Block 2. Week 6. Teaching: Reading off-piste

This data visualisation aims to represent the reading that I did over the past two weeks related to the topics we are thinking and writing about but that’s not on the course reading list. Call this reading ‘off piste’. Each branch represents a ‘reading path’: the straight one is the reading path set out by the course reading list; each branch off this main path represents off-piste reading on a particular topic; each leaf represents a 15-minute segment of reading; the dots the author I’m reading on each topic: Kitchin [2021], Véliz [2020], Brady [2019], Wu [2017] and Savage [2015].

Students often seek out readings by themselves: it’s one significant dimension to learning how to think for yourself and to becoming a more autonomous learner.  The visualisation could  be used as a contrast with standard teacher-facing dashboards which do not show teachers such ‘off-piste’ information. This kind of ‘datafication’ – ‘the rendering of social and natural worlds in machine readable digital format’ [Williamson et al, 2020:351] —  brings with it the risk that ‘only that learning that can be datified is considered valuable’ [Williamson et al, 2020:358]. Off-piste reading is hard to datify, so there is the risk that it, and the autonomous learning and thinking it engenders will be deemed less valuable. More generally, since dashboards seem to limit how teachers ‘see’ students [Brown 2020] by repeatedly directing their attention to features like ‘engagement’ – construed narrowly as platform engagement — off-piste engagement will no longer be seen or counted as such.                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                             

Block 1: Learning with data. Wrap-up

We’re now wrapping up this block on learning with data. At the start, I did not like the idea of tracking data about myself. It felt slightly narcissistic; I felt uncomfortable disclosing aspects of my life in public. Like [Veliz 2020], I associate privacy with a sense of autonomy over one’s own life. Likewise, Tsai Perrotta and Gasevic [2020:564] suggest that there might be a sense in which agency itself actually depends on ‘a relative lack of visibility and transparency’.

However, although I tried to control carefully what I revealed – and didn’t reveal — in the end, I unintentionally revealed quite a lot about myself. I think this might also be true of the design of, for example, student-facing dashboards: they inadvertently reveal how dashboard designers and the companies that they work for see students, for example. A dashboard telling students how they are doing compared to their peers inadvertently reveals the designer’s (rather low?) expectation that students are more interested competition and comparing themselves to others than, say, in what they are studying or in cooperation [Sennett, 2013] with their peers.

I learnt too that not every step of a visualisation design process is planned out (even if it looks that way to the audience); not every step has an explicit rationale. I think this intuitive, non-rational, unplanned point of the design process might make the visualisation itself (not just the data gathered) vulnerable to implicit biases of certain kinds, and, in some contexts, might render aspects of the visualisation process unaccountable. That might be fine if you’re interpreting yourself; it’s not fine if you’re interpreting others.

It’s interesting that learning (especially reading, for me, at least) tends to be coloured with emotion. That, believe it or not, came as a surprise. I didn’t start my first visualisation intending to track my emotional responses; it was something I noticed along the way as I was tracking what I was reading (and it wasn’t something I could easily control).  This in turn also suggests that we are not always transparent even to ourselves, never mind to others trying to interpret us. And things get even harder when you ask: What should we think or do when your interpretation of me disagrees with my interpretation of me, especially if we’re both in the dark to some extent (even if not quite to the same extent)?

What bothers me most about personalisation as Bulger [2014:4] characterises it is that it makes learning lonely to the point of solipsistic and fails to accommodate the value of communication with others, of teamwork, and of what Bulger calls students’ ‘need for relatedness’ [Bulger, 2014:13]. The socio-emotional learning movement tries to compensate for this with some companies (implausibly) boasting that they can now measure emotions (which they claim falsely to number exactly seven). But McStay [2020: 271] notes in passing that there is something creepy about scrutinising and measuring the intimate feelings of another like this (although he unfortunately never gets to the root of what’s creepy about it). More deeply, Eynon [2015:409-410] worries that quantification of the self risks changing a learner’s self-concept. Expanding on this, I’m inclined to think that excessive quantification of the self involves a kind of self-shaping without the consent of the one whose self is being shaped and that maybe this is the root of the reason that we find excessive quantification so creepy.

