Two recent blog posts, Structured Content is Like Your Closet by Val Swisher, and Content Strategy Can Save Us All From Slobdom by Meghan Casey, both illustrate how content management works today by analogy with a well organized closet. It is a perfect metaphor for current content management practice, and provides the perfect starting point for examining what is wrong with the current state of content management, and why (as I noted previously) people hate their CMS. The closet analogy shows us just how much people have tended to model electronic systems for managing bits on the same principle they use as humans to manage atoms. The basic principle of closet organization is to place like with like. Thus, we place shoes with shoes, coats with coats, and hats with hats.
But there are problems with like-with-like organization. Things can be like each other in different ways, all of which matter at one time or another. For instance, you could have red shoes, red coats, and red hats, and it is likely that you would want to wear these things together, to create a coordinated ensemble. But in a closet organized by garment type, rather than by color, you have to go to multiple places to find a red garment of each type.
Other complexities arise: do you place winter and summer clothing together by type, or do you segregate the seasons, at the expense of having two places for hats, two places for shoes, two places for coats?
The closet approach also has problems with scale. A typical closet belongs to one person. What happens when you want to start sharing? Do you create one closet for the whole household? For the whole street? For the whole city? A city-wide closet could eliminate a huge amount of redundancy but it would also create big problems in terms of storing like with like as it would introduce more aspects of likeness, such as size, into the organizational schema.By increasing the number of items in each group of like things, it would take individuals far longer to find the particular set of clothes they want to wear today (even without factoring in the commute time involved).
In the physical world, there is no way out of this dilemma. When you store like physical things together, you must choose one aspect of alikeness as the principle axis of organization at the expense of other aspects of alikeness. You can choose garment type, or color, or season, as your principle axis of organization, but you can’t choose them all equally.
But in the digital world, that restriction exists only if we impose it on ourselves. In the digital world, like can be stored with like for every aspect of alikeness without limit. Red socks can be stored with black socks on the basis of their sockness, and with red gloves on the basis of their redness, and with winter wear on the basis of their warmth.
In the physical world, putting something next to one thing means moving it further from another thing. In the digital world, you can put something near to an infinite number of other things in an infinite number of dimensions.
Organizing digital assets like clothes in a closet, therefore, is imposing physical world limits on your digital storage scheme. We do it not because we are constrained by the media, but because our thinking about how to store things is constrained by our habituation to the physical world. But subjecting ourselves to that constraint robs us of much of the power of the computer to manage, organize, connect, and display information.
The ability to place an object near to an infinite number of other objects in an infinite number of dimensions of alikeness is one of the fundamental properties of digital computers that makes them so powerful for managing information. The Web is but one example of this power. On the Web, any page can be placed near any other page simply by creating a hyperlink between them.
The only thing that makes any two points on the Web distant from one another is the failure to create a link between them. Of course, you can’t make every web page near to every other web page in this way, because it would involve creating an impossible number of links, but you can move a web page next to any other web page, for any reason, without in any way affecting its nearness to other web pages.
In the digital world you can move two things closer together without moving them away from any of the things they are currently close to. Relational databases (which, of course, lie behind many web sites and CMSs) also show this property of enabling nearness to an unlimited number of things in an unlimited number of ways. The relations expressed in a relational database are expressions of nearness between two sets of data, and there is no limit to how many relations you can form, and creating a new relationship does not make the existing relationships more distant.
And yet, though we work in the digital world, though we use the web and databases every day, we still do content management as if nearness in the digital world was constrained in the same way that it is in the physical world.
Why do we do this? In part, perhaps, because when we view relationships in the digital world, we do so on physical screens, where physical-world constraints on nearness apply. But the representation of nearness on a screen is just one projection of one set of relationship in the underlying data. Alternate projections can show other dimensions of nearness, but only if the structure is in the data to allow the machine to discover and display that aspect of nearness.
Ironically, the way that you free yourself to express nearness in multiple ways is to free yourself from thinking of relationships between objects in terms of nearness at all. Rather than thinking in terms of nearness, you have to start thinking in terms of alikeness alone. You must learn to express alikeness in the data itself, semantically marking up content to capture each dimension of alikeness that is relevant to your content.
