218 – SmileyCoin

[00:00:07] Santiago: Hi, and welcome to the IDEMS podcast. I am Santiago Borio, an Impact Activation Fellow, and I’m here with David Stern, one of the founding directors of IDEMS.

Hi David.

[00:00:17] David: Hi Santiago. What’s the discussion point today?

[00:00:21] Santiago: Well, we had a few episodes recently about different technologies we used in some way. We mentioned STACK a lot, you had an episode with Lily about CAST, these electronic statistics textbooks. We had an episode about PreTeXt, and you mentioned in that episode another technology or project, which you put it as sort of another piece of the puzzle, which is SmileyCoins.

I must say, I heard about them, I know a small amount about them, but, really, it will be you mostly telling me about them because my knowledge is very limited on this topic.

[00:01:17] David: Yeah, and I’m very grateful to Gunnar, who is the real driving force behind this. I actually caught up with him recently, Gunnar Stefánsson, I think he’s now a retired professor from Iceland, he instigated this work with a number of colleagues. But it’s really exciting, and something which I’m keeping tabs on, although we’re not pushing on this much directly at the moment.

[00:01:41] Santiago: So, what is this project?

[00:01:44] David: When I first met him, about 15 years ago now, we met at a conference and it was one of these networking conferences where there were lots of people presenting, the conference itself was just a sort of vehicle to bring people together. And I must admit there weren’t many interesting people at the conference.

But as soon as I saw Gunner presenting, I thought, I like this, I love what you’re doing. And the key thing that he presented at that point was actually very interesting, he had randomised, multiple choice questions, a lot of his work was based just on multiple choice, which we’ve had a lot of episodes about why multiple choice questions aren’t enough.

[00:02:26] Santiago: And this is mathematics.

[00:02:28] David: He was doing this for mathematics and statistics and fisheries and a few other things, there were a number of other things, but he had a system which was built out called TutorWeb. And his presentation caught my attention because he was analysing, this is 15 years ago now, he was analysing the data and he was saying there’s a real problem with our standard way of grading the multiple choice questions.

What he did when he analysed the data is he found that the normal way of grading it was sort of saying, if you’re doing mastery, your last 10 question count or something like this. And he was saying there’s a problem with this because what we’re finding in the data is, quite often, when you have this sort of grading in place, if people get a question wrong, they give up. And so they’re left with the impression that they’re bad at maths.

And this is because of our choices. And actually they found that if you analyse the student data, the proportion of students who got their last question wrong was higher than the proportion of students who got previous questions wrong. And so, more people were finishing getting a question wrong than they were getting questions wrong.

[00:03:43] Santiago: Sorry, that they were getting questions right you mean.

[00:03:47] David: No, no, no, no, it wasn’t that more people were finishing getting it wrong than right. It was that the probability of getting a question wrong or right was lower if it was your last question because people gave up after they got a question wrong, because if they were wanting a perfect streak of 10 correct questions, then, they’ve done eight and now they were gonna have to do a whole nother 10 questions, and so they couldn’t face it, and they gave up.

[00:04:16] Santiago: Sorry, but surely this means that the students are finding out whether they’re getting a question right or wrong instantaneously.

[00:04:24] David: Yes, absolutely, these were mastery quizzes, randomised in interesting ways, actually, with increasing difficulty. It was all built in in very nice ways, beautiful system, at that time it was called TutorWeb, it’s now called SmileyTutor, really nice little system just based on multiple choice questions. We’ve had this used, it’s been actually quite effective as drill based questions in Kenya, we’ve supported this in different ways.

But I’ve always felt that pedagogically, I want to go beyond just multiple choice. And until recently, the possibility of integrating these systems with systems that go beyond multiple choice wasn’t really technically feasible. But there’s reasons why right now, we could move in that direction.

But as multiple choice questions and quizzes go, this is the best I’ve ever seen. And his analysis of it was fantastic. Even 15 years ago, he had these insights about how people, how you can change the grading scales so that you change the incentives for students, so that they’re not incentivised to leave on getting a question wrong, that they’re incentivised to finish on a high note. And that changes their perception of the subject and so on. So beautiful, beautiful work, he was doing research on this.

And so he presented that and we interacted from then on, I got him involved in Kenya. And really, really interesting interactions since then.

[00:05:47] Santiago: So, let me see if I understand this correctly. One of the insights was ending on a correct answer, psychologically leaves you feeling better than ending on an incorrect answer.

