Remote learning: why hasn’t it worked before and what can we do to change that?

Posted on 14-03-2020

Back in 2008, the business professor Clayton Christensen made a prediction that by 2019, half of all US high school classes would be taken online. It’s now 2020 and this prediction is not even close to being true, in the US or any other country.

Similarly, big predictions that online courses would completely disrupt university education have also failed to come true (notwithstanding success stories like the UK’s Open University who pioneered remote learning in a pre-online world). Why is this? Is it fundamentally impossible to get remote learning to work as well as classroom-based learning? Are there things we could do to improve it that we are not doing? Or is it all just inertia – will an external shock like coronavirus make us realise that we can make it work after all?

Remote learning and disruptive innovation
Christensen made his prediction based on his general theory of disruptive innovation. According to this theory, disruptive innovation happens in the following way. You have an incumbent – let’s say an encyclopedia publisher, the Encyclopedia Britannica – who has a well-established product with a well-established customer base.

Then a disruptive innovator comes along – in this case Wikipedia. To begin with, the innovator’s product isn’t as good as that of the incumbent’s. It’s not as high quality or as reliable. But it is much more accessible than the incumbent’s product – in the case of Wikipedia, it’s free and easy to access online. And so it has great appeal for people who, for whatever reason, can’t access the Encyclopedia Britannica. And whilst to begin with Wikipedia might not be as reliable or high quality as the Encyclopedia Britannica, over time it will keep on improving until it gets to the point that it is not just more accessible than the incumbent – it’s actually better too. And then bingo, the innovator takes over and the incumbent disappears.

How would this work applied to education and remote learning? In his book Disrupting Class, Christensen lays out the essential ways you could apply this theory to education more generally. Currently, there are already a couple of percent of students who, for various reasons, cannot access traditional schools. Christensen’s theory is that virtual schools will start out delivering education to this small group of students. To begin with, it will not be as good as what you get in traditional classrooms. But over time, it will get better until it ends up being as good or better as what is on offer in schools. More and more students will then move over to learning remotely, in virtual classrooms.

I have a lot of respect for the general principles of Christensen’s theory. Unlike a lot of business theories, there’s some good evidence backing it up, and some pretty good examples from a whole range of different sectors. I also think the general principle of what he outlines about education is at least plausible. So why didn’t his prediction come true? I think there are three, linked, reasons: two that can be solved, and one that is harder and possibly insurmountable. These are the three important principles to remember when setting up remote learning – and which, unfortunately, are often ignored.

Use high-quality, well-designed content
Too many online learning programmes assume that if you are studying online, you’ve got easy access to all the facts and content you will need via the internet. For example, Google’s teacher training resources tell teachers that “we have access to every bit of info we need at all times. It is therefore no longer relevant to focus on teaching facts.” This might seem plausible, but it is actually a terrible idea. We need facts in long-term memory to be able to think. Content can’t be divorced from skills. Imagine reading a novel where you don’t know what half the words mean: sure you could stop to look up every word online but how much pleasure and understanding are you going to get from the novel?  Technology shouldn’t be used to eliminate memorisation, but to make it more fun and efficient.

Not only that, but all online content is not created equal. It varies in quality, reliability and design. One of the side-effects of the common idea that “content doesn’t matter” is that it’s assumed that if students can’t find what they need online, then teachers can easily throw something together that will do the job.  But creating high-quality text is not easy. Creating high-quality multimedia resources is even harder. There is an extensive research literature on multimedia instructional design: Richard Mayer has come up with over 30 principles of multimedia design, and creating resources that make the most of these principles is hard.

Barbara Oakley & Terrence Sejnowski are the creators of Learning How to Learn, the world’s most popular online course. They used Mayer’s principles in the design of their videos, and wrote a paper for Nature about the process.  In relative terms, their videos did not cost that much – but they still required more time, money and specialist expertise than most teachers have to spare. We can’t expect teachers to create all this content. Ideally we need online learning platforms that integrate high-quality, coherent and well-sequenced content; in the absence of those, we need to curate the best online and offline sources of content.

