Unstructured Thoughts on Information Management

Filesystems were useful but also a sub-par solution to information organization. Search has clearly surpassed the previous model.

Once upon a time, we put information in a “file”. That file was subsequently stored in a hierarchy of nested “folders”. We replicated this model in many different forms — Websites were under domains and had pages, often grouped in sections, Apps lived in folders on different screens of your mobile, Email was grouped into folders. This model’s primary strength was that it ensured there was a place for everything, it also (in theory at least) allows exploring content by [exactly one, arbitrary] category (i.e., the folder) and allowed a consumer to find content immediately if they just remembered the exact path. This model directly mapped to a physical experience of putting items in a physical location — a concept we are familiar with; however, it shared the weaknesses. Where did I leave me keys? Has someone stolen my passport or did I misplace it? This was the old model; search is the new model.

With global searches we still use documents as containers for information but we can find them in any location by using keywords from the content or metadata. This avoids the need to remember exact paths to content which, in practice, was often prohibitively difficult for all but the highest-priority information. Searching has no particular equivalent in the physical world but is more akin to memory recall. What do I know about Tesla? What countries border France?

As we become familiar with searching behaviour, we not only have an improved efficiency in finding information but we actually lose the need for folder/organizational hierarchies almost entirely. Having used Google for a decade or two, when was the last time you visited a website and navigated to a sub-page via any navigation other than search? Chances are, this would only happen in some esoteric website that neither offered its own search nor was indexed by Google (e.g., sites requiring authentication). Instead, we may as well dump all information in a single location (no hierarchy or organization) and simply allow consumers to search for what they need.

There is still opportunity for presenting information in a way that allows exploration — this is a key characteristic of hierarchies which maybe has not been improved upon much. That said, the prior model doesn’t work well for this either given the exponential growth of content creation. Instead, we find that particular information needs to be curated and presented — often on demand and in response to expressed interests. I can’t think of many examples where this is done well outside of social media — maybe you can?

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.