In order to organize content on more than one independent axis of alikeness, you have to become location independent. You can’t express alikeness through location if you have more than one independent measure of alikeness, so location has to go away as your organizing principle.
The interesting thing is that this can actually free you up to use the ordinary file system for shared space, which we see very commonly in the programming world, in which even the largest projects are managed on the file system via a version management system. The reason this works is that while each individual programmer organizes their code on their own individual file system, the overall organization of the code into a product isn’t done that way.
What happens is that the linker builds a database of symbols from all of the files in all of the parts of the file system included in the build, and uses that database to link the parts together regardless of their location. So, the overall system build is location-free, and the individual programmer is free to organize their own files on the file system.
The reason that content management systems are so big, so expensive, and so frustrating to use is that supporting the closet model of organization for a large shared collection is inherently difficult. It artificially makes some properties more important than others, which makes navigation, aggregation, and linking difficult, and it simply does not scale well when you start adding more objects and more users.
Isn’t it time to bring your content out of the closet?
Mark – What a fabulous post. I agree with you one-million-percent. I’m really glad you took the analogy of the closet and truly expanded it in a way that is extremely accessible to everyone. The organized closet metaphor is a good one for understanding WHY you need to structure your content. But, as you say, it completely falls flat when you try to describe HOW to organize your content.
This is a brilliant post. Thanks much. ~ Val
Pingback: Thinking About Content in Multiple Dimensions | Content Rules, Inc.
Mark – A brilliant post!
Here at onion.net we share your verdict that it’s “time for content management to come out of the closet”, and we have tried to avoid the pitfalls of conventional CMS concepts. So please let me share some of my thoughts:
Many popular content management systems are based on a “page-centric” concept. Content objects are basically containers which contain stuff to fill the placeholders in a site template. The user interface enables the users, to edit contents in the context of a (static) website’s preview. This has been a quite successful approach, to edit the three dimensional web sites of the past (i. e. a fixed grid with varying contents).
But today we are facing new challenges:
– With the advent of mobile the web has suddenly gained new dimensions (i.e. the three conventional dimensions plus screen size, resolution, bandwith, interaction device, …).
– Enterprise customers search to consolidate projects on a single infrastructure.
– There is an increasing demand to re-use content objects accross sites, in order to reduce the maintainance effort.
– Targeted content is promising to deliver a dynamic response to the user’s journey, often referred to as web engagement or -experience management .
– The next trend is possibly the semantic web.
All of these challenges indicate, that we need to re-think how content management systems are organizing information, i.e. the content objects – and it is clear that “potholes in a template” won’t do the job.
The first of our responses to this challenge has been a “content-centric” approach. The content objects are now organized independent of their presentation. This allows to re-use the contents in many differnt ways. In order to overcome the restrictions of a “flat” relational data base model, we have chosen to describe the information structure by XML schema. Then, the content presentation is based on rules, namely XSL templates. This allows to aggregate content elements from several (XML) sources and deliver dynamic results (i.e. targeted content) in different formats. And the best thing: the content controls how to present itself.
The experience from a range of customer projects has proven, that a couple of such basic concepts lends to an enormous gain in flexibility. And yes, we like the concept of linking information items (which is supported by drag-and-drop at the user interface). At the same time the XML schema can control, which types of information can be linked and which can’t. Thus, a given content structure (the one that has been defined for the specific project) is enforced, and the data is protected against human error.
I’m not quite sure if we did fully come out of the closet already, according to your definition. But I’m pretty sure that at least, we did blast the door away.
Cheers, Bernd
Full disclosure: I’m product manager of onion.net, a CMS framework “Made in Germany”.
Thanks for the comment, Bernd. Those certainly sound like interesting features, though when I look at your web site I find it very hard to figure out what your product actually is. Is it a development platform. Is it an out of the box CMS? I can’t tell. (Your are not alone in this, BTW. I find most CMS vendor sights pretty much incomprehensible. Not sure what that says about the industry.)