Get the right kind of interactivity
Of course, on its own high-quality content on its own is not enough. If it were, then much older learning technologies would have totally solved the remote learning problem: the printing press and camera! Indeed, Thomas Edison predicted this in 1913, arguing that within ten years education would be carried out via motion pictures. His prediction was wrong just as Christensen’s was, because a teaching and learning are interactive. Good teachers do more than just present content; they are constantly questioning, listening, and reacting to their students.

Of course, modern technology has the potential to be interactive, and interactivity and personalisation are major selling points of many modern ed tech programmes. Unfortunately, they often interpret personalisation in a pseudoscientific way. Ironically, this can be seen in Christensen’s own work. In the same book where he makes his prediction about online courses, he also says that they will be transformative because they will allow students to receive personalised instruction based on their own learning style. But there is no evidence learning styles exist. Study after study shows that what matters is not the preferred learning style of the student, but the best learning style for the content being studied. But the education technology world is not listening to this evidence. Summit Learning, an online learning platform which is funded by the Chan-Zuckerberg Foundation and has been praised by Bill Gates, is also heavily invested in learning styles.

What about different types of interactivity – video-conferencing, for example? This might provide some benefits, but the technical obstacles shouldn’t be underestimated: expecting every student and teacher to have a reliable enough internet connection to allow for video calls between 25+ different computers is highly optimistic.

Instead of attempting to replicate the classroom via a video-conference, we could think of simpler ways of achieving the interactivity that is crucial to teaching and learning. Adaptive learning platforms are designed to personalise instruction – not on the basis of non-existent learning styles, but using information about a student’s previous studies.  The most complex adaptive platforms consist of thousands of questions and videos and detailed information about the student’s understanding. These often need some kind of teacher and student training before implementation. But other adaptive platforms are much simpler. Some are essentially just collections of flashcards combined with a spaced-repetition algorithm that displays each flashcard at the ideal moment.

It sounds simple, but decades of research show this is one of the most effective ways to study. Some flashcard apps are free, and some support offline use for students with unreliable internet connections. They also provide teachers and students with data on the time spent studying and patterns of strengths and weaknesses. While learning styles have no research backing them up but are absurdly popular, spaced-repetition is the exact opposite: it has enormous amounts of research behind it, but remains underused.

Remember we’re not brains in a jar
The final reason why remote learning hasn’t worked so far – and the reason that may be hardest to overcome – is that there may be ways, particularly for young children, in which the physical presence of a human is necessary, or other ways in which physicality matters.

For example, we know that young children pay attention when they hear adults speaking, but not when they hear the same adults saying the same things on a video, and we know that handwriting has value over typewriting when it comes to learning to read and taking notes. Teachers provide motivation, encouragement and supervision, and you can’t switch them off by giving them a one-star rating in the app store. There are ways around some of these issues: for example, at No More Marking, where I work, we assess student writing online, but it can still be hand-written and then scanned in.

The best programmes
There are some terrible online learning resources out there, but there are also some brilliant ones, even if they are not always getting the attention or funding they deserve. In my research for Teachers vs Tech, I had a lot of fun trying out different learning apps and talking to the people involved in making them. Here are some of my favourites. (I don’t work for or have any financial connection with any of the below!)

  • Duolingo: probably the most famous language learning app, with enough data to be able to change our understanding of how we learn languages
  • Memrise: another language learning app – one thing I particularly like about this one is that they have lots of videos of real humans speaking the language, so you can hear the different ways words actually get pronounced.
  • Smartick: comprehensive adaptive maths programme for 4 – 14 year olds
  • Anki: a spaced-repetition flashcard app. You can download pre-made decks, but most people I know use it to create their own. Very powerful but not particularly shiny or gamified – perhaps better for older students. I use this myself and I love it.
  • Quizlet: another spaced repetition flashcard app – perhaps more suitable for younger students.
  • Times Tables Rock Stars: Making times tables fun.
  • Hegarty Maths: this is designed to supplement school maths lessons by providing homework practice
  • Diagnostic Questions: banks of questions designed to target misconceptions
  • UpLearn: online A-level courses that include adaptivity and spaced repetition
  • The Learning How to Learn MOOC: this is the world’s most popular online course, discussed above. It practises what it preaches, too: the creators wrote a paper for Nature explaining how the principles of learning science informed their video design.

And of course, you can always turn to an even older learning technology – the book! If you’ve liked what you’ve read here, you could start with Teachers vs Tech, the book on which this blog is based.