Knowledge work as craft work

Jim McGee, April 1, 2002

Knowledge work takes us back to a world of craft work practices. The "symbolic analysis" that Robert Reich identifies as the essence of knowledge work is designed to create the one-of-a-kind results that characterize craft products. Recent flaps over plagiarism by both students and professionals are a concrete reminder that the goal of knowledge work is the creation of new results.

Curiously, most of the discussion around knowledge management is decidedly industrial in flavor. How do we reuse knowledge assets more broadly in the organization? How do we raise the standard of practice in the organization to that of the best units? Standardization and reproducibility have been the watchwords of the industrial economy. Shouldn't they also be the watchwords for the knowledge economy?

There is a dangerous tension between industrial frameworks and knowledge work as craft work that needs to be managed. Forcing industrial models onto the management of knowledge and knowledge work accounts for much of the disappointing results of knowledge management efforts to date. Harking back to some idealized pre-industrial craft world is likely to be equally disappointing. Designing a synthesis that recognizes the essential features of the knowledge economy is the challenge at hand.

The importance of visibility in craft work

Almost by definition, the final product, process, and intermediate stages of craft work are visible. Consider your experiences at a glass blowing workshop or touring a silversmith's workshop. The journey from apprentice to master craftsman depends on the visibility of all aspects of craft work. (As an aside, Etienne Wenger's excellent work on communities of practice (e.g., Situated Learning or Cultivating Communities of Practice) grew out of observations about learning in craft environments.) The notion of trade secrets, for example, grew out of the fact that you needed to conceal elements of your process from the prying eyes of competitors.

Knowledge work today as invisible craft

One thing that differentiates knowledge work today from other craft work is that, except for final product, knowledge work is essentially invisible. All the important stuff takes place inside knowledge workers's heads. This has not always been true of knowledge work and need not be true. Consider this excerpt from "The Social Life of Paper" in a recent New Yorker (courtesy of the shifted libarian):

"But why do we pile documents instead of filing them? Because piles represent the process of active, ongoing thinking. The psychologist Alison Kidd, whose research Sellen and Harper refer to extensively, argues that "knowledge workers" use the physical space of the desktop to hold "ideas which they cannot yet categorize or even decide how they might use." The messy desk is not necessarily a sign of disorganization. It may be a sign of complexity: those who deal with many unresolved ideas simultaneously cannot sort and file the papers on their desks, because they haven't yet sorted and filed the ideas in their head. Kidd writes that many of the people she talked to use the papers on their desks as contextual cues to "recover a complex set of threads without difficulty and delay" when they come in on a Monday morning, or after their work has been interrupted by a phone call. What we see when we look at the piles on our desks is, in a sense, the contents of our brains."

One unintended consequence of today's technology environment is to make the process of knowledge work less visible just when we need it to be more so. The end products of knowledge work are already highly refined abstractions; a financial analysis, project plan, consulting report, or article. Today, the evolution from germ of an idea through intermediate representations and false starts to finished product exists, if at all, as a series of morphing digital representations and ephemeral feedback interactions.

Let me go way back for a counterexample. I started consulting before the advent of the PC. When you had a final presentation to prepare for the client, you started with a pad of paper and a pencil and roughed out a set of slides. You could see that it was a draft and the erasures and cross outs and arrows made that even more obvious. This might be two weeks before the final deadline. Then, you took it to Evelyn in the graphics department down on the eighth floor. After she yelled at you for how little lead time you had given her, she handed your incomprehensible draft to one of the commercial artists in her group. They spent several days hand-lettering your draft and building the graphs and charts. They sent you back a copy of their work.

Then you started another iterative process of correcting and amending this product. Copies got circulated and marked up by the manager and the partner on the project. At the end, the client got to see it and you hoped you'd gotten it right.

All along the way in this old style process, the work was visible. That meant that the more junior members of the team could learn how the process unfolded and how the final product grew over time. You, as a consultant, could see how the different editors and commentators reacted to different parts of the product.

While today's tools have made the journey from germ of an idea to finished product so much easier, they have also made it harder by making it less visible. It's my sense that most of us don't even see what we've inadvertently given up. It takes a conscious act of will to think about how to use today's tools in ways can give us both the productivity of the new and the process value hidden in the old accidental visibility.

