Structured Writing and Free Trade

By | 2017/05/22

In my last post, I promised I would reveal the unifying idea that I developed for my new book on Structured Writing. This is the post. So what does it have to do with free trade? Mostly it is that I see the same pattern in discussions of free trade that I do in many discussion of structured writing: a failure to focus on the big picture.

Free trade has been pretty much a given for the last several decades with nations and trading blocks negotiating ever freer trade. But of late the virtues of free trade have been called into questions by the Trump/Sanders wing of American politics, which has those of us in Canada, for whom America is our largest trading partner, a little anxious.

Map of the world's free trade zones.

Emilfaro [Public domain], via Wikimedia Commons

The basis of the Sanders/Trump objection to free trade is the loss of American manufacturing jobs. In truth, the decline of American (and Canadian) manufacturing jobs predate the trade agreements that Trump/Sanders complain about. But that is not the point. The point is that they are focussed on one part of the system and ignoring the whole.

Anne Krueger argues in Curtailing imports is a sure way to make America poor again that the kind of protectionism that Trump and Sanders are asking for would make America poorer and lead to a loss of American jobs and a lowering of American wages because of the way it would impact the overall flow of goods and money. Fewer imports would mean fewer import/export jobs, so it would cost longshoremen and truckers jobs. It would mean a loss of exports, as trading partners would not have the money to buy them because they can no longer sell their goods to America.  This would lead to a loss of American jobs. Most products today have a multitude of inputs and complex supply chains. Many of those inputs can be made cheaper in other countries, so if you refuse to import them, your cost of goods goes up, which means people can buy less, which means they buy fewer things, which means you don’t need to make as many things, which means loss of jobs. The supply chains partition low skill from high skill jobs, moving the low skill jobs overseas while retaining the high skill jobs in America. Curtailing inports therefore means bringing more low skill, low paying jobs to America, which makes America poorer.

That’s the gist of the argument. If you are interested in the argument itself, see Krueger’s article. But my interest in it here is not to argue for free trade (though I am very much for it). It is to point out that the argument for free trade is a holistic one. It looks at the entire system by which good are services are produced and wealth is created. The Trump/Sanders argument against free trade looks only at the loss of a certain class of manufacturing jobs. It does not take account of the knock-on consequences of tearing up free trade. Perhaps there is a holistic argument against free trade. Perhaps it is equally cogent as the holistic argument for free trade. But the point is that it is not the argument the anti-free traders are making. They are driven by a single idée fixe and don’t consider that the side effects of their proposals may cause more losses overall than gains.

Such is all too often the case in structured writing projects as well. Organizations all too often undertake a structured writing project to attack a single problem. They want to do single sourcing, or they want to do reuse. They bring in a system that does just that one thing, that disrupts how everybody works, just to accomplish that one goal. And sometimes they end up worse off that they were before, having created more problems than they solved. Even when they achieve their primary aim they are usually getting far less out of their investment than they should have done, and are not necessarily any better prepared for the next challenge that comes along.

How can attacking one problem — even if you attack it successfully — leave you worse off than you were before? Because the world is complex and the effects of every change you make ripple outwards, just as the impacts of trade tariffs ripple outwards to cost more jobs than they save.

Here’s the unifying idea: Content creation is a complex business. Writing is an intellectually taxing activity. It takes all of the focus the writer has to give to it. If the writer’s attention is divided by other tasks and responsibilities, the quality of their writing — both its rhetorical excellence, and its mechanical correctness — suffer. The full complexity of creating and maintaining a large integrate content set is too much to dump 100% on writers. Therefore we create tools and systems to partition that complexity and direct it to others who are better positions to handle it — either because it requires specialized knowledge or skills or because they can do it without their attention being divided between that task and the tasks of research and composition.

All tech comm tools, indeed, all computerized writing tools, partition and redirect some parts of the complexity of content creation away from writers towards experts and/or algorithms. Thus, unlike the scribes of old, writers are no longer responsible for creating consistent letter shapes or flowing text from one page to another. Those complex tasks have been partitioned away from writers and directed to fontographers and pagination algorithms. This allows writers to focus more of their attention on writing. The writers produce better content and the algorithms produce more uniformly formatted letters and pages. Everybody wins.

