It’s kind of wild when you stop and think about it.
Most documentation is arranged into files, folders and hierarchies. A document goes inside a folder inside another folder and on the surface it’s well tidy. The trouble is that knowledge rarely behaves that way.
A component accessibility decision can affect design, engineering, content and customer support. So where do you put the documentation for it? The more complex the subject, the more awkward that choice becomes.
The modern desktop was developed at Xerox PARC in the 1970s. Early graphical user interfaces borrowed heavily from physical office concepts like documents, folders and filing systems because they were familiar to office workers.
Fifty years later, we’re still using the same mental model. But how many people under 30 have even used a physical filing cabinet?
…
I do know that the younger people that work for me just shake their heads at my metal filing cabinets filled with paper. I do work for a company that makes metal office furniture though:>) I do love it when something pops up that they don’t know the answer to and I can quickly pull a document from one of my folders to find the answer. Doesn’t happen all the time but when it does I just smile:>) I love technology but sometimes old school rules!!!
I think the biggest takeaway to that article is that mult-modal searching is what’s really needed. If you already know what you’re looking for and where it’s supposed to be, a filing system works wonderfully. If you don’t know exactly what you’re looking for (or if it even exists) that can break down.
One thing we can all agree on though, is M$'s current model of saving things is absolutely horrendous!
This concept isn’t actually new, when document management became content management is the point where the concept of folders started to become less popular. Back in the 90’s Xerox help create Documentum it’s intention to manage unstructured data, Opentext also arrived at a similar time with the same focus I worked with both around that time and remember concepts like searching for key words and instead of actually revieing the documents, you could see information like who wrote the documents, so instead of reading anything you just worked out who the expert was.
I would say that the concept of folder structure and hierarchy still exists more as a legacy of the technology where the operating file system still needed some kind of structure instead of a million files in one folder. EPICOR’s own ECM is a great example of this, it does still hold onto the concept of folders, but they are largely irrelevant to manage the content itself, but still useful to manage security.
Full text indexing has been around for decades allowing key work searches etc. AI may add context offering some deeper insight (and potentially speed) into the content search but I don’t think this is something radical or new.
Maybe on the back end - but things like wikipedia have been around a long time. From the very beginning of my programing experience, I found stuff on MSDN by right clicking on things in visual studio or googling things. In both experiences you arrived at things from searching or from other information/pages, not by browsing folders and files.
I guess I am just agreeing with @steveh - we’ve known this for a long time, before AI, so the premise that AI has exposed a new problem or created a new paradigm seems like an attempt to make something old sound new an relevant again. AI is a better search tool than we had before, so it can make better use of meta data etc - but that doesn’t mean the physical location (file folder) metaphor needs to be eliminated - after all people use this as a memory technique too. The more paths/connections/metaphors the better, per the article.
Also on a different note, we are physical creatures in a 3D world. Its not always a bad thing to make digital things mimic the physical world our brains are used to. (E.g “The doorway effect”) So many people seem to think the goal is to become one with the machine, not use the machines to achieve things in the physical world. The file folder metaphor isn’t all bad.
It’s the difference between Outlook and GMail or Yahoo! and Google.
Google said, “I got it. Don’t you worry about where it is stored. Why waste cycles on choosing where it fits? Tag it and I’ll find it.”
Steve, and younger punks like Evan, know the information world isn’t two dimensional. It’s how Kirk defeated Kahn - the product of 20th century genetic engineering who was trained in two-dimensional thinking. AI is not two-dimensional. ECM knows this. SharePoint knows this but tries to keep a foot in both boats. Sure. The cabinet is comfortable for people used to it - just like writing a paper letter used to be, or taking a horse instead of a train or a car, or…
I think this is why Markdown with front-matter is easily consumed by AI. Unlike SharePoint and ECM, the meta-data is stored in the document divorcing it from any particular solution. This is one of the reasons that Obsidian has become quite popular. Hell, even SharePoint now natively supports Markdown.
Filing cabinets and the folder metaphor are just like smart numbering systems. They always break-down when they run into reality. AI doesn’t care how you store it, it only wants the tags and relationships.
The original neural network based on the single-layer perceptron never moved the AI needle because it was linear and flat. It wasn’t until they added multiple dimensions to neural networks that we saw these great advances in machine learning.
I struggle to find emails with gmails abysmal search. I never have a hard time finding my files. (Now helping other people find THEIR files.. different story)
I think if my world consisted solely of text files of human readable information, perhaps I wouldn’t miss the folder metaphor. After all, I don’t often find myself wishing each record in Epicor was a file that you could open from windows explorer. But I can’t image navigating around a OS without a file system.
Abstraction isn’t without cost. Yes, I don’t want to write assembler - but the person who knows how to write good assembler is going to have a much more in depth understanding of how the computer works than somebody who can only vibe code. File/Folders seemed like a reasonable way of connecting bytes on a specific magic hard drive to physical documents.
Now I find people who have no concept if something in on their phone or in the cloud or where exactly it exists.
Humans need a structure. The Filing Cabinet model fits it.
I said you shouldn’t stop there. Metadata / DB model etc, are all wonderful expansions on it.
I just think it’s best to start with a cabinet.
I take that back. If you have the forethought, add as many tags or whatever in the beginning, and it will make you and your tools have an easier time down the road. But a filing cabinet is better than nothing. Do something lol. Maybe you can have your AI expand it for you later. Good use for it.
Then you can use abstractions to put it in whatever model you want.
Absolutely. Filing systems are as individual and tribal as we get. We know our filing systems but not others. Sometimes we use a common system. My job at the college library was to “read shelves.” This meant we were looking for books not placed in their proper location using the Library of Congress classification system. If it wasn’t located where it belonged, people couldn’t find it. We relied on the card system to index the same book multiple ways, but each card pointed to one location.
As the original author stated, AI does not use the location of the document for analysis. If we want AI to surface interesting relationships within our documents, we need to add tags, meta-data, links, etc. The point of the article is not how to organize information so that humans can find it easier, it is how to organize data so machines can find it for us more easily.
Sure. The cabinet is comfortable for people used to it - just like writing a paper letter used to be, or taking a horse instead of a train or a car, or…
If I want to add structure, I add tags…
