The future of learning: Why xAPI is important

Shelly Blake-Plock from Yet Analytics joins Robin on the podcast and talks about how xAPI's data recording reveals streams of experience in learning, and the broad scope of possibilities for the future of learning.

Subscribe using your favourite podcast player or RSS

Subscribe: Apple PodcastsSpotifyGoogle Podcasts | Amazon MusicAndroid | Stitcher | RSS

Sign up to our email newsletter to get podcast updates

Links from the podcast

Transcript of The future of learning: Why xAPI is important

Welcome to the Learning While Working podcast. I'm Robin, the host of the podcast. This is another podcast in our series on xAPI. In this podcast I'm talking with Shelly Blake-Plock, one of the founders of Yet Analytics. Shelly talks about how he and his team came to need xAPI when they wanted to record informal learning.

Shelly gives some great examples of the power of xAPI, and there's one moment in the podcast where he talks about combining heart rate data with simulation data, that triggered for me a whole lot of possibilities around monitoring different types of data as people are performing or practising.

One of the really nice things about Shelly that's different to a lot people working with xAPI is he has a really nice way of thinking about it in terms of it just being activity data over time and really nothing more than that. That then has a lot of other powerful things about what can be recorded, what can be looked at and what can be thought about.

I really hope you enjoy this podcast. It was a great conversation with Shelly about why xAPI is important.

Shelly, welcome to the Learning While Working podcast. It's great to have you here today.

Nice to be here.

A fairly open question to get started in this conversation is, what really sparked your interest in xAPI?

I was working with a team on a social learning platform for educators and the goal was to help educators pull together both formal and informal learning from the everyday PD that teachers and educators do. As opposed to sort of the more structured PD.

This has been a long time coming. I had been a classroom teacher and then I was teaching in the university, the School of Education at Johns Hopkins, working with Baltimore City Public Schools teachers strategies on implementing education technology and social media into their practice, both from an instructional point of view and from a professional development - a self-guided professional development point of view.

What we quickly found was that because of the great surge that had occurred in the development of ed tech, sort of going back into 2007, 2008, sort of coinciding with the mainstreaming of social media, particularly Twitter, which had a lot of sort of both cachet and early adoption within the education community as a means to connect with other people to help find way-points as you were doing this type of less structured, more self guided professional development. As those things were all coming together: increase in the number of applications available, increase in the connections between people via social media, we saw very quick, an almost exponential rise in the amount of learning data that was being produced in classrooms and through these professional development practices. Very quickly, as we were putting together this platform, we were overwhelmed by the variety of applications, the variety of data sources, the variety of ways that people were using and purposefully misusing technologies to do things their own way. It creates this big knot of data that had to be unwound.

When we got involved with the Experience API, xAPI, it was really from the point of view of how can we use an interoperability specification for the purpose of helping to untangle - provide some structure to all of that different type of activity data that was happening across all those different apps, all those different events, all those different situations.

Then, actually, once we looked at that, we saw two things. One was that there was an enormous amount of potential for technological exploration. Secondly, in unpacking that data, you were sort of unpacking or uncovering the types of experiences and the types of experience pathways that the teachers - the educators who were using these apps - what it was that they were going through in terms of their own practise. It really hit me in terms of why xAPI mattered and what we were going to do as a team to address that in stepping back and saying the combination of technologies that can help to unpack the data and the fact that what you are unpacking are streams of experience, provides a means to create these really meaningful activity profiles around learning, around experience. That was sort of the spark that led us to much deeper investigation. Ultimately, the build-out of a company and products and solutions, really digging into what does it mean to produce the data of experience?

xAPI is almost the glue that holds together an ecosystem sort of approach of learning. People don't actually learn in set ways, that they find their own pathways through sets of resources and experiences. Also, I want to explore something a little bit, which is an interesting thing - I really like the idea that learning happens when you articulate and possibly collaborate around an experience. You don't just have the experience. It sounds like almost what you're talking about it that in that project that sparked all this, with that group of teachers, that they're actually starting to do this process of actively collecting what they're experiencing and thinking about that in a different way?