[00:06:04] David: Well, I would argue that the way he presented, now I’m talking about presentations from 15 years ago, you’re testing my memory here, but the way he presented it was that actually, the choice of grading methods that we generally associate with mastery and drill based quizzes create a scenario where people are more often finishing on getting a wrong question than they would just by chance. And so there is actually bias, which is coming in from our grading approaches, which is leaving people with that bad feeling. That was sort of how he presented it.

Really interesting, very simple things, and the sort of thing that you can do once you’ve got electronic assessment in place, if you analyse it well, you can find out these sorts of things. How are people responding to the assessment? How are they doing? One of the things I’m really pleased, and I’ve always been looking at, when we’ve been having the STACK work is that doesn’t seem to be the case with the STACK questions. Actually, when you have the STACK Mastery, when we look at it, people keep going until they’ve actually got the quiz right, there’s cases we’ve got where we haven’t seen that same problem in that same way with mastery in the STACK context.

That’s where he started, and then we got him involved in Kenya and he actually built these drill based systems with multiple choice questions related to the Kenyan curriculum. And we put this in these remote places and it actually was quite effective. But the real problem was motivation for students.

And then he had this incredible idea to say, well, actually, what if we could give them a financial motivation through a cryptocurrency? And he created this cryptocurrency. Now, I looked into this in great detail at different times, and there were always criticisms that you could lay these systems because of the energy problems, because of other things.

But he thought this out quite carefully in terms of actually building a charity behind it and having the SmileyCoin, which is a cryptocurrency, which he’s able to maintain a particular level, which means that if students do the drills really well, they can even earn a tablet, you know, through doing their maths homework. And this has been used in incredible ways.

[00:08:29] Santiago: Again, correct me if I’m wrong, but it’s not a standard cryptocurrency, it’s a cryptocurrency that only allows you to purchase certain things that have educational value?

[00:08:42] David: No, it’s a standard cryptocurrency. It’s on the exchange like any other cryptocurrency. But, different to other cryptocurrencies, it has the social aspect baked into it. It’s a coin, which actually when you mine it, if you mine it as you would any other cryptocurrency, a certain small percentage goes back to the educational projects.

And so it is a standard cryptocurrency, but because it was backed up by this not-for-profit charity model, they were able to get groups in Iceland to use it and to use it with students there. And then they got groups in Kenya and Ethiopia, and they’ve got Ghana, I believe, there’s a number of different countries. There’s more than 50 schools across Africa, I think, using this now. And they’ve had libraries using this.

They’ve had challenges with librarians actually then trying to just earn money by doing maths quizzes, teachers taking the tablets away from students and just using them to try and earn money from themselves, there have been challenges.

But the general process, it’s been incredible. They’ve had shops formed locally where kids have bought food for the family based on doing their maths homework. They’ve had all sorts of things come out with this in really powerful ways. It’s really incredible stuff, and an incredible use of a cryptocurrency to do something which is totally, well, it’s really socially positive.

Now, of course, the challenge is scaling because there are elements of funds about how you get funds in in different ways. But, those scaling challenges are exactly what Gunnar is now interested in trying to tackle. They’ve got enough evidence, they’ve been doing this now, I don’t think it’s quite a decade that they’ve had the SmileyCoins, but it’s close to that, and they’ve built these systems that work.

[00:10:34] Santiago: I feel that I need a bit of clarity here. You say one of the challenges is scaling and that’s obvious, but the challenge that comes to mind is sustainability, because the more the kids are doing their homework, the more currency is issued to them, and the lower its value.

[00:11:01] David: Well, yes, and the way they have been doing that at the moment is that some of the value has been made by the coin being just used internally in Iceland, which is where this idea emanated from. And so it’s just used as a cryptocurrency, which is taken by certain students in certain shops and they use it, and so on.

But the minimum value of the currency has been guaranteed by a sort of research fund, a small research fund, which then is enabling some of the African components to use this and so on without devaluing the currency. And so that’s why there’s been a limitation to the scale.

Now in practice, the way this is designed is that limitation of scale, if other people started using this cryptocurrency, like other cryptocurrencies are used, then that would actually mean that that use of the cryptocurrency in different ways, some of the value is then fed back into the educational program in a way where this would just be part of how the cryptocurrency works. And so that’s what’s actually designed and baked into the cryptocurrency itself.

[00:12:10] Santiago: So there is a social motivation for choosing this particular cryptocurrency.

[00:12:18] David: Absolutely. And if you are a miner of cryptocurrencies, then you do get slightly less from mining this currency than you would from others because of that social element. But it’s still something which is a valid and a sensible model for cryptocurrency.

