Writers love having written. Problem is the time, work and brain twisting necessary between an idea to produce something and actually having done it. Well, it’s not that bad, sometimes you love writing, but sometimes you hate it. Or it bores, is cumbersome, and annoyingly laborious. This is why the human species loves to create machines: to enjoy the fruits of life, ransomed from the need to plug, wash and process them. With the field of information production, it’s about the same. The invention of computerized information processing has led to the rise of numerous attempts to create machines supporting human efforts of thinking, understanding, and creating meanings. In a sense and high on the abstraction layers, this is what computing is about in general. More narrowly, the question is how and which kind of software can support individuals in their efforts to gather information, grasp it, recombine it, and create new insights, new meanings, new information, new knowledge. What would be the equivalent of exoskeletons for the brains, which would enable the average brain to easily jump on the notorious shoulders of giants and beyond?
The difficulty for software producers is that in the end their code needs to represent the real world or create new worlds by using a rather rigid set of algorithms and functions. As any social scientist, psychologist and student of complex system knows, applying mathematical rigidity to the real world has its issues. The masters of ceteris paribus are just learning the hard way that the ‘other’ doesn’t disappear by excluding it from sophisticated, yet under-complex models. Could one imagine a more challenging field than to computerize human thinking and innovation? It is highly individualized, it differs from person to person, individuals have different strengths and habits depending on their surroundings, the time of the day, their moods, their intensities, their socialisation, their education. Human beings think in and with images, texts, sounds, narrations, dictations, scribblings, sketches and all the other techniques for informational computation humanity has invented in its short history. Individuals get ideas while watching movies, reading books, dozing, dreaming, running, listening to presentations, chatting, brainstorming and so on. That’s why the venerable new economy provided their hard-thinking employees with sofas, table soccer, thai massages and luxurious brunches. And it’s not by chance that distributed collaboration and flattened hierarchies are so en vogue in knowledge intensive work and innovation creating endeavours: knowledge and ingenuity are everything, but impossible to measure and calculate. The complexity and abundance of varieties of thinking and creating pose a challenge for creating digital support systems for these activities.
In the last twenty-five years or so, there have been numerous attempts by software producers to address one or more of these problems. We have thus seen hundreds of attempts to create software for retrieval and creation of information, visualisation, for storing, categoring, linking, annotating, writing, or referencing of information. Note-taking software and text editors were among the first programs developed for computers and are probably as old as computers with attached monitors are. Word-processing was dealt with right from the start of the PC boom. Reference management software has proliferated in the nineties and early naughties. The 2000s have seen mind-mapping software (developed by Terry Buzan, with whom securitization thinker Barry Buzan has co-authored a book on mind-mapping) popping up everywhere. The vanguard of the more interesting category of individual thinking supporting software probably was asksam, created in the mid-late eighties and in a slow decline since late nineties, a DOS- and then Windows-based full-text database, with which you could easily create a digital version of Niklas Luhmann’s famous Zettelkasten (youtube video, in German though) or any other kind of text-based databases.
Asksam also spearheaded the concept of tagging, though implicitly, the wildly claimed feature of Web 2.0 with its still fashionable and ever more ubiquitous tag clouds. As of now, we have many isles created by software imitating certain aspects of human text-based knowledge production, some of them interconnected with different degrees of tightness. What is certainly missing, is the kind of round-trip interconnectivity between visualisation/modelling software and text production software that has been around for code writing and software design for years. Visualized idea generation in the form of diagrams can be transformed into respectively structured text (code), just as changes in code result in automatically changed visual diagrams. The category of outliners has tried to fill this gap (and more), easing the organisation, sequencing and hierarchisation of text. But this is only one aspect of the lack of integration and interconnectivity between the isles of idea generation and information production.
The emerging cloudification of computing and the rise of web-based collaboration has shifted a bulk of the collective software developer mindset to collaboration, social production, sharing and opening personal silos. And yet, a big share of knowledge production happens individually. The web is full of fora discussing how to apply which software for which use case, how to integrate which software with which feature of another software, whether to digitalize certain steps in your workflow, or whether a paper-based approaches are superior for certain tasks. A lot of the innovation of software adaption is happening there, partly in hand with software producers.
For writers – whether novelists, business report creators, or academics – many tasks in text production are similar: collecting and gathering raw material (digital text, articles, pdfs, movies, links; photocopied artefacts, graphics etc.) related to general interests or to a specific project or task; annotating, categorizing and highlighting the raw material; taking notes, writing down ideas, mindmaps, sketches; write drafts, digest third party comments, create new versions, add bibliographic data (researchers only). All of this is happening jumping back and forth, chaotically, not sequentially. It’s the non-sequentiality that requires tight round-trip integration of as many aspects as possibly either within one software or a set of distinct applications. This integrated feature-set can, speaking in ideal types, be created by a network of distinct application or by a single, bloated piece of software.