Yes, I think that's right. I guess there are two things. One is that xAPI itself and this way of thinking, which I think is maybe even more important, this way of thinking about data, it sort of comes at it from the point of view of - a friend of mine calls it structured flexibility. It gives an element of structure within which there is a great amount of flexibility and adaptability. When I think of that, we're talking about it in terms of learning, but I think it's probably the most commonplace sort of experience that we have in the digital realm. Profiles of our activity are used on Facebook, and used on You Tube, and used on Twitter for the purpose of helping advertisers get their message to the people they want to get their message in front of. Facebook collects all the things we comment on, all the things we like, all the things we post, all the things we share. It creates a profile of the way we interact and things that we like and don't like so that the advertisers have profiles to target with their advertisements.

I think that we all understand that that's happening just under the surface when we use these services. From the learning perspective, if we take that as a paradigm, where in collecting activity and experience in a time based format - we call it the Facebook timeline - it's happening over time. It's creating a chronology of our experience, a chronology of our activity. We're able to take that in the learning space. Instead of using it for the purpose of targeting content around an activity profile, we can use it for the purpose of targeting personalised instruction. We can use it for the purpose of targeting personalised content, based not only on what you may indicate as an interest but what your experience pathway suggests is content that makes sense for the type of experience, the type of activities, the types of things that you actually do that are tracked across that sort of system and contribute to your profile.

I think that one of the things that the Experience API allows for is the gathering of data from multiple platforms, but not just for the purpose of putting it into a box. Not just for the purpose of contributing some fancier graphs to an LMS, but rather towards a purpose of helping learners and instructors, mentors and mentees, employees and managers to leverage the same sort of activity data paradigm for the learning and training space that is already used and is manifestly mainstream in all of our lives in the digital media and social media space.

In actual fact, there's another podcast which is really about how what marketing can learn from, sorry, what learning can actually learn from marketing as well.

Just a slight technical diversion, I think there's a really interesting thing here Shelly, in learning we think of xAPI as the standard. It's actually built on another standard, it's actually built on the Social API, which as designed for doing exactly the timeline style things that you're talking about. It's really fascinating that we then bring that back in.

I'm actually going to make this sort of interesting - I'm not quite sure if this is going to work in my own head. It's almost like xAPI adoption Version 1, which is around bringing in different data sources, measurement and evaluation. Then there's this really interesting, exciting thing which at Sprout Labs, we're starting to work on at the moment, which is xAPI Version 2, maybe. Adoption 2. Which is collecting data in, self assessments, assessment work, interests, and then being able to really build personalised learning experiences based on the data that's going in there. It was sort of interesting in this particular - we're starting to get started on this particular project and the client brought up, "Can we do this in SCORM?" I just went, "No."

It was one of those moments where it was possibly technically possible, but the amount of complexity that we started to get into when we actually started to think it through - we went, okay, no. In actual fact this whole log and stream and this really elegant format just makes this so much easier and so powerful.

The other really surprising thing is we had had discussions with educators previously around this particular adaptive learning and personalisation approach and it just became really complicated. This particular time, in a couple of situations actually, it was people who were coming new to digital learning and they just expected it to be personalised to a particular person. This was sort of an enabling technology that they don't even really need to worry about. It's just the personalisation layer of it.

I think that another piece of that has to do with the type of data that we're talking about collecting. I'd mentioned that when my team first got into this, it was very much around people, who were using our prototype platform, were collecting information using lots and lots of different apps. The majority of them were in the ed tech space, but some of them were some social things.

Very quickly, once you unleash the ability to use activity data defined broadly within the learning paradigm, things get weird real quick. I'll give you an example. The very first demo of xAPI that we ever did was now over two years ago at a conference called ITSEC, which is a conference all about military simulations and training. What we did was we set up a very simple flight simulator and the flight simulator, as the pilot traversed through the simulation, we were recording all of the activity that the plane was going through, how fast it was going, how it pitched, what direction, going up, down, crashing, etc. All of those different elements, all those different verbs, were being captured as xAPI statements and being pushed live back into the LRS. Then we had a data visualisation layer that was sitting next to that, essentially, that was mapping out what was happening within the simulation. Literally creating a map, like flight paths and things like this, as a visualisation.

Right there, you get to something that's like, wow, that's kind of interesting, because you're taking activity that's happening within the digital realm and you are mapping it in a way that's sort of very different than what we would traditionally think of in a learning environment.

Then we went one step further, because what we did is we had hacked xAPI into Android-ware wristwatches, those smart watches. What we were doing was we were using that to capture the heart rate of the people as they were doing the simulation. We were also transforming the heart rate itself into xAPI. What we could tell was that this actor, as they were involved in this simulation, and the simulation is being mapped out in a visualisation as it's happening, this is a record of their heart rate and their response to stressors within the simulation. Live, in real time, captured in the same data store as it's happening.