But here’s the thing: Complexity cannot be destroyed. It can only be partitioned and redirected. If it is not handled by A, it must be transferred in its entirety to B, who must be able to handle it fully. If some of it is lost on the transfer, or B does not handle it fully, it falls through the system and lands on the reader. If the content is harder to find or to follow, if it contains inaccuracies or omissions or is out of date, the reader’s task becomes more complex. That is complexity that was not handled by the content produce and so must now be handled by the reader.

So what is structured writing? Structured writing is a tool we use to partition and direct the complexity of the content creation process so that it is handled by the right people or algorithms so that none of it gets lost and falls through to the reader.

But what happens when we attack a content problem piecemeal? Complexity get directed away from the point of attack, but where does it go? It practice, it often ends up going to the writer, and if the writer cannot handle it, it falls through to the reader. Thus when we attack the complexity of content reuse, we often create all kinds of complicated abstract content structures including ones that express complicated content management operations  or computer science concepts and expect writers to use all of them while writing. We also introduce complex content finding tasks into the author’s world and introduce complex and expensive content management systems to assist them, which is yet another source of complexity in the author’s world.

When all this stuff works, content can be reused — appearing in more than one publication at the press of a button. A lot of complexity has been removed from finding, managing, and using reusable content. But where did all of that content management and reuse complexity get directed to? It was not destroyed. It went to the writer, who now has vastly more complicated interfaces to interact with and operation to perform, all while they are trying to write. If they can’t handle all of that complexity in addition to their writing tasks, content quality will invariably suffer as the complexity trickles down to the user.

And what happens when all this added authoring complexity impinges on the already complex task of collaboration? What if you have occasional contributors to your content set who are not full time professional writers with full training in your content management and reuse systems? Suddenly contributing is very much more difficult than it was. Many of the occasional collaborators just can’t cope. They give up, dumping more work on the full time writers, or they fudge it, creating a mess that system administrators or writers have to clean up.

Whenever you displace complexity in a system, it goes somewhere else, and if it is not handled there, the result is often that you end up worse off then you were before.

Does this mean that it is impossible to implement an effective reuse system? Of course not. But it does mean that you have to think about more than reuse when you do so. You have to think about where the reuse complexity that you displace is going and how to make sure it goes to a person or process that is equipped to handle it.

Structured writing turns content into data that can be verified and processed by algorithms. To partition and distribute the complexity of content creation in your organization, you need to make sure that the content passes from one partition to another in the form that the next partition expects and with all the information that that partition needs to process it successfully. A structured writing language is a description of the interface between the partitions of a structured writing system. When you change how your system is partitioned, therefore, you change the structured writing language that is used at each stage of your process. Conversely, of course, this means that when you adopt any structured writing language you are implicitly adopting the partitioning of the content system that it is designed to implement.

If any part of your process is dumping more complexity on a person or algorithm than the are equipped to handle, the cure is to change how the system is partitioned. In the new book, I will show how the complexity created by various content development functions can be partitioned in different ways using markup in different structured writing domains.

But the unifying idea is this: you don’t start with a content modeling exercise (the traditional starting point for many structured writing projects). You start with a partitioning of the complexity of the content system to ensure that all the complexity gets handled by people or algorithms with the right skills and resource. You then design you content models to create robust and usable interfaces between the partitions of your content system.

Just as with trade, you have to treat the system as a whole, not fixate on individual parts.

9 thoughts on “Structured Writing and Free Trade

  1. cud

    Not sure how this applies, but…

    Probably the biggest argument (perhaps understated because you don’t want to arouse the sleeping dragon Smog) is not a loss of jobs, but a loss of CONTROL. When a corporation can sue a government to keep the government from protecting its citizens from pollution or other mischief, that is a loss of control. Another argument is simply that there is no such thing as free trade… Never was, and never will be. What we call free trade is the handing of control over to a shrinking group of people.

