We used to know where to put things if we thought we might want to find them again later. It was called filing. Filing got less fashionable in government about 15 years ago, not coincidentally at about the time that paper was getting increasingly displaced.1
The way we think about and manage work and information has changed a lot since then, but to a surprising extent the idea of paper has survived much longer than the reality – all too often, we organise information as if it were on paper, even when it never has been and there is no expectation that it ever will be.
That has the advantage of keeping things simple and familiar. But it also has the much bigger disadvantage of obscuring opportunities to do things better.
There are five big differences that a new approach needs to reflect
- paper is not the medium
- files and directories are not the location
- metadata is data
- people are search engines
- friction is failure
Let’s look briefly at each of those, and then put them together to explore the wider implications.
Paper is not the medium
Word processing programs and web sites both have pages, but they are very different things. Word can’t do anything unless it knows what size paper you have in mind, and what you seen on screen will then be driven by that paper size, even if the document never has been and never will be printed. The physical structure (the point where there is no more room on a page) has no connection with the logical structure (the point where one section ends and another begins). And in most systems, you have to take possession of a document in order to open it. Moving and sharing such documents is still all too often stuck in a world of email as transport system and endlessly self-replicating document as payload.2
There is a small advantage in doing that: it allows the same mental models to be effective in the new world as in the old. But there is a much larger disadvantage: it gets in the way of developing new mental models which are better aligned to the greater power of search and organisation which new tools allow, it means we risk getting stuck in thinking about the new world as if it shared the weaknesses of the old.3
In short, as Mark Foden has put it:
It is time to move from circulating documents to visiting texts.
Files are not the location
Paper documents live in files. The key to finding the document is to find the file. And if you might need the file, you need a filing cabinet reasonably close to hand where it can be safely stored with lots of other files, almost certainly on related subjects. That of course has enormous consequences for the organisation and physical structure of work: if the unit of work is a paper file and that file is a unique (and therefore precious) assembly of information, the location of work is driven by the organisation of information.
Translating the analogy of the filing cabinet into the digital world is arguably even more pernicious than the analogy of paper. A hierarchical folder structure reinforces the idea that there is a single, canonical, right place for a piece of information. More subtly, it supports the idea that that right place tells you all you might want to know about the nature of the information to be found there. Yahoo gave up the attempt to create a structured directory of the web many years ago: it turned out that the connections within and between web pages were a more useful source of information than knowledge of their location – and so began the ascendancy of Google. The absence of those of strands of connection – of hyperlinks – is the fundamental reason why searching a file structure is always more frustrating than searching the web. The web was, of course, invented as a tool to support the organisation of work, for just this reason, but that’s another story.
In supermarkets and filing cabinets, physical location is a direct representation of the approach to cataloguing. Hierarchical filing structures aspire to be like supermarkets, applying an unneeded constraint of the physical world, when they could be much more like Amazon, where the physical layout of goods in the warehouse has nothing at all to do with how they are found on the website.
Would we get further by focusing on tagging rather than filing? Tagging is social, collaborative and can be game-like; filing is lonely, bureaucratic and dull – prompted by this recent post on applying tagging to books – could it be more effective in a working environment too?
The title of that post – everything is miscellaneous – is a a reference to a book by David Weinberger published in 2007. I wrote a blog post about it at the time which I don’t think has dated too badly. This is the sophisticated version of the argument that we should give up on fixed classification as a way of finding stuff: essentially that even now the way we think about filing and retrieval is dominated by the constraints of paper. There’s even a rather splendid five minute video which summarises the argument.
Metadata is data
Tags (and file locations) are of course forms of metadata, as are authors, creation dates, and a host of other data scraps which get attached to files. The level of creative imagination needed to dream up three keywords was itself an insuperable hurdle to the effective adoption of document management systems. That’s not to say that metadata is unimportant. Some of the obvious stuff – date ranges, authors’ names and more – can be extremely valuable. But there is potential which we have scarcely begun to explore and make use of. I may be the creator of two apparently very similar documents (perhaps even so similar that it’s impossible to identify which (if either) is canonical). Understanding who has visited that text may be critical in searching an otherwise undifferentiated mass. Understanding who has contributed to it (and when, and in what order) may be what separates the historically interesting from the ephemeral.