Sockness! Oh, my. This one definitely will go into my vocabulary for special occasions 🙂
Your post describes a general problem of humanity. I’m talking about Information Management. People tend to put things into order the way they are used to. And don’t you dare questioning that!
This is the problem with every structure in your office. Every single structure in your office grew in the mindset of a few individuals. They imprinted the structure with their understanding of “putting the things into order”.
If you look at those structures, you will only find those useful, that share a common denominator with your own mindset.
Yesterday, we discussed in a team meeting the future structure of our the organization of our support materials.
In fact, the current structure evolved years ago, maintained by one person. The guy mainly using it at the moment, adapted to that structure over time and also put his own mindset into it. Problem is, no one else besides him does find anything in that structure. It’s like sorting apples and pears by their sockness. If your understanding of sockness of an apple is different to the understanding of the guy sorting, you will never understand why the apples and pears are sorted the way they are.
So, this is a problem of how to organize information. We need to define our structures in regard of common denominators.
This is true for every structure we use, be it structures of a CMS, folders on a shared drive or whatever.
Those denominators have to be simple and they need to be accepted by everybody using them.
Take the closet example here. The first common denominator here would be, that stuff is stored in closets. The second one would be, that stuff is organized together with likes. This will do for a low scale already.
If your scale increases, you have to find new denominators, like size, brand, or season, depending on usability (a new category is useless, if only one or two individual pieces fit into it).
In the digital world different denominators can share the same content, which you pointed out Mark, increases findability a lot.
If sock is categorized by its attributes, in a digital world you could find a pair by size, color, type or material – at the same time.
Actually, Alex, I have to disagree. It’s not a general problem of humanity. Much of the rest of the world seems to have got out of the closet model long ago. Database developers, and database users, don’t work on this model. Software developers don’t work on this model. Big sites like Amazon and Wikipedia don’t work on this model. The Web itself, taken as a whole, does not work on this model.
There were many who said, back in the day, that the Web could not possibly work because it was so disorganized. What the Web has shown, though, it that orderliness does not scale. The Web is not orderly, but it is eminently filterable. Filterable scales.
So, the content management industry, and writers in particular, are in many ways the last holdout of the closet model. Perhaps it is because, as David Weinberger pointed out in Everything is Miscellaneous, “hypertext subverts hierarchy”. The organizing of content gave the writer power. Today, the reader can organize content for themselves, and the writer’s power is diminished. We still keep trying to take the power back, still keep insisting that we have to organize it for people. That is what keeps us in the closet.
If your point is valid then the language used to search for specific information needs to also be more specific. Naming strategies i.e. metadata is often so similar across platforms as to defy finding information by name:type. If your search algorithm also skews results to: most requested/most used, then good luck to you.
It is not a matter of writers having more or less power, it is a matter of having a living intelligence create information that a living intelligence, in the end, will use.
If your point is valid then the language used to search for specific information needs to also be more specific. Naming strategies i.e. metadata is often so similar across platforms as to defy finding information by name:type. In your example of Amazon.com maybe the method works fine for finding toasters. Try searching even you local company for a common term like “configuration”. If your search algorithm also skews results to: most requested/most used, then good luck to you.
If your set of metadata is designed for a set of users, say car buyers them your theory is appropriate.
It is not a matter of writers having more or less power, it is a matter of having a living intelligence create information that a living intelligence, in the end, will use.
Mark, as usual, you touch on important and timely themes here. I like the way Bernd Burkert is thinking in his response too.
I tend to share the notion that the semantic web is a good direction to follow, but there is a huge problem. The semantic web is based, for a large part, on ontologies. Ontologies are just the way to show the various relationships that you talk about, including “sockness.” 😉
However, to do ontologies properly involves enourmous human effort. It takes humans to make intelligently machine-readable relationships. Most researchers working on semantic webs today admit that the World Wide Web will probably never be all semantic – only the parts of it that people really see as being worth the labour investment.
