[Lifestream]: a time-ordered stream of documents that functions as a diary of your electronic life.

My Lifestream Analytics

• Date of first post: 13th January 2015
• Date of last post: 29th March 2015
• Total posts: 211
• Total number of comments (excluding spam): 76
• Comments from course tutors: 10
• Comments from other students (& pingbacks): 65
• Total number of comments (including spam): I wouldn’t want to guess!

What does the Lifestream show?

The Lifestream re-constructs my real time experiences of learning, which would be difficult to produce via other modes of assessment (e.g. the digital essay).  Initial posts during weeks 1-3, for example, illustrate my engagement with key principles and concepts within digital culture, especially the philosophical approaches of trans- and post-humanism.  Through this, I developed a critical understanding of dystopian futures, as well as attempts to construct binary distinctions in humanist culture.  I was particularly drawn to the concept of homo-faber (‘tool making man’).  These are all represented in my Week 3 digital artefact and my discussion.

Content during weeks 4-8 re-construct my engagement with the concept of community, and how communities are formed and negotiated online.  I explored analytical typologies of online interaction, but became particularly interested in the role of the architects of online environments (especially their role in shaping community identity and interaction).  These themes are presented in my Week 8 artefact, which summaries findings from a micro ethnography of the Massive Open Online Course Programming for Everybody.

My interest in digital infrastructure fed nicely into the final block of the course, where we explored the role of algorithms and learning analytics.  There are a lot of tweets in this section of the Lifestream, as  through the twitter storm in week 9, I debated a number of concepts and ideas with tutors and fellow students, especially the reliance on socio-material theory, the ‘agency’ of algorithms and the relevance of theories of power and political economy in understanding the role of algorithms in digital culture.

Reflections on Lifestreaming

As the weeks progressed, the use of automated feed (via IFFT) helped to transform the Lifestream from something which I need to create, to something which largely generated itself.  In this respect, the  lifestream became organic (as opposed to mechanistic).  Whilst twitter was the primary automated feed, Pinterest Vimeo and Youtube (and later Tumblr and Flicker) became additional (multi-modal) feeds into the Lifestream.  It is interesting to see how the different feeds reflect different learning activities: Twitter posts, for example, tend to reflect my interactions with tutors and fellow students, whereas Youtube, Vimeo and Pinterest feeds reflect my engagement with a range on non-academic materials (surrounding Digital Culture, Netnography, and Big Data).

A potential limitation of the Lifestream is that, whilst it provides a detailed history of online activity (course participation), this does not (in and of itself) evidence effective learning.  Hence, there is a danger in assuming that extensive lifestreams can serve as proxies for student engagement, and (in contrast) sparse lifestreams signal limited learning.  This is where the incorporation of weekly summaries into the Lifestream provided crucial opportunities for critical reflection.

My critical discussion of the analytics from the Tweetorial highlighted the over-representation of males as top-participants.  Analytics such as this, I argued, can help facilitate a level of reflexivity in teaching; for example, by highlighting teaching techniques that could, inadvertently, re-enforce gendered divisions in learning.  I also argued that such analytics need to be treated with caution.  In particular, we need to avoid the temptation – as highlighted in Ben Williamson’s presentation – to view big data as ‘revealing’ the true social facts and patters in education.  Instead, I argued, learning analytics are best viewed as part of a broader set of socio-material relationships, through which the learning process is socially constructed.  This, I suspect, is not the prevailing approach to the promise of Big Data in higher education!

This led me to consider the focus of my digital essay.  I am keen to explore Big Data further, so I emailed a provisional title to Jeremy and Sian (which was looking to explore humanist and anti-humanist frameworks to using Big Data in smart universities).  This was to broad a focus for an assignment of 2500 words, so I  narrowed the focus to the following

The promises and pitfalls of Big Data in higher education

I then did some searching via Pinterest on visual representations of the promises and pitfalls of Big Data.  As you can see from the results, there tends to be more promises associated with Big Data than pitfalls, which will be a useful theme to explore in terms of the emerging ideology of Big Data.

Q: ‘How has the Tweet Archivist service represented our Tweetorial (Use the drop down arrows on the menu items on the right of the page: ‘Top users’, ‘Top words’, Top URLs’, Source of Tweet’ etc.)’

Tweet Archivist presents a series of numeric, textual and visual summariesof the twitter activity.  On the top bar is a series of overall summaries: total tweets (n=325); ‘impressions’ (not sure what this refers to!); date range of the archive (3rd-16th March 2015) and whether the archive is active/inactive.

