Week 8 Summary

Attempting to play with algorithms

Last week was spent attempting to affect a change in my Soundcloud weekly playlist by liking and playing every track I could find with the title ‘algorithm’.

That’s why there are so many Soundcloud posts currently in the lifestream. I switched off the IFTTT link after a while to avoid flooding the stream as I ended up with more than 100 tracks.

However, perhaps fortunately, it didn’t have great deal of effect. In fact the one effect it did have was that I received a message from Soundcloud telling me to slow down the liking of so many tracks. Weird.

I guess that shows how the algorithm has been played  by users previously.

Besides playing with the algorithm itself, I found it interesting to see just how diverse the material of the songs was. So many different takes on the idea of an algorithm. Of course, there was a preponderance of futurist EDM, but also country, folk, and spoken word. Love songs, songs embracing algorithmic culture, songs protesting it, thrash metal songs predicting a gory future of perpetual war with robots more in line with the fear of cyber culture.

Algorithm Analysis for Big Data in Education Based on Depth Learning

This article highlighted for me how the belief in the organisation and analysis of big data from educational institutions has become systemically accepted. Or at least a primary goal of IT.

Democracy and the Algorithmic Turn

Makes a case for the extent to which algorithm design has become a global force. And shines a light on how these can shape flashpoint events like elections but also become inscribed in the digital mediation of democracy. It reflects what Williamson (2018) proposes regarding the capacity for ‘big data’ to shape policy and for policy to determine what data is collected.

Introna’s  major work on Turnitin

This groundbreaking piece is a sharp analysis of not only Turnitin’s algorithmic design but the impact it has upon the written word itself and the way students write.

Excavating AI

“in which the Internet’s distorted picture of us becomes who we really are.” – but who are we anyway? Do we expect there to be a true self which is freed from the algorithms perception of us. How is that any more authentic and real than the ‘distorted’ picture?

Anatomy of an AI System

A revealing of the materiality of Alogirthms. From the extraction of rare earth materials to the indentured / prison labourers who built Amazon echo devices.

The Prevalence of Algortihms in Education

Pretty banal article detailing all the ways that algorithms are already in schools shaping how teachers, students and institutions behave. But don’t worry, it says, the heart of school is still its human faculty.

So the algorithmic turn is in fact another part of the humanist project.

Simon Denny’s Mine

The exhibition from MONA also brings the brutal reality of data mining and the growth of AI powered products into focus.

Reflectacles

A company selling glasses that befuddle facial recognition technology. We’re opting to subvert the surveillance rather than legislate against it.

Turnitin’s podcast on the written word

Interesting to see the company producing this. Getting on the podcast wagon and occassionaly interviewing some interesting people, though never challenging the assumption that the algorithm is right.

Limitations of algorithmic recommender systems in Soundcloud

  • Based upon tags
  • Based upon musical style
  • Based upon track image
  • Not thematic content or name

This results in more o the same style of music being pushed -to the extent that one would assume the platform only houses one type of music.

Simplistic recommenders that worked just by title might render more obscure and bizarre recommendations – less suited to what the algorithm producers assume to be the way people listen to music.

This produces a way of listening to and thinking about music which reflects commercial interests rather than artistic ones.

Internalises a belief that one’s musical tastes are of a particular flavour fixed and dictated by the algorithms in our digital platforms. Does the algorithm shape our tastes or is it merely an accurate reflection of how our musical tastes take shape. The challenge is not necessarily what the algorithm includes in its recommendations  but rather how it excludes those who don’t adapt to its patterns – If you don’t tag your uploads, the algorithm is unable to process and recommend them as effectively, thus making your work invisible to all except those who seek it out directly.

The artificially social nature of these platforms (many accounts are bots) also heightens the presentation of self in everyday life – One presents the music one likes in order to collect more followers. Resulting in a system where users like and share music that will attract the most followers rather than what they may actually enjoy listening to.

The platforms and entrepreneurs pushing ‘personalisation’ of learning are based upon similar fundamental principles of commerce and advertising. They miss the key problem that education is not advertising, no matter how much they may want it to be. Students are not just another demographic to be catered to. The educational experience cannot be turned into a recommender system.

 

 

Leave a Reply

Your email address will not be published. Required fields are marked *