    How does that compare to structured writing? Maybe structured writing is part of a progression toward loss of control on the part of the writer? The tool that gives you the most control as a writer is the pencil… If you know how to use it. You want formatting? You want fonts? You want illustrations? You want page breaks? Nothing can beat a pencil. (This would be like an artisan-based economy, I guess.)

    The more mechanized the act of writing has become, the less control the writer has had. Typewriters broke it down to words and white space… e. e. cumings being the handiest user, perhaps. Word processors and DTP mechanized a wider range of actions, and so they seem to return some control. But that could be an illusion — by adding more actions into a machine, you push more onto the writer but take control away from other people… It’s likely there’s a net loss.

    Structured markup is another step in that progression. Now you must explicitly state a schema, and authors must explicitly satisfy it. This is so other machines can do more with that content (reuse, multi-channel, dynamic updates, etc.). But to hand control to even more machines means you take even more control away from the author.

    What we do for a living might warrant this transfer of control. I certainly enjoy exploring it. But I can understand how some people just don’t like it.

    1. Mark Baker Post author

      Indeed, loss of a control is an issue in both cases. Of course, in a free trade agreement, the loss of control is mutual. But one does tend to chafe more when their corporations sue our government than when our corporations sue their government. It is part of what can make free trade a hard sell.

      Loss of control is certainly an issue with structured writing as well. But while the pencil does indeed represent the ultimate in freedom for a writer, writer’s are somewhat atypical in their attachment to it. For a programmer, for instance, machine code is the pencil. Yet few programmers want to work with machine code and most would not even know how. Programmers have always recognized that higher-level languages give them the freedom to focus on algorithms and usability and just generally to get things working more quickly. In other words, they give them the freedom to make working software without the complexity of having to twiddle millions of bits by hand.

      Higher-level languages, therefore, are not seen as a form of constraint (though of course they are) but as tools of liberation.

      Writers often view structured writing very differently. They are far more likely to chafe at the constraint and far less likely to rejoice in the liberation. The question is, where does that different come from?

      There is no question that, for many projects, the right structured writing techniques can vastly improve both the efficiency and the quality of content. Structured writing can bring repeatability to rhetorical design, which (for the first time) makes content genuinely testable in a way which allows you to extrapolate the results of testing reliable to new content. It greatly improves consistency, coverage, and findability. It is a tool that you can use to make better content faster, just as high level languages are tools that programmers can use to make better software faster.

      The pencil may represent the ultimate in freedom but by the time you have made a few thousand copies of a document using a pencil alone, you are probably going to find that liberty irksome and wish for higher-level tools. And, of course, writers have embraced higher-level tools in the form of word processors and DTP systems.

      The question is, why have they not gone the next logical step and embraced structured writing. My working theory is that the reason is that most of the current structured writing systems were not designed for writers at all. They were designed as the front end of publishing and content management systems with not a lick of thought given to their impact on writers. Writers, in other words, don’t get to choose their tools or to have any input in their design.

      My theory is that if we can give structured writing back to writers, they will see the power it offers and claim it as their own, just as programmers have done with high-level languages. The book is an attempt to show both writers and content owners what is possible.

  2. cud

    I look forward to the book, BTW. It promises to be interesting!

  3. Larry Kunz

    Thanks for this, Mark. I agree with your overall premise: that structured writing requires a holistic approach. And I love the definition you provide in the “So what is structured writing?” paragraph. I think it’s best one I’ve ever seen.

    However, in stepping back to look at the big picture, I’m not sure you’ve stepped far enough back. Rather than starting “with a partitioning of the complexity of the content system,” as you advise, shouldn’t we first do that content modeling exercise — specifically, the part of the exercise in which we identify the goals we want our content to achieve?

    Without knowing the goals — and they’ll vary for each project — I don’t think we can talk meaningfully about the content system and how to partition it. Perhaps our problem isn’t that we’ve been starting in the wrong place. Rather, we’ve been starting with the goals and then rushing to address each goal individually, skipping the vital step of analyzing the whole content system.