There is one piece of software that is mentioned in every discussion about software for writers and researchers, albeit it’s Mac-only: the aptly named Devonthink application. To judge the usefulness of software, it is helpful to recall use-cases in which feature of the software can result in productivity gains for the user. A long list of feature doesn’t help you to be productive with a piece of software if it doesn’t fit your behavior and working style. And simplicity, though widely claimed, can turn out to be adverse if it misses to encompass features necessary for a certain task. Imagine – not that imaginative in this context, but you feel free to imagine any other topic – you would want to write down a couple of remarks about the individuals’ thinking supporting software in general and devonthink in particular. It’s a topic you’ve been interested in for while, you’ve read a few things about it on the internet, copied a few snippets on your local computer and filed them with more or less discipline and consistency.
Devonthink’s features make it a strong, the leading contender in the category of “everything bucket” applications, a term used not only for FBI database systems. (More on that in the next paragraph) The characterizing feature of “everything bucket” software is that it serves as the central repository of whatever information you would like to store on your computer. Whatever you get and see and consider to be worthy to remain on your computer: it goes straight into that bucket. As such, it serves as a sophisticated replacement for less feature-rich file-systems. File-systems historically lacked full-text search capabilities, a feature heftly missed by researchers until it finally became a commodity in the last couple years as Apple integrated Spotlight into it’s Mac OS X and Microsoft drew equal with Vista and Windows 7. Before that, user’s had to rely on third party systems like venerable dtsearch or the intrusive and talkative Google desktop search.
Twitter engineer Alex Payne argued against everything bucket applications, as they would deprive users to use external application which expect files to stay in the normal file system hierarchy and not within buckets. While he has a point to some extent, and Doug’s concept of the “File System Infobase Manager” is geekily charming (Doug draws on stunningly sophisticated debates in Scrivener’s user forums), and it might hold true for some of these bucket applications that: “By using Everything Bucket applications you give up functionality for compactness and eventually that equation works against your creative process. By working in the file system you use best in class apps for each specific purpose.” (ibd.) However, devonthink’s indexing feature does allow files to remain in the file system hierarchy; and even if a file resides inside Devonthink’s own databases, you can still use, say, Word to edit your documents. As far as I know, devonthink is the only consumer-level software that can both act as a file-system indexer and as a bucket at the same time.
One of the core arguments against “everything bucket” applications is the threat of lock-in, of non-exportability, of the potentially high costs of moving out and abandoning a software system in favour of another. Devonthink supports those who want to walk out, and thus lures those who want to sneak in: Mark the top level groups in your database, press alt-command-E and wait an hour or so until DT has shuffled the gigabytes of your database into a file system hierarchy that copies the group structure of your DT database. In addition, devonthink creates one file per group that contains meta-data related to the documents within that group. And out you are.
So, feeding the “everything bucket”. Sucking the more important sediments (thus ‘devon’ in devonthink) of your hard disk into devonthink, is one of the first things you will do after starting using the software. The procedures are described in dozens of other places on the web. More tricky is the question of data organization and structure. If you have multiple personaes, you might want to create several database, one for each, e.g. one related to research, one for administration of your business, one for your writings. Others prefer project oriented approach, but a one-database-for-all approach is also possible. Devonthink databases can contain gigabytes and still be functional and responsive. Devonthink imports virtually any file format on your hard disk. If you prefer your files to stay in their filesystem hierarchy, DT can alternatively just index them instead of importing.
The second way of feeding devonthink – a popular one among all the internet consuming researchers out there – is to store snippets cut out of web pages. PDFs from journals , news articles from media sites, blog entries from RSS readers, emails – press a keyboard shortcut to copy the respective document to DT or select the text that’s relevant to avoid the clutter of ad-loaded news articles. After you’ve pressed the shortcut, a semitransparent window will pop up where you can tag the information-piece and file it in a group. All in all, it works remarkably well and efficient and devonthink can probably rightfully be labelled as the hegemon amongst all “everything buckets” applications. It robust, stable, at times lightning, at times liveably fast. In certain cases, however, e.g. when you’re using fifteen application at the same time, among them memory junkies like a browser with fifteen tabs and an Adobe application and your database is a couple of gigabytes in size, Devonthink has to claim back a few hundreds of megabytes RAM. In such cases, waiting time for your search to be completed can easily amount to a dozen of seconds.
Devonthink truly shines, when you start harvesting the repository and digging in the sediments of what you’ve neatly filed or hectically thrown into it. It kind of resembles sneaking around in an old dusty library, looking for the magic gem, the eye-opening book written in the allegedly dark ages by an unknown monk. Umberto Eco described this approach so lucidly in his book on academic researching. Transferring this method to the new ages: As mentioned earlier, imagine you would want to write down a few ideas about individuals’ supporting text-based information production software. What I do in DT is this:
- I create a new group in in the “current activites” group of “general interest” database. Create a document that will contain your final article or first drafts. Just start writing or crawl through the database first.
