you’ve written. It is even worse for discovering structure in what you would like to write.

Jorge: So, I’m thinking of the contrast, then, between a tool like Tinderbox and a tool like Microsoft Word. Microsoft Word might be a good tool to use for the final step, where you’re defining the presentation of the text somehow, whereas Tinderbox is better suited to earlier stages when you’re still trying to make sense of this miasma of ideas.

Mark: Exactly.

Jorge: In The Tinderbox Way, you also say that understanding is the true goal of Tinderbox. And I’m wondering how the sort of note-taking that we’re talking about here — this kind of hypertext note-taking — can drive understanding?

Mark: The fundamental problem of all research is to take all the information that we can gather that bears on a problem that we cannot yet solve and find a solution to the problem in front of us. That problem may be artistic. That problem may be technical. We may not yet know what sort of a problem it is or what sort of a solution we can contrive. But the first thing we need to do, of course, is to figure out what is the nature of the problem in front of us and what the constraints are.

For example, a friend of mine in architecture school was part of a class that was assigned to… everyone had to design an embassy on a certain lot in Washington, DC. And everyone had different ideas for great things in an embassy, a place for garden parties, an office for receptions, and so forth. One of the big constraints is that it wasn’t a very big lot, as Washington DC lots often are not. And one of the things that an embassy simply must have is a driveway into which you can drive a limousine and get the limousine out again.

So that constrains the turning radius of the driveway, which constrains the amount of the lot that you end up needing to spend on the driveway and what you’ve got left for the house. Lots of people left the driveway to last, and as a result, their house was simply too big to be used as an embassy. You couldn’t get diplomats into it. And that’s the sort of thing that happens more often than you expect. You find late in the process a constraint that changes the forces in front of you in which you had not given much thought to.

Jorge: Would it be fair to characterize the limo example as a case where getting deep into the problem’s constraints revealed a driving factor — sorry for the pun! — that would only be apparent bottom-up. Like, once you’ve actually started diving into the details.

Mark: Yes. All research… All significant research is, in some respects, bottom-up. There is no alternative. And so, the only research that you can do top-down entirely is research for which you already have the solution. You already know what you’re doing, and you’re just punching the time card. So yes, this is very much a bottom-up process.

Jorge: So, what I’m hearing is that the approach becomes one of getting as much of the stuff that, you know, down onto a — I’m going to use the word “surface” — but onto some kind of system that allows you to explore these bottom-up relationships to see if some kind of pattern emerges.

Mark: Exactly. And some connections are evident at the time you first make the note. For example, if you are interviewing subjects and they’re telling you things about events that they witnessed, you want to keep their observations about the events connected to the interviews and connected to their identity. You know from the beginning that if this is useful to you, there will come a day when you’re writing this account, and it may be five years from now, and you won’t remember exactly who it was who told you this unless you write that down and unless you keep the note about who they were and how you get in contact with them connected to the observation. So that’s a piece of structure that we want to create right now.

The fact that two people you interview may be connected in ways you didn’t expect may only emerge much later. And that’s terrific to have, and so you need to be able to move those interviews or those events around later to establish that newly discovered connection.

Object-oriented note-taking

Jorge: I just want to clarify: I think that in this case, the structure that you’re talking about is the fact that when you are interviewing a person, the minimum you know about them is that they are, in fact, a person. And a person can be described using several attributes. It might be that those attributes include their email address, their place of work, or whatever, right? And there’s a relationship here with object-oriented systems, right? Like, object-oriented programming systems, where you’re defining the object of a person, except that instead of doing it for some kind of software application code, you’re doing it for your notes.

Mark: Yeah. And I actually think the Agile revolution in software development is software development catching up to the fact that it’s a writer-ly art. Writers don’t know where they’re going or how they’re going to express it when they start out. Neither, it turns out, does software developers. They can pretend by writing it the first time in a spec language and then coding it and then, checking the specification, then finding out that they’ve written the wrong thing and writing a new specification. That was when I was getting started, the right way to write software.

None of us actually wrote software that way if we could help it, but we all knew that we were doing it wrong. And what Agile did was say, “No, this isn’t wrong! You have permission to do experimental writing in the IDE.” Now, we think that IDEs are for experimental writing, and we should use formal specifications for our books. Well, that’s nuts.

Jorge: Well, this goes back to the idea we were talking about earlier, the difference between bottom-up structure and top-down structure, right? The top-down structure presumes that you know the answers up front like you were saying. And for something as complex as developing software, you don’t. Right?

Mark: Exactly.

Jorge: I think in one of your writings, you compared the process of taking notes in this way to sketching, which I thought was illuminating.

I want to shift gears here because you used this example of doing research, and you said something along the lines of, “Well, you might want to share what you learned later.” And one of the intriguing things about Tinderbox is that even though we’ve been focusing the conversation so far on these kinds of early-stage, idea-exploration, disambiguation processes, Tinderbox also has facilities to actually share those ideas with the broader world. And I’m wondering about that phase of the process. Where does making — and here I mean making not just for your own sake, but making to share with others — where does that fit into the process here?

Sharing with others

Mark: Well, first of all, at some point, you may need to create a paper document or an electronic document that other people will read or a website that other people will read. One of the important lessons of the entire personal computer world is that you need to be able to have your text in a form where you can put it into new places and use it for new things.

So, from the beginning, Tinderbox has been very conscious of finding ways to work on other machines and on other media. From simply letting someone else look at your Tinderbox document and see all the different views of an information space so that they can either extend it or think about it to being able to export to selected pieces or the whole thing to text, to XML, to a PDF, or to a website.

