Landing : Athabascau University

And so it ends...

Well, actually, I'm very happy to continue the conversation and develop this group space if anyone is interested. However, we are reaching the end of my week at the Change11 MOOC so this is probably the last post in this space for a while, back to my usual blog for the next one.

We had a lively and, I think, fruitful and interesting discussion today to wrap up my week at #change11. Stephen Downes has made the audio available at I suppose I should provide some kind of summary of where we got to but all that stuff is there online already and in the audio of the session, so I'm going to finish with some observations about the MOOC itself and some of the self-referential ways it relates to what I've been discussing this week.

I'm not the first to observe that a big problem with connectivist-influenced MOOCs like this is that they are, well, chaotic and lacking in centre. People are contributing all over the place in a hundred different ways and certainly not in an orderly fashion. This is not your grandmother's kind of course and that would be fine, apart from the fact that such a small percentage of people wind up getting fully engaged and so many drop out, one of the main reasons being the complexity of following and keeping up with the course. If we had drop-out rates of this magnitude in our universities there would be some very serious questions asked. But this is not the same kind of thing as a formal institutional course and it would be silly to apply identical standards here. Apart from anything else, the only motivation for most people being here is intrinsic - apart from a very few who are getting some kind of professional or academic credit indirectly or directly as a result, no one is going to punish them for failure to attend, no one is going to reward them with grades for pleasing the teacher or demonstrating knowledge of a fixed set of stuff. But that does make me wonder a little - if we had such an intrinsically motivated crowd in a traditional course we would be pretty pleased and would have very high expectations as a result. And yet, many fall by the wayside.

I don't think it's too much of a problem that many people do not write anything public - people learn in different ways at different times and respond in different ways to different things, so (though it greatly helps the learning process to write about it, especially in public, as well as helping to provide one of the pillars of intrinsic motivation, connection with others) it is fine that only some of the participants are visibly 'there'. And it is equally fine that people pay attention to some sessions and not others - there is no particular narrative in the various presentations and there is no single body of knowledge to absorb (that's part of the point) so people should only engage with what they find engaging.

But wouldn't it be great if more people stuck with it? Wouldn't that show that it was really working?

We wound up talking quite a bit about balance this week - reaching that Goldilocks spot that is not too hard and not too soft in not just our technologies but the whole system of which technologies are a part. i think that the change11 MOOC technology, though decidedly flaky in places (Stephen Downes is brilliant but he only has so much time to build and manage tools along with the rest of his commitments), is evolving nicely, employing precisely the kind of principles I've been advocating this week and for many years, of building with small, hard pieces, and aggregating them well. I think I might adapt the interaction design a little here and there but there is now a fairly strong sense of narrative that emerges through the deliberate aggregation of blogs, Tweets and so on, and a good centre to the course on the change11 site, with strong and simple interfaces to other systems so it can itself be reaggregated as we wish. It relies a bit too much on soft technologies - my own sessions this week came close to disaster because Stephen was left having to handle almost the whole thing while George and Dave were away, which meant 'my' page labelled me as Erik Duval for most of the week, my sessions didn't appear in the calendar and today's session was only linked and announced five minutes before it began. We harden stuff because it makes things less error-prone, faster, more efficient, and there is scope for a little hardening here though, having said that, it was the brittleness of hard technologies that made the announcement of today's session so late: an automated system had broken down, and a soft technology (Stephen manually running the job) that allowed it to recover.

But, though there is a good bit of top-down structure and some good aggregation of bottom-up content, the scope for emergence is slightly limited because what gets aggregated is presented as a single, flat stream of content, the good and the bad, the useful and the useless, the helpful and the obscurational. It is left almost entirely to soft technologies (ie the reader's means of constructing structure or recommendations of others) to sort it out. Because it is not easy, this will be demotivating and inefficient. Although a lot of soft cues are available (titles, tags, reputation of the poster, etc) and an extra layer of other social media sits on top (e.g. Tweets from people we respect helping to draw attention to good stuff) and some people are creating their own edited aggregations, there needs to be a lot more texture here. A single view of any course is always going to be a compromise that suits some and not others, but that is even more of a problem when an almost unfiltered stream of stuff comes pouring in with nothing to counterbalance it but the top-down structure of the course leaders. And it's not enough to subject it to editorial control (e.g. a simple rating system) because different things will be valuable to different people at different times. A number of relatively simple collective-based solutions immediately spring to mind, all of which would require a bit of serious programming somewhere down the line but any of which might make use of existing tools:

  • collaborative filtering - matching people whose interests and tastes seem similar, whether through explicit profiling or mining of implicit preferences, would help to draw attention to things that are more interesting. There are big weaknesses with a CF approach in learning that I and others have written about so it is not the final solution, but it would be a start.
  • tag clouds - this would be especially valuable if it dovetailed with a soft technology, whereby people tagged things not only with #change11 but also other tags to help identify the purpose, theme, pedagogical value of things they post
  • reputation management - perhaps as an adjunct to CF or on its own, a means of helping to distinguish different individuals. I would not be happy with a simple ranking though - my own CoFIND system used an approach that would be useful here, whereby ratings covered a range of dimensions (technically achieved through the use of fuzzy tags with non-binary values) rather than a simple good-bad scale. Slashdot uses similar and many other mechanisms to allow tuning that is highly tailored, a very scrutable user model, but that would be way too complex and nerdy for this purpose 
  • visualisation - combined with other approaches, tools to help show threads, trends, changes, patterns etc
  • adaptive hypermedia - eveyone who joins the course registers on the site and could be encouraged to provide a profile, which could be used as the basis for a simple user model to provide intentional recommendations. 

None of these alone are sufficient but together these, and things like them, might help provide the structures that different people would find useful. Whatever system is used, it is important that it is just employed to provide signposts, not fenceposts. I would resent having things hidden from me, but would value a bit of help in alerting me of things that I might find interesting, or of helping me to find stuff that is most relevant to my current needs.

I think MOOCs are brilliant and I really like what Stephen and George and Dave are doing with this one. It's an evolving model that is becoming far richer as the years go by but, to be really scalable rather than just big, MOOCs need to begin to mindfully employ some more tools to help different structures and guidance to emerge out that mass of interaction, to really use the crowd. It has been really interesting and a privilege to be involved in this and I look forward to the ongoing conversations and discoveries that will occur over the rest of the course!


  • Jon,

    I like what you've done here to take a look at the course itself. I wish I had read some of your thoughts on the MOOC a few weeks ago leading up to your week. I find that I am relying on my own content filtering system, people I follow on twitter, following the #change11 hash tag and reading blogs I find particularly interesting but I always still have a nagging sense I am missing something. 

    I have the most time on Sundays and the calendar on the site wasn't updated with your information and you're right that did play a part in my lack of participation. That said, a tweet from you or from someone else linking your reflection, or even some transitory discussion from one week to the next would be particularly helpful. It's almost like every week is a new course with a new facilitator parachuting in.

    Anyway, I just wanted you know that my lack of direct, observable participation this week wasn't because I wasn't interested or I had 'dropped out'. I still intend on going over your recording and the other materials at some point. Right now, I'm imaging doing a goldilocks comic for you :)

    Finally, I realize that this is a volunteer thing for you, so I'd like to thank you for your work. What I have read so far seems fascinating and quite relevant.

    - Giulia Forsythe

    an unauthenticated user of the Landing November 26, 2011 - 1:20pm

  • Jon, thank you for leading this week's MOOC topic. I found it to be very helpful. I also appreciate your critiques and suggestions regarding the structure of the MOOC. This is my first time participating in one, and so far I have benefitted immensely, though I have not written much about my experience publicly. I think the reason I have not interacted much is that navigating, processing, and applying each week's material to my own context takes so much time, that I don't have time left to write blog posts, or interact with fellow class mates. I plan on sticking with the class until the end though. Thanks again!

    - KellyABurton

    an unauthenticated user of the Landing November 26, 2011 - 10:00pm

  • Interesting reflections and I appreciate the comments and the participation.

    It's easy enough technically to implement some sort of collaborative filtering or reputation management system, but the result would conflict with the objectives of the design of the MOOC.

    To over-generalize, things that pull out one (best post, most reputable writer, etc) out of many are exactly the sort of things I wish to avoid. I think you sense this - you write "A single view of any course is always going to be a compromise that suits some and not others" - but my response is to attempt to avoid the single view.

