Desktops, metadata and filing
In 1973, Xerox’s Palo Alto Research Center released the “Alto” personal computer. This was the first machine to boast a graphical user interface (GUI) instead of the traditional character user interface.[1]
|
JC pontificates about technology
An occasional series.
|
A bad metaphor
To lessen the cognitive burden on users — at the time, bowler-hatted civil servants, sleeve-gartered clerks and others whose mental framework comprised a typing pool and mailboys running memoranda around the office in reusable manila envelopes, and whose idea of “information technology” was a pneumatic tube system that launched invoices around the clanking pipes of the organisation like mortar bombs — Xerox PARC created the metaphor of the “desktop”.
Yes, you were looking at a computer screen. But on that screen was not an impenetrable wall of green code and a flashing cursor, but a cartoonish depiction of a literal desktop, with its comforting iconography of manila folders, a blotter, filing cabinets, in-trays and out-trays and even a dinky little wastebasket. All very familiar. All just so. Perhaps this transition to the atomic age would not be so bad after all.
A better metaphor
In 1979, Dan Bricklin and Bob Frankston created VisiCalc for the Apple II computer. VisiCalc was the first spreadsheet program. It would revolutionise business computing by allowing even modest undertakings easily to create and manipulate data about their operations. VisiCalc wasn’t nearly quite as dinky or intuitive as the desktop. It didn’t have a graphic user interface. It was a much purer expression of what a personal computer could do, though: it promised a powerful means of storing, augmenting, filtering, analysing and manipulating structured data. VisiCalc was, of course the ancestor of that beast we all now know and love as Microsoft Excel.
Why it’s is a good metaphor
Aspreadsheet is a much better way of thinking about how to organise information on a computer than is a desktop.
Being a conceptually infinite number of rows and columns — limited in practice, but these days not by much — a spreadsheet extends in two infinite directions:
- Downwards: You can add items to your filing system without limit, unconstrained by the area of your desk or the volume of your filing cabinet, where each item occupies one of an infinite number of rows.
- Across: You can categorise each item in an unlimited number of ways by creating a new column. Columns have no fixed hierarchy between them, and need not even bear any relation to each other. Whereas a subfolder is necessarily a sub-division of the folder it sits in, this is not true of a new column. Its only condition is that it must relate to the original item.
Whereas in a “desktop” structure the subfolders are divisions of a folder, each further level in the folder structure becoming more fine-grained and subordinate, the category columns in a spreadsheet have equal standing — they are, er, pari passu — and their combination has a multiplicative effect: if an existing column, or an artful combination of columns, doesn’t yield the information you need, you can always add more columns.
A front in the battle between substance and form
The desktop prioritises form.
A spreadsheet prioritises substance.
The last thing to notice is our old friend the struggle between form and substance: if we take it that, whatever your metaphor of choice, the “item” — the thing being filed — is the substance and the organising system it goes into is the form, we can see that the desktop and the spreadsheet have fundamentally opposed philosophies.
The desktop priorities form over substance — the “item” is buried at the bottom of a rigid formal structure of folders and subfolders which cannot easily be altered. This is why it is so hard to find things you have misfiled. You cannot put anything into the database until you have fully specified the folder path that comprises its formal structure.
By contrast, a spreadsheet prioritises substance over form. The “item” is the first thing to go in the database, naked of any formal structure. It therefore sits at the top of the structure. Only once it is in situ can you assign it any formal metadata properties. The item wears its metadata lightly, and is not affected if the metadata is later altered, removed or augmented.
Metadata
Each folder or column is its own item of metadata — literally, “information about information” — about the item being filed.[2]
A desktop folder structure can use a very limited range metadata: a limited number of characters of text. So limited that it is hardly worth thinking of it as metadata at all.[3]
In a spreadsheet, by contrast, metadata can take the form of text, calculable numbers and dates, checkboxes, people,[4] colours, flags, choices, lookups, comments, or calculations. It can be validated, managed, controlled, compulsory, optional, pre-populated or free-form. Each extra piece of metadata enriches the existing data in the row without detracting from it. Indeed, even if the metadata is wrong, the inconsistencies between the fields allow a user to triangulate and identify likely wrong — or problematic — material.
Metadata is, in this way, “non-destructive”. It only augments. Each metadata field creates its own way of ordering information. Each is its own hierarchy. If metadata is non hierarchical, it can be much more specific — you can date-stamp to the second, where you might have had a folder for a month.
Suddenly, you can organise the same information in multiple different ways, simultaneously, without upsetting anyone else’s existing categorisation.
You can then filter and group your items by one or more columns. You can sort, chart, pivot and triangulate. The more metadata you have, the more ways you can look at the data.
You can even sort your data using data about how much metadata there is.
This is metametadata.
The spreadsheet approach to file management is, thus “multi-hierarchical” and non-destructive. What about unused metadata? Ignore it. Unused hierarchies are almost costless. And you just never know —
The desktop clings on
Yet on our modern, hyper-networked, cloud-based work environment the desktop metaphor hangs on. We still call them “desktops”, though now for the prosaic reason that they generally are the only thing that sits on top of our desk. The desktop was a nice, quaint idea, and it got old men in green visors to sit down at a keyboard, and for that the ranks of middle management can be truly grateful, but it has well-outlived its purpose now.
Because physical information that sits on a real desktop, and digital information that sits on a computer are very different ontological propositions.
The desktop metaphor asks us to put our files in folders, as we would do on a real desk. If a folder gets too big, we create subfolders. And, just as with a real desk, once we have put a file in one folder, we can’t very well put it anywhere else. Just as with a real filing cabinet, if we misfile our subfolder, we might never find it again.