Jorge: But that stage of the process, if I’m grokking it correctly, it comes much further downstream, when you have somehow gotten greater clarity around the structure of what you’re sharing.

Mark: Yes, though it may not be the end point. In the computer science world, we have a regrettable tendency to think about work as filling out forms or performing specified measurable tasks. But work isn’t really like that work involves studying and then creating a work product to use what you’ve studied and use what you’ve learned. Then you’re probably going to study some more, learn some more, if only from the reaction to what you’ve written. And so, a project document is likely to go on and for years. And this is why Tinderbox has so many different views and so many ways to look at the same document and to reorganize the document because however you thought you were going to organize it when you started the project, a year or two later, you will be wrong, and you will want to have a completely different organization. That’s what learning is about.

Information farming

Jorge: In one of your papers, you used the phrase “information farming.” The image of farming, in my mind, is powerful in that it does talk about something that is growing over time, and then it literally yields fruit, right? I was hoping you would unpack that for us. What did you mean by information farming?

Mark: Cathy Marshall at Xerox PARC originally started speaking about information gardening. She developed an early tool that’s the inspiration for the Tinderbox map view, in which you would have boxes but no lines. It was a spatial hypertext system, a system for connecting things by placing them near each other rather than drawing a line between them. Very interesting abstract representational problem, but also it turned out to be tremendously useful.

And so, dragging things around the map view is a terrific representational solution. Outlines… again, sometimes you want lists. Sometimes you want lists of lists, and there’s nothing better. You can also have charts and all sorts of other views. Kathy’s boxes on an infinite plane did look a little bit like garden plots. And I wrote a very early essay on information gardening, emphasizing that it’s not necessary to turn everything into a trivial outline or a seven-bullet point list. Sometimes you need to use indirection. Sometimes you need to surprise the reader. Sometimes you need to delight the reader, even if you’re doing technical writing. And so, you’re cultivating a garden space.

Information farming is intended to be more deliberate and more focused on gathering information that’s pertinent and important though you may not know how you’re going to use it and keeping it ready to hand so while that it’s organized and you can go find the right kind of apple when you want to bake a pie. And this is a helpful distinction from things like information mining, which were at once to be essentially the same problem. Information retrieval! Digging stuff out of a database or out of the worldwide web.

Jorge: The metaphors we use have an important framing effect, and I love these more organic metaphors around gardening and farming. I just have one more question for you, Mark. You’ve been working with and on hypertext for a long time, and I don’t know that it’s possible to speculate there can be counterfactuals on that, but I’m wondering how you expect that working in this way has changed the nature of your work or affected the nature of your work?

Mark: Oh! I think over the years, I’ve been more open to seeing influences and directions from disparate fields and disciplines, and schools of thought because this sort of hypertextual approach to working and organizing knowledge spaces makes it more difficult to simply discard what doesn’t seem to fit directly into what you had thought, or what you would expect it to do.

Sometimes this has not had entirely good professional consequences, but I think it’s the right way to work and is interesting. There’s an architect at Georgia Tech, Lars Spuybroek, who wrote a brilliant essay on Ruskinian aesthetics and their relation to digital computation and design, which I think is a perfect example of this sort of reasoning. Thinking about the ways that our concept of beauty and our concept of roughness can translate into the world of the digital. And also thinking how the ideas of arts and crafts work in a world of robots and CNC machines.

The answer Spuybroek comes up with is surprisingly optimistic: that where arts and crafts floundered on the cost of unique handcrafted work with the wonders and flaws of the individual craftsmen, in a world of 3D printing and computer-controlled blades, we can make every cut a little bit different and pick the ones that we like.

There’s a guy named Larry McCaffrey who is a potter in Cape Cod. And he has a building that’s filled with coffee cups like this one. And each one is the same, except each one, is a little bit different. Both because they’re handmade and because they’re full of stochastic processes with the glazes and the clay. So, even if you tried to make them all identical, the chaotic atmosphere of the kiln is going to change them. He doesn’t know which one is better.

I don’t know which one is better, but you look at a shelf full, and eventually, you find the one that’s better, and that’s the coffee cup you’re going to have for the next five years. That’s the future of a world of things that aren’t mass-produced and aren’t one size fits all.

Closing

Jorge: That’s a wonderful, wonderful image. And I want to thank you before we close the call. I want to thank you for your work, and in particular, I want to thank you for Tinderbox, which has been very important to me in my own career. It’s a tool… now that you’ve been saying it like that, it makes me think that it’s a tool to help you think for yourself and to look for the outliers. And it certainly helped me in that regard. Where can folks follow up with you, Mark?

Mark: Well, you can find Tinderbox at eastgate.com/Tinderbox, capital T. There’s a free demo, and there’s an amazing community of Tinderbox users from all sorts of disciplines, many of them very unexpected. We have one of the world’s experts on cranes frequently drops by our weekend meetups, for example.

Tinderbox, because it’s a big product and a deep product, can be frightening at first, in particular since it has to accommodate lots of people who are working in lots of fields and work in different ways. It’s got lots of menus and lots of options, and you don’t need all that stuff. You need the first few commands, the first few bits, and that’s all you need. And then, when you need more, well, it’s there, and you can look it up, or you can just ask people and often get answers in minutes.

Jorge: I’ve been using it since, I think, around 2004, and I’m still learning new things about it. It’s an incredibly powerful tool whose value and functionality unfold over time, as you’re saying. So I’m going to include that in the show notes. Thank you again, Mark, for being with us.

Mark: Thank you. It’s been fun!