What do we do about the rest of it? What methodologies can we find that might be more realistic in terms of effort, so that we can benefit from the kinds of relationships you are writing about?
Might we be able to crowd source relationships by offering the possibility to in-build user-suggested links, in a wiki or other similar structure?
Could we, in fact, crowd source universal ontologies? What concept!
Thanks for the comment, Ray.
I think universal ontologies are a pipe dream. All language is local — something I have blogged about before (/2011/05/23/findability-is-intractable/).
But locally, one can create focused, structures, taxonomies, and ontologies that allow you to organize and manage your own content dynamically. That is how you get your stuff out of the closet.
Much of the focus on taxonomy and ontology in the content management tends to be outward focused — trying to provide a universal way for people outside to find you content. That is doomed to fail, and often just makes things worse.
On the other hand, if you focus your content structure and labeling internally, focusing on creating the maximum semantic richness and integrity in your internal content store, you gain both the ability to organize and present your content in any way you like, and the ability to map your internal taxonomy to any taxonomy an external client may want to receive your content in.
Precisely because there are no universal ontologies, the way you make your content available semantically to the widest audience is to structure it as richly as possible using your own semantics, so that your have the maximum amount of information and structure on which to base an output tailored to the semantic needs of each customer in turn.
An additional benefit is that if your semantics are locally focused, meaning that your staff are thoroughly familiar with them, with what they mean, and with the distinctions that they make, you get much better data entry, and much better auditing capability.
Pingback: Organizing Hard So Information Is Why? | How to Write
Mark, Between your post and Val’s, my brain got to buzzing beyond what a comment here could hold. Thanks for the stimulation. See “Organizing Hard So Information Is Why?” (http://howtowriteeverything.com/organizing-hard-so-information-is-why)
Interesting take, Marcia.
I agree about tags, though I would make a distinction between tags as a folksonomy device where many people can tag the same picture or topic in any way they seem fit, and structured and typed tagging.
Each star in the sky is cataloged by astronomers using a large number of strongly typed tags. See the right hand column of this Wikipedia article on Alpha Centauri for some of the typed tags that astronomers apply to stars: http://en.wikipedia.org/wiki/Alpha_Centauri.
As I noted above, organization does not scale. Cataloging scales. When you get past the point where there are either too many different aspects of alikeness or simply too many objects, organization break down. But cataloging, provided you have the right fields in your catalog, scales infinitely. A computer can dynamically present any kind or organization that your cataloging supports.
We can’t organize the stars, so we catalog them. We have to stop organizing content and start cataloging it.
Delightful post, Mark.
I routinely get myself into trouble by observing, usually in public, that content management systems (CMSs) almost always exhibit a disregard, if not outright contempt, for “content”. The charge being levelled in this case, and none too delicately, is that these systems are propelled by a desire and even a need to impose very specific restrictions on the structures and relationships that the content under management will exhibit. With rueful predictability it usually takes a few days, and sometimes only a few hours, before the intrinsic qualities of content to strain against these boundaries and to eventually break across them in mischievously inventive ways.
Now as you know, these observations figured prominently back in the heady days of HyTime, Architectural Forms and Topic Maps (as in the ISO standard as opposed to what we find in DITA). And, to Ray’s point about crowd sourced universal ontologies, this was precisely one of the areas where the Topic Map movement directed a fair amount of attention and fashioned support (at least technically) for sourcing and merging public ontologies.
Back in the summer of 2009, I wrote two posts on one day the first of which lines up with this one of yours to some extent – The Trials and Tribulations of Content Management. The second post on that day attempted a frontal assault on the question: What then do we mean by the word “content”? Although I am due to revisit the exploration that I then put forward, and I know I have a few additions to make, I still stand behind my attempted answer. See The Truth about Content.
Thanks for the comment, Joe.
I too feel that the word “content” is an expression of contempt. As I noted in my post I am a Content Strategist (/2012/05/04/i-am-a-content-strategist/), content is a word for anything that fits in a box. Content management systems don’t manage content. They manage containers. And they generally want all the containers to be labelled according to a single schema.