In the real world of physical information, that does no more than reflect grim corporeal reality: a thing can only be in one place at one time so that’s that. If the boss wants to file by customer, and you want to file by industry, then tough.
Physical filing systems reflect this: there is a unique physical location for any single document. So do physical filing methodologies: older readers may remember the Dewey decimal system, which numbered the entire corpus of non-fiction wisdom from zero to 1,000.[5]
If the same document does need to be categorised in different ways — say the legal department needs to file by customer and the credit department by industry, this could only be achieved by duplicating the document and holding one version in each location. Legal would have a filing system, and credit would have another.
Plainly, this is an imperfect state of affairs. It created a basis risk. Which was the canonical version of the document? What happened if one of them, but not the other, was updated? Where it is a “living thing” plotting its own miserable trajectory through the cosmos — a contract under negotiation, or a periodically updated legal template, for example — then duplicating it is a bummer. It duplicates the manual task of updating all copies of the document as it changes, and that introduces the opportunity for human error. There may be miskeys. A document may be forgotten. Version control is a pain.
In the physical realm, duplication was slow, imperfect and expensive and so, limited. At the time this seemed to be a drawback; with hindsight, it appears a valuable discipline.
Also a preferred hierarchy can change, as personnel, managers, business priorities, or circumstances change. Changing your preferred hierarchy means completely re-engineering your folder structure. It would be lovely to not have to do that every time there is a corporate reorganisation.
So: a folder structure assumes a single hierarchy and multiple copies of each item.
Substrate neutrality
These are all problems of the physical realm; the spreadsheet metaphor shows us we need not be so troubled in a digital realm. In the digital world, the physical “substrate” of a document — the paper it is made out of — is, to all intents and purposes, irrelevant. What matters is the ASCII code embedded in that document. In a digital world it has been abstracted from the substrate and floats free. Within a diverse network of collaborators, this is immensely empowering.
It did not take people long to realise that email was amazing.
From, more or less, a standing start in about 1993 — by lucky coincidence the year JC entered the workforce — the corporate world fell head over heels in love with electronic communication. Whatever reverence it had for the sacred substrate fell quickly away.[6] The expression, “this document is not worth the paper it is written on” has lost its meaning because the paper it is written no longer has much value at all.
Now we recognise the digital content embedded in the substrate is the valuable bit; the paper bit is just annoying. It is an inconvenient reminder of our erstwhile physical analogue reality. The better metaphor than the “desktop” here is the spreadsheet. A spreadsheet is, of course a rudimentary form of a database.
In a spreadsheet that inconvenient imposition of substrate has gone: a “document” is nothing more that an information string: more or less costless to generate, transport, replicate and store. By simply appending metadata, we can enrich it and put the same thing in several places at once. We transcend the Euclidian geometry of physical space.
Now, I said our reverence it had for the sacred substrate fell quickly away. It did not entirely fall away. We still revere wet ink, for some reason counterparts clauses, and the dear old desktop. For still, as we file, we cannot resist the siren call of folders. Folders in folders in folders in folders in folders. Why do we persist with folders?
More than twenty years ago Tom Zingale taught a young JC a valuable lesson. Battling with some byzantine folder structure, and losing, JC cried out in anguish, and Tom said this:
JC: How on earth am I meant to organise all this?
Tom: With metadata.
JC: Er, with what?
Tom: Metadata. The answer to your question is metadata. Metadata, metadata, metadata. Whatever your question is, the answer is metadata.
JC: Well, my question is, “How do I use metadata to fix this filing problem?
Tom: Oh, right. Simple: SharePoint.
Now a lot of good people viscerally hate SharePoint. And, to be sure, Microsoft seems to have gone out of its way, over 20 years, to make SharePoint as hard to love as it can. But at the same time, it has based its entire Office 35 Suite on the SharePoint platform. It is, to be sure monumentally confusing, the Teams integration is baffling. The utterly dismal online versions of its Office suite drive people righteously up the wall.
But, still, a good part of the enmity for SharePoint arises from this fundamental misunderstanding. SharePoint is the first, philosophically, digitally native operating system.
SharePoint has abandoned the desktop metaphor.
SharePoint uses the spreadsheet metaphor.
In SharePoint you organise by metadata, not by folders.
DO NOT USE FOLDERS IN SHAREPOINT.
Folders are top-down. Metadata is bottom-up. Folders prefer form over substance. Metadata prefers substance over form.
SharePoint allows you to do exactly the same thing with a document library as Excel allows you to do with a spreadsheet.
So it is odd — isn’t it? We intuitively understand the power of metadata when we are presented with a spreadsheet. But the same power does not occur to us when we are presented with a file management system. The desktop metaphor is burned on our retina.
Even though it is, in essence, a supercharged online spreadsheet, SharePoint continues to be resented by almost everyone.
See also
References
- ↑ It was well ahead of its time: the GUI would not become mainstream until Apple released its Macintosh a decade later, in 1984.
- ↑ Grammar pedants’ corner: Even though “data” is plural, “metadata” is generally treated as a singular mass noun. Please direct your letters to the Royal Statistical Society — not because it is their fault: rather, they might keep metadata about this sort of thing.
- ↑ Indeed, the Windows operating system doesn’t treat folder names as metadata at all, which is mad.
- ↑ As in, a lookup to an object in a people directory, and not just a text name.
- ↑ My favourite was 001.9.
- ↑ I have a lengthy essay about the gradual extraction of data from the substrate but can't for the life of me find it at the moment.