Data, in other words, which captures the history of a document (or better still, of an idea, of a policy, of a ministerial decision) can be useful and powerful. Texts can tell their own story – with not an arbitrary keyword in sight.
Traditionally, we have thought of document repositories as self-contained entities. Whether they are filing cabinets or databases, the raw material for finding stuff is contained within the stuff to be found.
In the long run, that is unavoidably true – you will get no help from the authors of papers written a hundred years ago about why they are as they are and what thinking lies behind them. But in the short run, it’s not true at all: knowing who knows stuff is often as useful as knowing where that stuff might be filed. Knowing who else has worked on similar issues in the recent past and getting their insights into what the key documents might be is often the most useful thing which can be done – but whether it is practically possible is often a matter of chance. And while filing is a chore for most people, a request to share knowledge is a mark of respect and is usually welcomed as such.
So if we want to make better use of the knowledge stored in our systems, we need to be better at finding the people who created or who understand that knowledge. And if people are the best search engines, we need the best search engine to find people.
Friction is failure
All of these problem were supposed to have been solved many years ago. Yet somehow they persist, seemingly impervious to the wonders of new technologies and new ways of working. The underlying argument of this post is that one important reason for that is that we still too easily remain trapped in framing the problem around paper, files and working practices – and that as long as we do so, the promise of digital will remain only a promise.,
But that’s not the only reason why we are still grappling with these issues. Another critical one is that we underestimate the power of friction. People are very good at balancing costs and benefits, even (perhaps especially) if they do so completely unconsciously. If benefits accrue in the indeterminate future and costs are incurred today, those costs will be avoided to the greatest extent possible. And if the present value of those benefits is at or close to zero, the acceptable cost will also be zero.
My instinct is that that explains why it has proved so difficult to persuade people even to do things which appear trivial to those who have designed the systems. Assigning keywords or navigating to the right folder are tasks which take just seconds – but if those seconds are pure cost, they will be avoided as far as possible.
There are two ways of dealing with that. The more obvious is to attempt to reduce the cost. It’s true that trying to get people to do a smaller and easier thing is more likely to succeed that getting them to do a bigger and harder thing, but if they are resistant to doing anything, it will still be a long and continuing struggle to make any kind of difference
The second way is less obvious, but I suspect is more powerful. Instead of reducing the cost, let’s aim to increase the benefits. The time value of money (or in this case, the time value of time) strongly suggests that the scale of increase needed for benefits far in the future is likely to be unachievable. So instead, let’s look for ways of bringing the benefits into the present. That makes it much more a social challenge than a technical one. If we recognise, reward and value people who manage knowledge effectively, if we set real expectations that a piece of work will be seen as successful only if it is captured in a way which maximises its medium (and longer term) information value, then perhaps the trade off changes.
Shredding the paper
So where does all that get us? My conclusions are still tentative; we need to do more to test and explore how we best manage information in a digital working environment, and learn from those who are already doing it well. I am pretty sure though, that there are three traps we need to escape from:
- seeing paper as the underlying metaphor for everything
- seeing email as the standard transmission mechanism of information
- seeing the long term value to posterity as more significant than the more immediate value to efficient and effective working here and now.
Once we have got all that out of the way, we are in a position to be much more positive – addressing the challenge of using information to help civil servants help citizens.
The point of all this is not to prescribe – still less proscribe – how we might want to manage these things in the future. That thinking needs to be done, but this post is not it. The point is rather that we are more trapped in thinking about information in ways constrained by the office of the mid-twentieth century than we like to realise. If we want better solutions, we first need to find better problems.
- Not just in government of course. But as in other areas, the same external drivers can result in very different rates of change. ↩
- This is beginning to change with the growth of cloud based services, but the cultural experience is still to download a document, then open it, and only then be able to edit or create. ↩
- Getting rid of paper as a metaphor for digital information is a very different matter from getting rid of paper as a physical object. There are still valuable uses for paper, which is why the paperless office is another piece of the future which keeps receding, but that metaphorical use is not one of them. ↩