    Speaking of complexity, I experienced major cognitive dissonance at the mention of a “Trump/Sanders wing” of American politics. True, both of those men have criticized free trade, but for vastly different reasons and with vastly different agendas in mind. By conflating Trump and Sanders, even if only to make an illustration, you’ve kind of done the very thing you’re trying to warn against: you’ve overlooked (or misunderstood) a complex aspect of American politics and forced me, the reader, to untangle it.

    1. Mark Baker Post author

      Thanks of the comments, Larry.

      There is a bit of a chicken and egg problem here. If the goals vary with each project, can we afford to create a brand new content system for each project? Or do we seek to create a content system that is highly adaptable so that it can be adapted for different goals in different projects? Systems usually outlast projects. It has to be a pretty big, and pretty long-term project, for which you can design an entire new system. (In this regard, the lean movement has a lot to teach us about the value of being able to retool quickly.)

      But when I say content system, I don’t just mean the publishing tool chain. I mean the distribution of responsibilities in the whole content organization. Whose role is it to set content goals? How are those goals formulated, communicated, and tested? Whose role it is to model content for various types of documents? The writers? The information architect? The content strategist? How are those models, formulated, communicated, tested, enforced, processed?

      All these are questions about the partitioning of the content system. You need answers to all these questions before you can actually sit down and do goal setting or content modeling.

      At the same time, the goals you set and the models you develop can influence how your system is constructed. Design for adaptability often proves to be superior to optimizing for any single capability. Part of looking at the big picture is acknowledging the unpredictability of future goals and the models that support them.

      As to Saunders and Trump, politics makes strange bedfellows. (And they will vehemently deny ever having been in bed together!)

  4. Michael Andrews

    Mark, I want to add a different perspective. I feel the debates about structured writing are often misguided when they focus entirely on content production. People have different views about the efficiency of structured writing: is it more efficient to be able to reuse content, or does trying to do that entail more complexity than it is worth? These debates get rolled into issues about who is more valued: the needs of the individual writers, or the needs of the publisher as a whole. I’m not a writer, so I don’t focus on those issues. To me, the issue that matters is whether structured writing best serves the needs of the users of content. Everyone, whether writer or publisher, should be focused on that. Structured writing can either deliver a wooden content experience with repetitive, cookie-cutter content. Or it can deliver highly specific and relevant content where people get precisely what they want, and no more. When everyone on content teams have a shared purpose for structured writing, then merits of different approaches can be evaluated against common criteria.

    1. Mark Baker Post author

      Thanks for the comment, Michael.

      I agree wholeheartedly. Content quality should be the first concern of structured writing. And when it is implemented purely with process efficiency in mind — a failure to treat the system as a whole — the quality of content frequently suffers.

      The irony of this is the when people do create systems for purely process efficiency reasons, they are usually significantly less efficient than systems that are designed primarily with content quality in mind. The is partly because systems designed for content quality help writers write more efficiently, but it is also because they do a far better (and far simpler) job of capturing the metadata that can drive efficient automation.

      By far the biggest offender in terms of narrowly focused approaches today is content reuse. Its emphasis on creating reusable chunks often leads to the creation of monotonous and poorly integrated content. The problem with narrowly focussed approaches is that they usually take the simplest most imperative approach to implementing one function.

      A holistic approach to structured writing requires that you consider many different functions and accommodate all of them. This pushes you to a more declarative design which may be conceptually less direct, but which turn out to be far simpler to implement and understand once you have your head around it.

  5. Diego Schiavon

    Dear Mark,

    The following typos caught my attention:

    every freer trade
    both is rhetorical excellence.

    Sorry about that: job conditioning.

    Having said that, thank you for the post. Partitioning of complexity is an interesting viewpoint on which to think about my job and the relationship among readers, documentation and writers. I am looking forward to reading more in your book.

    1. Mark Baker Post author

      Don’t apologize. I always appreciate the proofreader’s eye. I just wish I had one myself!

      Fixed now. Thanks!


Leave a Reply