One of the mantras I’ve championed is "curation over creation." And I still believe it. It’s just that the task has changed. For the better, notably. In particular, there is not only backup support for turning content (created and curated) into learning paths, but there’s also support for curation. There are still issues to be addressed, however. It’s time to look at the directions in curation.
As background, in the context of learning, knowledge plays a particular role. By itself, it’s not enough to facilitate learning. Learning needs to be engaged in meaningful practice to develop new skills. However (and this is why lectures at conferences can be relevant), if you’re engaged in meaningful work and have a question or interest, the right content can facilitate learning. You’re already in practice, so the content serves as guidance, either just-in-time performance support or as ongoing development.
Why curate learning content?
The issue of why was really brought to life for me during an engagement with a manufacturer. Their manufacturing was complex enough to require multiple pieces of software to support design. And, not only was there was more and more relevant software, but each program was getting more and more complex with added features. The internal unit responsible for supporting software training and support couldn’t cope.
The engagement was about investigating social uses of software, and in this case the unit had devolved the responsibility for supporting users to the users of the software. They took what they had, and migrated them to collaborative documents (wikis, at the time). Then, they supported the users in maintaining and updating the information. Certain users were the ones who had real knowledge about how the software was used, and interacted with the vendors. The approaches taken were both content creation and curation, but by the users with facilitation by the learning unit.
In another instance, the head of learning at a large social media organization similarly shared how he coped with the quantity and diversity of learning required. The solution was not to try to meet all needs with a course. Instead, if it wasn’t company-specific he would opt to recommend books that addressed the issue. (It could be other resources as well.)
If it’s not something proprietary to your company, it’s likely generic and somebody’s written about it. And at a finer grain, there’s likely an answer for most needs (hence the success of Google).
And, if you’re already engaged in practice, you don’t need a full course. You know why it’s important, you know what you need, and you just need it. And even if it is a course that’s needed, again unless it’s proprietary, why build it? Truly, you should be developing courses only for things that are specific to the company, move novices to the minimally useful performance level, and are likely to be stable enough to be worth the time. Then do it right.
Approaches to curating training content
Once you decide to curate, how do you go about it? Let’s be clear, there are many things to curate. Courses, job aids, and independent learning resources. That latter, arguably, is the most dynamic, since new things are popping up all the time, faster and faster.
The first approach is, of course, by hand. David Kelly has synthesized much about the specifics. This can be either the learning unit or the staff. The former is typical, and the latter has been viewed as problematic. But is it? The argument against employee curation is that they don’t know learning. True, at the novice level. But once they become practitioners, they begin to know what they need, and why it’s important. They may tap into their own networks, and find something relevant. And their output can be facilitated, instead of subsumed.
You and your employees may find things on your own, and recognize the potential value for others. You’re also likely tapping into expertise from their community of practice. That includes sources such as society sites and e-zines that publish specific material. There’s also what comes through social media, including community groups and trusted relationships. Building a suite of individuals to follow is important.
It’s also important to facilitate your community to also learn how to be good trackers, and generators, of information. Harold Jarche’s Personal Knowledge Mastery is a good model of practice here. Choosing the most relevant content for the community is important, as is recognizing that your employees may belong to multiple communities that may need to be tracked.
What's next for content curation?
There’s a new dimension, however. Increasingly, there’s support for recommender systems, that provide curated content in a variety of ways. Using algorithms, machine learning systems, and other AI approaches, systems now can push content based upon a variety of models. Several mechanisms can come into play.
There are two underlying models: push and pull. If an individual searches for something, a system can look for appropriate things to recommend. Alternatively, a system can be proactive, for instance tracking a learner's development path and finding relevant things to suggest. This is in lieu of a mentor/guide doing the same.
Like Amazon and Netflix, one mechanism is to leverage what other people are visiting. If many people are visiting a particular piece of content, it can increase the likelihood it will be shown to others. Thus, the most popular stuff rises to the top.
A related approach is what people rate highly. Here, people have to be rating what they come across or share. It’s more work, but it’s slightly more valuable information. Both are data analytics-driven, which isn’t bad. But, can we do better?
Material can be auto-tagged for categorization. This can be via one or a combination of methods. This includes keyword identification, source parameters, or more. It could also be a combination of automatic and manual effort.
A more complex approach is having the system actually do semantic analysis of the content, and then push it to queries that match. That is, it parses the content, and therefore has a way to document what the content is about.
On top of this, it’s now possible to have the engine answer questions on the content. So, for instance, you can have a system semantically parse the HR policies of an organization, and then answer questions about it.
In fact, such systems can actually generate questions. Thus, you can prepare learners for meaningful practice automatically. Note that such systems, as yet, can’t actually generate meaningful practice, only knowledge-related tasks. Technically, they don’t truly understand what they’re processing, they’re playing relationship games. And that’s ok for purposes "in action." It’s just not sufficient by itself.
These auto-curation systems can meet both immediate (performance support needs) and longer-term personal development (learning). There are still limitations, but they’re an increasing powerful adjunct for the task of supporting organizational performance and development. Understanding them, and recognizing the opportunities, is an important component of the new L&D workplace.