Finally, I’m not sure how to define learning but I think that it does involve expanding one’s conceptual space in various ways. Certain kinds of conditions and turns of mind are necessary to support this. To learn, you need a space where you are not constantly bombarded by distractions and where you can engage in slow but dynamic thinking extended over time, thinking not broken up into static little bits by distractions and interruptions. You also need a mind that is willing and able to listen to what others are saying. Otherwise, the only person you’re in conversation with is yourself.

I’m going to sign off this block with some more Aphex Twin. Enjoy.

Block 1. Week 5. Learning: Listening on Twitter

It’s widely recognised that conversation makes a difference to learning. Historically –- especially in Socrates, Plato and Rousseau — the kinds of conversation that make a difference to the learning process are assumed to be between teacher and student [Friesen 2020]. Much contemporary work on learning emphasises this too [Laurillard, 2013: 71]. In particular, acts of speaking or redescribing are considered critically important [Laurillard, 2013]. But listening is as important to learning as speaking and, significantly, when we are speaking, we are not listening. Genuine conversation requires listeners as well as speakers.

Crawford [2009] develops the concept of listening as a metaphor for paying attention online. In some quarters, it is still fashionable to cast listeners derogatively as lurkers – supposed non-participants who ‘prefer to inhabit the margins of debates, rarely or never contributing in public’ or as ‘freeloaders who leech the energy of online communities without offering anything in return’. [Crawford, 2009:527]. But, as Crawford argues, this inhibits our understanding of online spaces. Expanding on her point: if everybody is chattering and nobody is listening, then what is the point in saying anything? Chattering is not the same thing as communication and conversation. Without listeners, we have neither; there is only emoting into the void.

This data visualisation aims to represent the things I paid close attention to on Twitter online from Monday to Friday. It could be used to demonstrate one way to use Twitter to learn: focus on what a small number of people are posting in a given week; read and reflect on one or two things they tweet or retweet that fit with your interests; don’t worry about everything else flying by on your feed; avoid doom-scrolling.

I found it tricky to represent paying attention and listening visually. I ruled out lots of things: ear shapes – too literal-minded; sea shells – too flat and lacked the dynamism suggestive of a receptive activity involving paying active attention to what is being said. In the end, I’ve run with a sonic mode of visualisation to suit the metaphor of listening. I’ve kept it simple: it’s not possible to listen to everyone on all topics at once, after all. The clefs represent people. Kate Crawford is a blue treble clef and that’s Huw Davies as a pink one. The visualisation’s sense of movement tries to convey to what Crawford calls ‘the more dynamic process of online attention’ [2009: 527]. The act of selective and careful listening is not passive. Being selectively receptive and attentive to what other people tweet is one way of learning on Twitter. Finally, this visualisation contains one error: one of the key signatures is not quite right. I decided not to change that mistake because, as Kitchin [2021:39] notes, data are often still published even when those producing them are fully aware that they contain flaws. I’ll leave you to work out which key signature I’m talking about — assuming, of course, that you are in the relatively privileged position of being able to read music in the first place.

The Whig view of learning

The non-linearity of learning

I don’t think the process of learning is purely linear. I started noticing this a little back in Week 2. We had some set readings and I was doing those. But I was also still doing that Gitelman and Jackson reading from the book “Raw data” is an Oxymoron from Week 1. So there I was, mid-Week 2, scooting back and forward between the Week 1 readings and the Week 2 readings (and getting anxious about falling behind). In Week 2, I also read an extra reading from the same book just out of interest — the second paper by Daniel Rosenberg. But the general point is that my reading is not straightforwardly linear: relative to how our course is structured, it’s going forward and then backward, and occasionally sideways (onto Twitter). Relatedly, I don’t really think the point of reading is to get to some pre-ordained destination either. That leaves one a little unclear about where one is going sometimes; but maybe the process of learning involves tolerating some degree of uncertainty about that from time to time. So if my reading were a person, it might be someone who isn’t quite sure of where they are going. Or, maybe it would be someone who has a vague sense of where they might be going, but who doesn’t mind dancing back and forth, and definitely doesn’t see themself as having any kind of ‘learning destiny’, if you know what I mean.