So what? Only the final product is relevant, isn't it? What possible value is there to the intermediate versions? If anything, recent experiences such as the document shredding at Enron would suggest that anything other than the final product is a potential liability. We're more likely to see organizations working more diligently to eliminate these traces of knowledge work than to organize and benefit from them. That would be unfortunate.

Direct value of improved visibility

As an individual k-worker, the value of increased visibility might seem a bit abstract. You know what you did. What's the incremental value of an explicit audit trail?

One value is in the ability to backtrack to a previous version when a line of analysis fails to pan out. Moreover, that ability to backtrack can make it more likely that alternatives will be explored because the effort and risk of doing so is reduced. Spreadsheet modeling is a good example of both the value of flexible tools for exploring alternatives and of the problems inherent in the invisibility of many knowledge tools.

In the days of paper spreadsheets, you spent a lot of time thinking about how to set them up and you didn't go off exploring multiple alternatives. Dan Bricklin invented Visicalc to make spreadsheet analysis more thoughful and less mechanical. An electronic spreadsheet is dynamic.When you type in new numbers or enter new equations you see the new results immediately. This frees you to explore multiple alternatives. What new spreadsheet analysts often fail to grasp is that there is no trail of breadcrumbs to lead you back to your starting point if you reach a dead end. There aren't even eraser marks to suggest where you've made changes.

While the newer spreadsheet tools provide an array of features to make auditing and backtracking easier, most spreadsheet users have still to make the conceptual leap that they need these features to begin with. From my observations of most spreadsheet users, best practice appears to be structuring your models to separate parameters and assumptions from the main model and to save periodic snapshots of an evolving model as separate files. What that means is that the bulk of the audit trail still exists in the head of the developer. For one thing, that's a cognitive load you don't need. For another, you're ability to reconstruct the audit trail depends on the quality of your memory vs. the thousand other demands on it.

Similar arguments apply to other categories of knowledge work tools. One useful direction for some additional poking around would be to investigate how to apply the lessons learned in software development around version control and source code library management to more general forms of knowledge work. Wouldn't you like to have that level of tracing over your powerpoint presentations or correspondence files? Doing that today takes a level of sophistication that the average knowledge worker or knowledge work group doesn't have.

Indirect value of improved visibility

The indirect values of improving knowledge work visibility may be the heart of realizing the promise of knowledge management for the organization. Some of that indirect value will come from pushing knowledge into the industrial frames I mentioned earlier. But more will come from two other aspects of visibility. The first will be increasing the value of knowledge work as a learning environment for other knowledge workers. As craft work, knowledge work fits more into apprenticeship learning models than in conventional training approaches. Making the work process and its intermediate products more visible will make the apprenticeship process more effective.

The second aspect of visibility is better leverage of communities of expertise and practice. More and more of the difficult problems organizations face require groups of experts to coordinate their expertise and invent multi-disiciplinary solutions. These problems don't identify themselves in advance. They show up. They generally get addressed by whatever team can be identified and assembled quickly. The more visible you can make those experts and their expertise by making their thinking visible, the more likely you will be able to field a team that will work.

Weblogs and visibility

With all the attention to weblogs and knowledge logs, this aspect of visibility hasn't yet been emphasized. John Robb's discussions of k-logging are closest to this point, but still emphasize final products without calling enough attention to process.

John Seely Brown helps show how important the process dimension is in knowledge work in The Social Life of Information. An excerpt from the relevant chapter ran in CIO magazine.

The concept of a k-log is a huge step in a good direction. The fundamental chronological structure of k-logs makes it straightforward to capture an audit trail of an evolving knowledge work product. Tools like Radio and Instant Outlining push the envelope further and offer possibilities for realizing new visibility in doing knowledge work. The additional step that I am trying to articulate is that we need to become mindful of the process of knowledge work. That process is nothing like the process of industrial work. We will collectively have to invent it and refine it and think about it as we go.

The discipline that needs to be developed is that of organizing the rest of your surrounding knowledge work to make its evolving nature more visible. The tools make that more manageable than in the past and weblogs provide an overarching organizing structure. All that remains is for you to believe that it's worth doing.