I have engaged in this over the years, but the details of it, I always felt something where if we really wanted to engage in this, then I think there are questions I would have about, if this were to take off at scale, are the details right or not? And I don’t know, I don’t know enough about what is possible.

I think there’s certain elements of innovation that were taken with this as a cryptocurrency, which I think are brilliant. Baking in the social aspect I think was fantastic. But there are other elements, such as the environmental sustainability of these approaches, which weren’t thought about. That’s a concern for all cryptocurrencies, that’s one of the reasons that we haven’t engaged more generally. But it is something where I think it’s a step in the right direction.

The actual process of mining cryptocurrencies is computationally intensive and therefore this is using power and energy in a way which is questionable as whether this is the most efficient way to do these things and so on. And that is baked into the algorithms that are used in, I believe, all cases. I know there were some people working on innovations around this, but I’ve not heard of any real breakthroughs.

And so, this is one of the reasons that we’ve not engaged more seriously in this because it’s just a big task. We’d need to actually think deeply about how to do this and actually think the whole thing through.

But what Gunnar’s done already is fantastic. You know, he’s baked in the social element, he’s got that fed in, it’s a standard, he’s building into the standards of a cryptocurrency processes. It’s a really powerful approach, and I think very exciting. And I’ve been watching it keenly for years, and they’re still going strong, they’re slowly growing, they’ve got a little bit of other interest in other ways.

I heard they created games that actually use the SmileyCoins, and issue SmileyCoins as well. This is something I’m sure you’d be really interested in, actually embedding the SmileyCoins as part of how you can engage with certain games and so on.

[00:14:44] Santiago: I had heard of certain games that allow you to gain Bitcoin, but I hadn’t heard of any other cryptocurrency used in that way, and presumably these are educational games.

[00:14:58] David: Well, actually I believe these, I don’t know, I’d need to check with Gunnar on this, ’cause he mentioned them just in passing ’cause I didn’t know about this. But he said that there were a few other applications of SmileyCoin, which they’ve tried and which had been done. But I believe some of them were just games, that they had a games developer who was developing a set of games and one of them embedded the SmileyCoins in that, whether it was an educational game or not, I don’t know. I don’t think it was, I think it was just a game, it’s just a games company who produced a game and they decided to embed the SmileyCoins in it in a way, which was part of their marketing.

Anyway, I don’t think it was that successful, but it’s something where that at least has been tried and there’s ways that these things could, there’s a lot of exciting things that I think could come out of this.

[00:15:43] Santiago: And the challenge of scale comes from the number of people actually choosing to use this cryptocurrency?

[00:15:52] David: If this cryptocurrency did get picked up at scale, then of course it would just be in circulation, and then these education programs fitting into it could scale more easily and that would be relatively straightforward. And so in some sense, yes, people using the cryptocurrency is the big way that this could scale.

But if it isn’t widely used as just a standard cryptocurrency, then the challenge of scale really does relate to, well, are there studies using this or are there business models which actually come around, which then close the loops on this? The point is if you are closing the loop on it, it doesn’t matter, the amount of money going to students in different ways and the currency going to students in different ways, that’s just part of creating a system where the coin has value. You’ve just gotta have a system where enough value is created through the use of the cryptocurrency that the system is sort of sustainable.

And so, yeah, there’s a lot of different ways this could happen. It’s challenging, but it’s exciting because think of this more generally. You know, the reason these cryptocurrencies, it’s possible to do this sort of thing, is because the friction on these micro transactions is so low. And so actually what you can do is you can have micro transactions happening in ways which are efficient and which are effective at scale.

So actually, within these systems, having these low friction micropayments, it can be very effective because you can really remove a lot of the friction so a small amount of money going to support these initiatives goes a long way here because these are very frictionless systems.

Iceland is not known as a big international donor, but it’s something which has done something so powerful. This particular innovation is one of the best I’ve seen and one of the most scalable I’ve seen, because it creates these systems which are almost frictionless for microtransactions.

[00:17:51] Santiago: Okay. You are saying scalable now, when you said scale was a bit of a challenge, but I think I understand that it is scalable if certain conditions are met in terms of using.

Now I have two questions for you. First one is, I believe there was a trial in a very remote place in Kenya where even electricity was a challenge and they managed to get solar powered tablets in a school, in a sort of village school.

[00:18:27] David: On an island in the middle of Lake Victoria, a small island, a school which was just remote and which has had problems with the students and supporting them in different ways. No electricity other than, as you say, a solar powered sort of system, which could then be hooked up to…

It was a fantastic trial. It really was very exciting, and it showed incredible results that students from the island then progressed in education in ways that had never happened before, and not really been possible in that school.