    This makes the parceling or highlighting problem an order of magnitude more difficult. Basically, it amounts to wanting a way to do it for each participant, but also to provide each participant maximal choice, and a reasonable but not excessive amount of homophily.

    I think a tag system is an excellent alternative, but simple keyword tagging is clumsy and ineffective - it depends far too much on what you are calling soft technologies (and specifically, the act of applying the tag) and means the only resources available are self-selected materials.

    I do have a 'topics' system that preserves the best of tags but greatly automates the process, but I've been frustrated by some technical difficulties. It requires a lot of caching, and my cache system has its issues (if comments haven't been appearing when you make them on posts, it's because I'm still trying to make the topic system work).

    I don't think a parceling system will be by itself sufficient, however. I'm not even sure it's necessary. I think that the problem of participantion lies elsewhere. Because we could send a post with only a small number of resources to people, which would be easily managable, and participation would still decrease.

    That's why, in my talk on engagement this week, I tried to explore the various things that would cause people to commit to doing things. I don't think any of the formulae are quite right yet. And nothing will be perfect - people take these courses in their spare time, which means they may stop for any reason at any time.

    I don't think the answer will be a _simple_ thing, like badges, levels, competition, rewards, etc. - I expect it means getting the basic design (open, connected, interactive) right, plus providing focus (attractors, parcelation, personalization), and then stimulating actions (signs and symbols, loyalty, campaigns, progress indicators, etc).

    - Stephen Downes

    an unauthenticated user of the Landing November 27, 2011 - 7:02am

  • Thanks Giulia, Kelly and Stephen!

    Stephen, I thnk we agree wholeheartedly on this, though I suspect I didn't make that clear enough in this post. The last thing needed here is a simple system of badges, rankings and rewards: that's the quickest way to make a stupid mob rather than a wise crowd. I also totally agree that filtering imposed by the system would be a bad idea. I think your last sentence sums up my thoughts pretty well.

    Signposts are potentially very helpful, fenceposts are not. Information hiding appeals to computer geeks who want to be in control, but only works when you control everything else in the system and have some very low-level learning goals.

    On the whole I'm a fan of making things adaptable, rather than adaptive. Or rather, I like using adaptive tools (CFs, recommenders, adaptive algorithms etc) as part of an adaptable system, allowing people to choose the automated tools as and when they need or want them. It's about giving people lenses that can overlay structures and patterns that are useful to them. Tag clouds are good for this, though simple binary tags are (apart from in quite closed communities) pretty coarse-grained and only take you so far. Over the years I've developed some variants - taggable tags, fuzzy tags and aggregatable tags - that can offer a more refined view, though I've yet to solve interface issues (they get over-complex very quickly) and the greatly magnified cold-start problems that occur as soon as you shift away from simple binary classifications. Similarly, i like Judy Kay's ideas about scrutable user models but her systems suffer the same problems of usability and clarity. The most sophisticated mainstream approach I've come across is the vast array of tunable and intelligent methods used by Slashdot but you have to be a well-motivated nerd to use them properly and they are not the answer here. Simple preferences, however they are used are used, even the most sophisticated of collaborative filters, are not very useful in a learning situation. They can lead to filter bubbles, preferential attachment, confirmation biases and all sorts of woes that work against the interests of the learner and the learning community. However, tools that allow for changeability, diversity and shifting contexts, catering for the fact that to learn is to change, are too complex to manage. It's a tricky tension that makes this a really interesting place to do research!


    Jon Dron November 27, 2011 - 6:17pm

  • hi,

    em um MOOC tudo fas a diferença e principalmente o tempo de resposta em direçao a sua semana. Eu gostei muito da sua semana, parabens!

    Cada pessoa tem um tempo para" digestao" , creio que muitos entram no hange11 para saber e sem participar efetivamente eles criticam... facil falar e dificil por em pratica.

    Este Change11 tem um programa, so que em cada semana uma pessoa diferente, 

    necessitamos ser rapidos, e nem sempre sera possivel. 

    - Tereza

    an unauthenticated user of the Landing November 27, 2011 - 6:49pm

These comments are moderated. Your comment will not be visible unless accepted by the content owner.

Only simple HTML formatting is allowed and any hyperlinks will be stripped away. If you need to include a URL then please simply type it so that users can copy and paste it if needed.