It doesn’t work. What makes the “content” interesting and useful are its specific characteristics. You don’t use the same set of labels for stars as you do for flowers or for furniture or for musical compositions. It is not all “content”.
The tragedy of all the fruitless striving for universal ontologies, taxonomies, and schemas, is that in the process people largely ignored the enormous potential of local ontologies, taxonomies, and schemas.
In response to the post, I think what gets lost in the digital world (that’s harder to lose in the physical world) is that content still requires intensive thought and continued care over time. One of the problems with websites is that they create way too much digital “stuff” because they can, because it’s so “easy” – to create new pages, link pages together, add photos, etc.
We Americans are rarely taught to think strategically and instead we’re taught to think tactically. This is never more clear than on websites where tactics rule the day but weigh content managers down and retard the business at hand. I’m currently reading a book by Susan Weinschenk, “100 Things Every Designer Needs to Know About People,” and in it she says that there are three kinds of mental activity: cognitive, visual and motor. We think, we see and we do. The most difficult is cognitive – it requires the most mental energy. So my view is that we as web content creators lapse into “doing” (creating content) and avoid “thinking” (strategically processing content before it even gets published).
The result is crowded and distracting websites. I guess I’m unclear about the purpose of your post. It’s a fanciful idea to think that we’re freer digitally than in the material world but I don’t see it really playing out that way, online. I think the most rigorously effective online marketers and other professionals put tremendous thought into what they’re doing – they back it up with strategic rationale. They’re careful and methodical. They pay attention to organizational capacity before committing to campaigns they can’t sustain. I n other words, the most valuable commodity in the present day isn’t our online universe – it’s still our ability to think through what we’re doing before we do it.
Jo, thanks for the comment.
I’m not arguing for a lack of organization in the digital world. What I am arguing is that a model of organization taken from the physical world, which consists of putting things in “places”, and which is constrained by the inability to put one thing in multiple places, even if, by the rules that define the places, it belongs in more than one place.
Organization is the digital world is not done by putting things in places but by labeling things with metadata. The great difference is that while you can’t put something in more than one place, you can give it more than one label. You can give it as many labels as you want.
You can label things in the physical world as well, but the problem is that it is hard to retrieve things by label in the physical world unless things with similar labels are all stored in one place. In the digital world, no such restriction exists. Labels can be indexed and items can be retrieved by label, or by combinations of label, almost instantaneously, even if they are located in very different places.
In terms of displaying content to a user, this means that the website needs to be able to select appropriate content by label for the user’s current interest. This is how Amazon, and just about every other online retailer works, for example. Here is a smaller scale example of organization by label: http://autocatch.com.
In terms of cognitive load, creating, maintaining, and navigating hierarchies creates much more cognitive load than labeling. For an author to figure out where to place some new page in a complex hierarchy is very difficult, and for a reader to find a page in such a hierarchy is even more difficult. By contrast, people don’t have much trouble finding things on Amazon despite the millions of items it offers.
For an author, the discipline of labeling content, especially according to consistent labeling standards that make sense in terms of the information the author has available to them, presents much less cognitive overload than inserting content into a complex hierarchy.
The only place where hierarchy seems to win in terms of cognitive load, is in the initial conception or architecture of the site. Site designers clearly find it easier to think in terms of static hierarchies rather than dynamic selection and/or creation of pages based on data and metadata. From that point on, however, both for contributor and users, static hierarchies create significant cognitive overload that makes using the sight progressively more difficult as the information set grows.
One of the reasons that some content strategists want to start by reducing the amount of content on a site is, I suspect, that they simply don’t understand how to create a site with lots of content that is also easy to use. But it can be done, and some of the biggest sites on the Web, such as Amazon and Wikipedia, illustrate that it can be done. It is simply a matter of freeing yourself from the constraints of physical-world organization.
There is, of course, the issue of removing redundant or outdated content from a site, but here too, labeling wins. It is much easier to find and remove outdated content that is appropriately labeled, than it is to hunt it down in a hierarchy.
Pingback: Reusable Content is Like Your LBD | Content Rules, Inc.