- The second step is to create a subfolders for the, well, ‘raw material’ that I replicate (replicating in DT is creating a link to another database item instead of a physical copy) into the subfolder.
- Search database for documents and groups to link to the ‘material’ group.
- Start or continue writing.
After a while, you have written down a couple of ideas. Devonthink’s most unique feature surely is the “See also & group” button, iconized as an upside-down semi-filled stovepipe hat. This is the magic, devonthink does with my first draft of this review document.
Even if I hadn’t set up a project folder with its replicants, devonthink would have supplied me with a few dozens of worthwhile links in my database. Interestingly, most of the files I created links to in earlier paragraphs above are listed in the lower part of the window, whereas the upper half suggests a couple of relations that are somewhat oddish at first sight. And at second sight. And third sight. Or is that “machine memory creates ideas we’ve never considered” (Clive Thompson, A head for detail; see image above)
But what is the idea behind setting Karl Marx’s Kapital on number two on the list? Devonthink plays Google and keeps the algorithms close to its chest. What could it be? Identical words? This article’s language is, well, English (possibly plus denglish and anglish), Marxen’s “Kapital” is a German edition. The words devonthink rates as the most significant ones of both articles differ substantially: Grossejean, rearchitected, patinized, oligopolist vs. Grundeigentumsverhältnisse, Arbeitsprodukt, Fetischcharakter, Entwicklungsgrad.
This machine’s firewall software hasn’t yet logged any traffic from my machine to the internationale.org or fifthcolumn.net, making a political conspiracy unlikely. What then? I opened the “Kapital” in a second tab in this “devonthink review” window, it takes a minute or so for the document to open – the memory consuming “see also & group” drawer is open, which is used by DT to list and visualise statistical similarities between the current document and those in the database. The Kapital comes with some 376.000 words, giving devonthink a few words to crunch. But, still, I really don’t get the link between this review and the Kapital. Because of the “Oligopol”? Das Kapital, however, doesn’t contain the string ‘Oligopol’. Should DT know that a monopoly is the big sister of an oligopoly? Anyway, I’ll take it as joyful news that artificial intelligence isn’t inhumanely clever, yet.
As said, the lower area of the “see also” window lists many helpful and relevant articles: Stephen Berlin Johnson’s article on devonthink in the New York Times; “Devonthink and other Mac Apps for History and Humanities Research”; “Delve into Devonthink”; “An attic called Devonthink”; “A digital academic workflow”, and couple of other documents, some of which were already mentioned above. “See also” obviously works very well. So I can continues with the odd findings…
What about the second in the list, “Billmon on war”. Billmon (“War is to important to be left to the think tank nitwits”) was a blogger who portrayed and analyzed the preparation and later the implementation of the US engagement in Iraq. He was renowned for his stunning sources, and obviously had a terrific citation database with hundreds of sharp quotes that he managed to deliver on the spot. Again, devonthink doesn’t reveal the reason why “Billmon on war” (a document containing his articles on war) is on rank #2. You have to find the connection between war and devonthink. Do you see it? It’s speed and maneuverability. Military history and military strategy highlight the superiority of the aforementioned two qualities in many modern warfare scenarios. Both require sophisticated logistics and advanced, durable technology. If you’re at war with your brain because e.g. it’s too late after midnight, devonthink might give you a rescuing hand. Your Devonthink databases amass tons of pdfs, texts and writings from which it sends you off deep into unknown territory or over unknown bridges. This is exactly where you would want to go. If that doesn’t work out, take rank 4: Innovation studies tell you that it takes a bunch of failed efforts to for one major innovative leap to strive.
Next to its capabilities in gathering, searching and drafting information, Devonthink’s fourth major feature is its position within the network of information-support systems on individuals’ computers: it has managed to acquire the status of production the master hub, the fat node where all the remaining applications link to or are linked to. Devontechnologies has laid ground for this by redesigning the underlying architecture and increasing speed and scalability of devonthink. It remains to be seen, whether they will be able to capitalize on their hub status or whether devonthink’s hegemonial position in its software segment will be contested by new contenders in the long run. The biggest challenge for all these small-size Mac software shops will be lie in the contestation of the indie-devoper model by a more capital-intensive development approaches – which will only be a natural development caused by the increased size of the Mac market. The business models of Evernote or particlarly Mendeley shed a light on how the post-indie Mac world will look like and it might force the code-smiths behind devonthink to substantially change their approach to the market.
The territorialisation of the land of information-production support systems is for the most part still happening in distinct, distributed isles of application. With their maturation, the focus for desktop software in this area will in the years to come be to integrate functionalities and features across the borders of single applications. The second stream of innovation will continue to be in the area of collaboration and finding new ways and possibly new granularity in the degree of publicness or peer-level sharing of knowledge work and its pre-products. Looking back in the history of this area of computing is a akin to development in science in general: it’s sobering how little progress has been made and thrilling to see what has been achieved at the same time.