Lost” learning: why it matters that learning is non-linear

Since January I’ve noticed that there’s this way of talking about, for example, children’s learning at school, that strongly suggests that the learning process is linear, or, even more strongly, that it ought to be linear. You can see this when you listen to the way many people in public life talk about children’s learning. There is much hand-wringing about the need for children to ‘catch up’ on ‘lost learning’ (see here, here, and criticism here).  What if they don’t memorize the list of kings and queens of England! This is surely regression. Or is it? If learning is not linear, then all this talk of needing ‘catch up’ and all the worry about ‘lost learning’ is misplaced. If it is necessary — assuming it is, indeed, necessary – to memorize the list of kings and queens of England, well, that might just happen later than, say, Key Stage Two, and there’s nothing wrong with that. In the meantime, perhaps we should give children in particular the space to do and learn other things and move away from sitting in distress at their laptops (assuming they have one), trying to do homework by themselves, or perhaps, if they’re lucky, with the assistance of a frazzled and exhausted parent.

An interesting consequence: algorithms and the idea of a ‘correct’ order

Many platforms use ‘smart’ learning algorithms to adapt content to meet the requirements of an individual learner.  An algorithm is “a set of defined steps that, if followed in the correct order, will computationally process input (instructions and/or data) to produce a desired outcome” [Kitchin, 2017:16] This means that in using many platforms for learning, one is often implicitly signing up to the idea that there is a ‘correct order’ for the learner to proceed. That in turn seems to leave very little room for non-linear dimensions of learning. (And, of course, I’m not saying that all learning is non-linear — I just need the claim that at least some of it is to make my point.)

The Whig view of learning

I want to draw on a famous view of history to conceptualise the more linear picture of learning and the attendant notions of ‘learning loss’ that often implicitly assume it, in order to show why such notions are problematic. The famous view of history is called the ‘Whig view of history’ and so I’ll call the analogous view of learning the Whig view of learning. Here’s the general idea behind the Whig view of history. Once upon a time, some influential British historians took an approach to historiography — the so-called Whig view — that imagined the past as an inevitable progression towards ever greater enlightenment and liberty. Here in the UK, our current forms of liberal democracy and constitutional monarchy were, and in some quarters still are, somewhat self-servingly imagined to be exemplars of such necessary, inevitable, and, of course, glorious, progress.

Now the Whig view of learning — and what’s wrong with it — comes into view. Imagine you are eight years old. Due to the disruption caused by the pandemic, you have not learned your kings and queens of England. Your parents are anxious; serious people on the telly are fretting about it too. But is this really such a great loss? After all, you’ll probably just pick it up at some other point in the future (it’s not as if the BBC is short of histories of British monarchies!). The idea that it is some sort of significant loss (often cast in terms of a loss of future ‘productivity’) seems to assume the Whiggish view that it is necessary, even inevitable, that when learning, children, for example, must be frog-marched forward, lockstep, in the correct order, towards some pre-ordained notion of what counts as progress (and often without much regard for what they want, or for the particular circumstances in which their learning is currently taking place either). But if, by the ripe old age of eight, you haven’t yet learned your kings and queens, maybe you — or your parents, not to mention the serious people on the telly — shouldn’t be so worried. Maybe you’ll learn something else instead. Or maybe — outrage of outrages — you’ll do a little idling. After all, as Russell once argued, there is much to be said in praise of idleness.

Sweet serendipity

I’m still thinking about our Tweetorials from a couple of weeks back. During our Wednesday tweetorial, Jeremy asked:

#msccde folks gave lots of interesting answers to this question. At one point, we started talking about recommender systems and Tracey mentioned in passing that she had a certain recommender system suggest that because she listened to the Jesus and Mary Chain, she might like The Proclaimers. She joked that this is why she doesn’t worry too much about the machines taking over. (Ben joked that the same music recommender system keeps trying to play him Bing Crosby long after Bing season.) There’s something to what both are saying: supposedly ‘personalised’ recommender systems get our preferences wrong all of the time in many, often comical, ways.