And the other example I had was they tried this in prisons in Kenya. They did this in Kenyan prisons and it was fantastic, it enabled education to happen in these really challenging environments. They’ve got these things, they’ve got really interesting case studies, worked out in different contexts, in different places. It’s incredible.

And Kenya has now become actually one of their biggest sort of trial sites, they’ve got trial sites all over. But they have so many different learnings from Kenya, they’ve had challenges, but it’s really been fascinating to see.

This was not designed for Kenya, before I met Gunnar he wasn’t thinking about doing this internationally, you know, he wasn’t thinking about SmileyCoins, he was thinking about doing things in Iceland for Iceland. And then he got interested and we got him started to work in this African context, he had this sort of whole setup where he could bring his suitcase of these tablets and a offline system and all the setup which enabled this to work in remote areas, limited or no access to internet.

In the prisons, they couldn’t have access to the internet, so they had to do this in an offline system. They ran that off of Raspberry Pi, I believe, at one point. It’s great. Really interesting, really fascinating learnings.

The challenge, which is the challenge, which I come back to is that they’ve based everything on the multiple choice questions, which pedagogically, I believe should only ever be part of a solution like this. Getting it so that this could integrate in with other systems like STACK and other, that requires a whole different mindset of how this works.

But, I can tell you that this is a discussion Gunnar and I had just a few weeks ago, and I’m really excited that this is something where we could have this, we could integrate this. Now there’s no reason, as he put it, why the infrastructure of the SmileyCoin couldn’t be attached to the textbooks that Mike is building, so that we could actually build this on the STACK questions in the mastery in these contexts and so on.

And that would potentially enable something like this to scale to a national level. These things are theoretically possible, but there would need to be work at all the different levels to integrate. It’s always been my dream to integrate these things together.

[00:21:08] Santiago: And that’s why you say it is another piece of the puzzle.

[00:21:12] David: Exactly.

[00:21:13] Santiago: Now, maybe my final question to close the episode, my understanding and one of the things that I was most interested about with TutorWeb, not so much with the SmileyCoins, the SmileyCoins were an aspect that I found that I didn’t quite understand because I don’t understand cryptocurrencies too well.

But I’m interested in the educational value, and the bit that interested me is that Gunnar, I think with a bit of your input, managed to develop a sort of algorithm to decide which questions to present to students according to their previous responses so that students could get that sense of finishing on a high.

[00:22:05] David: This was not with my input, this is exactly the work that he was doing in different ways. And as I say, the interesting point of this is he’s based all of this work and this research around multiple choice questions. And good multiple choice questions do lead to good learning.

Now there’s actually criticism about that, there’s a lot of literature in the pedagogy about how having open questions is really important. But what I love about the work that he’s done is he said, well, yes, yes, maybe we can do better, but multiple choice questions can give a better education than people can currently have, and this is something which is valuable, useful, it’s leading to better learning if we give people access to this. And so why don’t we just do that?

That’s what he’s really driven forward with. And I have so much admiration for that in a sense that in many of these contexts, we wanted to have content which covered the whole school curriculum, but it took us 10, 15 years to put in place structures that could create it.

Because he was just doing multiple choice questions and the way he was doing it with these powerful algorithms, he was able to do that much faster, he was able to get things out and get them tested with people in ways that were really powerful. Is it perfect? No. But is it good? Yes. And exactly this element, as you say, of the algorithms that he’s got, actually, there are elements of his multiple choice questions which we can’t really compete with yet in STACK, simply because actually the cognitive load of writing good STACK questions means that we’ve not yet actually developed as many as he has.

He was able to churn out lots of them and then find ways that within the internal algorithms, they provide this student progression. There’s some really fascinating approaches, which I think there’s so much to learn from these different people taking different approaches, I just love the work they’ve been doing.

[00:24:06] Santiago: And this is perhaps a lead up to our next episode. This is a really lovely example of adaptable personal learning. And I think we should stop here and focus our next episode on the potential benefits and how personal learning could be achieved.

[00:24:29] David: And bearing in mind there are lots of systems that are already doing bits of this. I mean, TutorWeb, which is now of course SmileyTutor, is just one of them. But this is not the most well known. There were many others which are better known and have built in these elements of personalisation.

[00:24:46] Santiago: And we can discuss that next time.

[00:24:48] David: Sounds great. Look forward to it.

[00:24:50] Santiago: Thank you very much, David.

[00:24:52] David: Thanks.