In general, a more quantitative approach is more amenable to computational, *quantitative* integration; if, however, it doesn't include rich, explicit mappings between the number and the context behind the number, we're nervous about its utility in practice
This "shrinking" is often accomplished by abstraction. While powerful, abstraction can impede synthesis if context isn't easy to recover (i.e., if the compression is "lossy"), because Z: Compression and contextualizability are in tension, and Z: Contextualizability is necessary for synthesis
If I understand it correctly, Liane's approach of using quantum theory (see, e.g., R: aertsConceptsTheirDynamics2013a)to model concepts makes Atomicity important, but also inextricably tied to context. This is also a central idea in Niklas Luhmann's idea of the sys/Zettelkasten, where individual notes (which are supposed to represent individual "ideas") obtain their meaning only by being connected to other ideas; that is, their meaning is determined by the context of the notes they are connected to.
But more powerfully, allows me to tag in the context of lit-notes, and then pull up papers that discuss X (could be concepts, could be zettels). This is far more natural a way to formalize content and associate with a paper than trying to tag stuff ahead of time, out of context (following Z: Organize by using)
But the notion of "levels of evidence" is contentious, especially in settings with high interdisciplinarity - the explicit articulation and negotiation of degrees of belief and uncertainty, about epistemology even, might be where true interdisciplinary progress happens. Prematurely "compressing" these nuances into a single context-less number threatens the real usefulness of that number for synthesis.
Z: Specifying context for future reuse requires predicting trajectories of future reuse, and it's hard to predict the future! Particularly if you want to creatively reuse something in quite different contexts
entropy in the information theory sense is supposed to answer the question "what is the smallest size of package I can send that retains all of the information I want to communicate?" this seems very relevant again to our question of how to compress ideas. I think the connection is a bit loose, but the principle of entropy *might* be relevant for thinking about how to decide what context is likely to be needed, given an entry point. Maybe also for conversations as context, to make it easier to share a sys/Zettelkasten (cf. idea: multiplayer zettelkastens)
Without understanding why an idea was written, how it relates to other ideas, what its precise meaning is in the context of the knowledge sharer, and how that idea is warranted or grounded in observations, details, and evidence, the "mileage" of that piece of information is extremely limited.
This might be one reason that Andy Matuschak and Michael Nielsen think that Z: The most transformative insights must come from a single "mind". Because it's really challenging to (or we haven't yet figured out how to) enable a medium with data model properties of compression, context, and compositionality that spans multiple minds; a distributed system that enfolds or spans more than one mind and their workflow.
authors think that there is enough regularity in the types and locations of context information that we could build an automated system to extract htese bits of information, which would in turn enable exploration of a variety of hypothesis projections
To properly deal with a discrepancy in benefits from the caller, a HLG employee needed to reuse information from the CARL database, among other sources of information, to create an escalation to the benefits group. To do this, she needed important context that was missing from the CARL record itself, such as details of the record's creation or maintenance (was it authoritative?), and any circumstances surrounding the caller's employment. The HLG employee dealt with this missing context by consulting an expert (a senior agent) and her own memory, rather than searching databases for additional information, even though that information could in principle be in there.
need to coordinate to make sure things happen well and in a timely fashion, but coordination requires insight into the context of production and work for individual team members, and that's hard without special tech
some types of context information were more contextual, depending on the particular "hypothesis projection" of the review, which varied across the lifecycle of the project studied (e.g., location of medical condition, amount of exposure, confounding risk factors), while others were more constant regardless of hypothesis (e.g., study- and population-context information)