That's interesting because we're working not just with the interoperability of those two different data sources, the simulation and the Android-ware. Of course we are, we're collecting those two things and using xAPI to push that somewhere where we can work on those two different data sources at the same time. That's fine. What we're also doing, when we're talking about interoperability, is we're linking what I'd like to think of as almost like the interoperability of digital and physical experience within a learning situation. The ability to pull digital information and biometric information at the same time. The ability to pull digital information and environmental sensory information. The ability to pull digital information and look at it live in motion in the real world. Those are the types of interoperable connections that don't just demonstrate the power and potential of xAPI as a technical specification, but which suggest the value of xAPI and the potential to completely disrupt the way that we design instruction. The way that we design learning experiences, therefore, is not limited by learning applications, or ed tech, or e-learning content. Rather, everything which produces an activity feed is fair game for the instructional designer to use within the type of learning creation that they're building.

A huge amount of possibilities are just running through my mind, including what my heart rate would have been during a role play this morning. The facilitator was deliberately insulting the participants and seeing how we responded! It was just interesting, because it would have been a nice way of being able to get personal feedback and for her to give feedback afterwards. That's a powerful thing.

You're almost seeing it as this possibility of being able to bring together all data into a universal format, be able to understand interrelationships.

Well, I would temper that in this way: I would say that especially a bit earlier, once xAPI formally hatched as xAPI, there was a lot of talk about should this be used to pull together IOT data? Should it be used to pull together different types of business intelligence data? I would argue that actually it is perfectly fine to keep those different data sources that handle those different types of data well, to keep them running as is and allow innovation to occur within those spaces. I don't think it necessarily does anyone any good to have smart temperature gauge data being collected as xAPI just because you can. Rather, I think that xAPI is especially good at capturing the nuances of an activity that an actor engages in. Whether that actor is a human or whether that actor is a robot, it's capturing the activity that is engaged in.

That's a specific way to look at data. Not all data does that. Rather than say, xAPI becomes the universal hub of all data, I would say instead that xAPI does a very good job expressing types of activity that often are related to formative events. We have an event based data store that can capture these things. Rather than try to wrench other data types into that format, we've been doing a lot of work on our team at Yet Analytics around using contemporary developments in big data stream processing to bring xAPI in as one of many types of data streams. There, when you're looking at things, for example, from sort of the point of view of the Kappa architecture and streaming data, xAPI doesn't have to be the be all and end all of everything. Rather, xAPI can be one of many contemporaneous, contributing factors, which I think tells a much bigger and more compelling story.

Especially when we talk to large corporate organisations, they're not very interested in translating their existing business intelligence into xAPI. What they're extremely interested in is taking their learning and activity related data and their activity performance level data at the employee level, finding what makes sense for xAPI there, and then turning that xAPI data into another form of business intelligence data and running that through these more comprehensive stream architectures. I think it's a developing and maturing approach towards how the specification sits into a broader business architecture.

I'd actually say one in three of our clients at the moment are sitting there doing: that's just somewhere we store data, we actually want that out, we want to get it into our business intelligence systems and then we'll do our analysis work back into those systems.

Shelly, this has been a really fantastic conversation. A question I've been asking everyone with the xAPI series, it's a really nice spot to quite often finish the interviews with - and this has been a really rich conversation: If someone's getting started in Experience API, what would be your advice?

Begin with design. Don't start by looking at what technologies are available out on the market. Don't start by looking what vendors do this, versus what vendors do that. Don't start by any of those things. Rather, start by getting a piece of paper or a whiteboard and figuring out whether you can use, whether you are already using, or whether it would make sense and open up some opportunities for you to get what you want out of your programme. To use activity time, the ability to graph profiles of learning pathways, more nuanced approaches towards understanding how people learn and what content they use to get there. Start by doing that. Map that out.

Then challenge yourself to think, "Is there a way that I could leverage activity and experience for the purpose of making my programme, my organisation and my people better?" If, at the end of 90 minutes, you've got a whiteboard full of ideas and you've actually come to the conclusion, "Yes, this is something that I would really like to pursue," then there are many ways to reach out through the open source community. There are ways to reach out through people like me and my peers across the industry and to actually get involved in putting this in motion.

Cool. Thank you Shelly. That's a really nice way to finish the interview. It's been great having you on the Learning While Working podcast.

Thank you very much Robin.