Around 2011, I worked on a project for long-term archival of digitally captured cultural heritage. During the design phase, as we spoke with representatives from libraries and other cultural institutions, it became clear that we would have to bend on a principle we had previously held firm: non-destruction. The idea was simple—once archived, never deleted. To square that circle—satisfying our partners while still honoring the principle in a way that didn’t feel like lip service—our professor came up with the diplomatic formulation of “deletion under exceptional circumstances.”
Self-imposed constraints are hard to live by. External ones work much better. That’s why deadlines are so effective.
In the digital world, we often mimic the “real world” by reaching for metaphors that make systems feel familiar to users. Think “desktop” and “windows.” In that vein, blog software imitates print publishing by giving us a publish button—except that no matter how seriously we take it, deep down many of us know that the cost of “going back” to fix something minor is nowhere near as high as stopping the printing presses. The best of us tell ourselves it’s only a typo, and that’s fine. But from there it’s also only a tiny step to rephrasing a sentence slightly, and then—where do you draw the line?
In what follows, I will argue that we should, for once, fully embrace the reality of digital publishing—that there is no constraint requiring a “publish” metaphor. But first, let me steelman the other side. Nobody forces us to eat our veggies or go to bed on time. And still, we know it’s best to do so. Hitting the publish button and finally being able to forget about that one thing and move on to another can be very liberating. This is a matter of habit, and once the habit is established, it can be quite healthy—like eating our veggies. Relaxing that constraint might have its drawbacks.
London Big Ben Inner Clock Face 1070925-PSD [commons]
So, with my version of “this is not financial advice” out of the way, let’s move on. What can we do if we relax that “constraint” and assume that updating, in addition to publishing, becomes a first-class citizen?
-
Offering transparent work in progress. The writer Derrick Jensen has adopted the habit of making his book drafts publicly accessible while he is writing. In fact, you could argue that the same is true of all open source software—with the difference being that there is never a final publication event, since software is always a work in progress.
-
In the sense of open source development just mentioned, such an approach makes no distinction between living and published documents.
Let’s further assume we keep a history of all versions (like version control in software!) and make it easily accessible. Then:
-
Addenda become completely transparent, whether they are typos, rephrasings, or the rewriting or addition of whole paragraphs.
-
Citation, at least between digital artifacts, becomes less awkward. Citing links that might break and adding the proviso “last accessed” has always seemed idiotic to me. “Last accessed” can at least now become implicit in the updated date of a specific version of an article.
-
Hyperlinks between articles inside a time-versioned system like ours can refer to contemporary article versions.
As we can see, a lot of unexplored or underexplored opportunities open up if we follow this radical approach.
Recently I wrote a whitepaper and implemented a beta version of Personalist,[1] a knowledge management system with a lot of overlap with what is described here. The main difference from what is described there lies less in the fact that Personalist—as an encyclopedia (note the metaphor!)—is concerned with definitions rather than articles, as we are in the publication platform we are designing here. Rather, it lies in a design choice: I found it important to place great emphasis on how relations are modeled.[2] Since terms are always defined in terms of other terms, I found that relations should be emphasized far more than existing systems do.[3]
Here, in contrast, I want to go with the traditional hyperlink model, which, from a modeling perspective, is just “empty relations”—that is, relations that are not further annotated. A text passage is marked up to point to another resource available on the internet, without further comment justifying that choice.
What the system described here should do, similarly to Personalist, is—where it can, that is, inside the sphere of systems that support this logic—point to a contemporaneous version of that resource (perhaps with an option to point to an even older version).
As soon as I have a first working prototype, I will post a link here.