This conversation prompted me to look said music-recommender-system-that-I will-not-name and yes, to be sure, there were some fairly hilarious options presented to me. But there were also many more that were more or less accurate. I listen to lots of small indie bands, and, there it was, recommending lots of little bands that I had never heard of that, upon listening, I kind of liked.  Kind of convenient, if you ask me!  (I like other things too, of course, but this is a good example for what I’m about to say). Here’s the rub: while this is nice in some ways, it’s also rather conservative. The system looked at what I listened to in the past and served up a new menu of new options based on that. But there’s something crucial that the system doesn’t know about me. What it doesn’t know is that I’ve been rather bored of its recommendations of late. I mean, I like indie music, but cripes, I wouldn’t mind listening to something new; otherwise, it feels like being trapped in some sort of Groundhog Day. On the evening of our tweetorial, it was hard to see a way out of the godforsaken indie loop, so out of the blue, mid-tweetorial, I just emailed someone (who wouldn’t think me a total weirdo for randomly emailing them like this) and I asked what they were listening to. Answer: this new album.  For the rest of that week, I listened to this one track from it (you can listen below) more or less on a loop when I was doing the readings. It’s very beautiful and it’s great to listen to if you just want to concentrate on something for an extended period of time.

What’s interesting is that I really liked this new album even though it wasn’t something that I’d ever pick myself, and certainly not something that the recommender system would have given me as an option. And this, I think, is what I love about interacting with humans as opposed to machines: the sweet serendipity of encountering others who are different from ourselves and of discovering all of the new and interesting things that go along with that. And maybe too the path out of one’s own mind and towards the minds of others: the escape from the ‘personalised’ solipsism that I talked about here.

Block 1. Week 4. Learning: Slow thinking in non-networked spaces

This data visualisation aims to represent slow thinking in non-networked learning spaces. I’m imagining such spaces to be — roughly — spaces without ‘smart’ devices or learning platforms that gather data on students and that will support a slow thinking movement analogous to the slow food movement, with fewer distractions, and more learner autonomy and privacy. The visualisation is more pared down than this one to reflect these concerns.

The visualisation could be used to demonstrate how few non-networked spaces we have right now and how little time there is for extended reflections. Each snail-shape is inspired by this and represents a line of thought. A snail minus a head represents an interrupted thought. The dots represent the length of time spent thinking. I don’t disclose what I am thinking to highlight the significance of privacy — as Veliz [2020] argues, privacy is power — but I do show you where I am thinking.

The capacity, and opportunity, to carefully attend to, pursue, and sustain an extended line of thought is a significant dimension of learning. It’s not at all obvious that networked learning environments with constant notifications from online platforms like Twitter – ‘the most invasive attention capture apparatus yet invented’ [Wu, 2017:288] – accommodate this, even if they are fruitfully used for some learning events (‘Tweetorials’). Twitter also collects substantial amounts of data on users, meaning if Tweetorials are the way group activities are organised, students wanting to participate to interact with their peers don’t have a meaningful choice but to give up data (and privacy) to do so.  More generally, work by Tsai, Perotta, and Gasevic’s [2020: 562] suggests that when it comes to data practices in education, the lack of meaningful choice, and, ultimately, lack of meaningful and informed consent, diminishes learner autonomy.

Block 1. Week 3. Learning: Conceptual Space

This data visualisation aims to represent how a student’s conceptual space expands in response to provocations made via the use of a range of media. It hones in on four of many aspects to this topic: (1) The question (or theme or subject) that provokes a thought (in this case, on either learning, personalisation, or data); (2) The media though which these thoughts are provoked (either via course readings, Twitter, or our course blogs);  (3) The emotion associated with my own response in each case (interested, excited, puzzled); (4) The dimension of response (small, medium, large) to what is asked or under discussion.

This visualisation could be used to demonstrate how our learning processes are not just cognitive or social. They are also charged with emotion. Influenced by Bulger (2014:4), I took a shot at defining personalisation here. But personalisation so understood makes the learning process an almost solipsistic endeavour devoid of emotional tone. Individual preferences and competences are adapted to; social and emotional dispositions are ignored. When Jeremy misquoted me I was annoyed – an appropriate emotional response to what Peet (2015) calls interpretative injustice. The learning process is laden with emotion. Personalisation fails to accommodate this. It’s missing from what Eynon (2015) calls the quantified self.

Word count: 200 words.