On the right hand side of the page is a series of representations.  There is a pie chart of ‘top users'; a word cloud of ‘top words'; a list of top url links; a pie chart depicting the source of the tweets; a pie chart depicting languages used in tweets; a line graph depicting spikes in activity; a word cloud of twitter mentions and hashtags; a link to images and finally (and interestingly!) and ‘influencer index’.

Following Williamson (2014) these summaries are a selection of all possible summaries that could have been produced.  Additional summaries, for example, could have included time of tweets as well as the number of replies to tweets.  Another pragmatic choice is the medium in which these data are represented e.g. using word clouds and pie charts rather than bar.  Statistics are kept to purely descriptive features i.e. there are no analytical statistics presented e.g. between top users and top word returns.  The designer of this algorithm has, therefore, had to make a series of pragmatic choices yet the rationale for these decisions are not visible on the website (so the reader cannot evaluate the quality of the data presented).

Q: ‘What do these visualisations, summaries and snapshots say about what happened during our Tweetorial, and do they accurately represent the ways you perceived the Tweetorial to unfold, as well as your own contributions?’

I would take issue with the notion of ‘accuracy’ as this implies there is a ‘right’ and a ‘wrong’ way to present these data.  From an ethno-statistics perspective (Gephart, 2006), these data should not be taken as a series of truth statements, but (instead) as artefacts produced through a process of meaning-making, by which the tweetorial and its impact on learning are socially constructed.

These data do, however, highlight some interesting findings.  Top users, for example, are primarily male whereas less participating learners were female.  Does this point to the potential for Twitter to re-enforce gendered divisions in learning?  Twitter in general, and twitter storms in particular, have a strong association with discussing controversy and for thesis vs anti-thesis style learning (Hegel, rather than co-operative inquiry.  My impression of the tweetorial was that it was based on challenging users statements and beliefs, rather than building a shared consensus (e.g. through a Delphi style process).  In this respect, the archive data lends support to my initial impressions.

Q: ‘What might be the educational value or limitations of these kinds of visualisations and summaries, and how do they relate to the ‘learning’ that might have taken place during the ‘Tweetorial’?’

Facet methodology offers a useful approach to using learning analytics in education.

As above, I think the key to the educational value of such analytics lies in the approach taken to their interpretation.  If educators adopt a quasi-positivist perspective, for example, and treat learning analytics as objective ‘facts’, then their value will be limited.  This is because treating analytic data as real ‘things’ ignores the role of pragmatic decision-making (e.g. by programmers and software designers) in how data are mined, interpreted and represented.  If, however, learning analytics are treated as social constructions then they do have a useful role to play in highlighting patterns of social interactions, through which (following a network learning perspective) learning takes place.  Facet methodology, as outlined by Mason (2011) I think offers an ideal approach to interpreting learning analytics.  This approach treats data and data sources (such as Twitter Archives) as a facet (akin to a side of a cut gemstone) which illuminates certain aspects of social phenomena whilst simultaneously obscuring others.  Thus, from this perspective, learning analytics represent one of many possible facets in understanding the learning process.

References

• Gephart, R. (2006) Ethnostatistics and organisational research methodologies. Organisational Research Methods 9(4): 417-431.
• Mason, J. (2011) Facet methodology: the case for an inventive research orientation. Methodological Innovations Online 6(3): 75-92.
• Williamson, B. (2014) Calculating Academics: theorising the algorithmic organization of the digital university

Images

• Round brilliant cut diamond

I engaged in several debates with tutors and fellow students via twitter around the role of algorithms in education.  Particular debates focused on:

• Whether or not researchers need to learn how to code
• Who ‘owns’ algorithms and the potential of algorithms to behave in unpredictable ways
• The extent to which we ‘give’ information to algorithms to use
• The dominance of post-humanist/modernist thought in framing debates about algorithms and learning analytics

During the course, I’ve become increasingly sceptical of the ability for socio-material theory (and other post-modernist frameworks) to offer adequate frameworks for these debates.  In particular, they fail to  consider adequately the role of power and vested interests in shaping online learning spaces and practices.  There is a strong preference in the literature for post-humanist/modernist models, yet alternative models do exists!

One approach I’ve been exploring is the political economy of digital culture.  This approach, in my view, offers a more coherent framework for exploring the role of algorithms and learning analytics in the (envisioned) 21st century SMART university. I intend to explore this approach in my assignment.

References

Ben Williamson (2014) Calculating Academics: theorising the algorithmic organization of the digital university