One of the features of a good knowledge management system is its ability to bring together information or insights gathered from different contexts. When brought into juxtaposition with one another, knowledge from disparate domains can form new constellations of thought. As Luhmann writes, "information ... originates only in systems which possess a comparative schema—even if this amounts only to 'this or something else.'"
When we collect information in a personal knowledge management system like a zettelkasten, we often try to make information modular. We tease out the core insights and articulate them in such a way that we can build new claims and arguments. In other words, we want knowledge that is transferable across domains.
Sometimes, this involves abstraction: we try to understand a more general principle that the specific insight suggests. We have to make a bit of a leap. We ask, "What might this mean?" or "What is going on here?" This is the work of synthesis: the recombination of ideas into a new theory.
But transferring knowledge isn't always easy: it can be hard to tell whether or not a particular kernel of knowledge is extremely content-specific. A specific observation on its own probably doesn't translate well. But a few observations can point to a larger pattern. Once we start to collect many observations, we may be able to infer new possibilities. Then, we can reason abductively. We begin to speculate, given the information we have, what else might be true.
The most familiar example of this in action is probably the method of Sherlock Holmes. Contrary to popular belief, Holmes typically did not practice deduction but abduction. His observations of discrete facts—the mud on a shoe, the character of a man's hands—formed suppositions that he then tested with a dramatic flourish, much to the amusement of Dr. Watson (and Conan Doyle's readers).
But I think abductive reasoning is underestimated in many settings. It seemingly lacks the apparent rigour of a more scientific deductive method, where a hypothesis is put to the test and determined with some clarity whether it is true or false. The act of constructing new mental models and new theories feels a bit arcane in comparison. People want numbers. They want things they can count and measure—not new theories, metaphors, or models.
It's unfortunate, because it's precisely the kind of thinking that we need most. When faced with complex, wicked problems, abductive reasoning is our best tool: as C. S. Peirce wrote argued, only abductive reasoning is capable of developing novel insights or knowledge. We need to reason abductively to make sense of uncertainty. After all, we need to build a thought model before we can undertake to deductively validate (or falsify) it.