Commercial interests often ensure that algorithms remain unseen. Furthermore, the complex mathematical formulas usually involved mean that even if we could view them, our ability to understand how they work would be limited. Nevertheless, interest in this subject is growing as the likes of Apple and Netflix have revealed some technical information about the algorithms used to recommend content. Even without technical information, what we can always see are the results an algorithm produces, and it is these results that we want you to document and consider within this activity. In the first two weeks of this block you should spend time playing with algorithms and then documenting and sharing the results. Once again, this exercise is not assessed in itself although it is an expected part of your contribution to the Algorithmic culture exhibition we are building. The task here is quite simple: to have some fun exploring the kinds of algorithms we are reading about in this block.
A good place to start your algorithmic play is Christian Sandvig’s blog post Show and Tell: Algorithmic Culture, featuring three examples of algorithms shaping our attention: Google Instant, Facebook News Feed, and Google Ads. You might want to try out these examples for yourself, before choosing somewhere to start your own algorithmic play. Below are some suggestions, although you are free to choose any space for this task – there is likely to be some kind of algorithmic operation behind most of the web content you encounter. The important thing is to find an example that your personal interaction can affect, usually requiring you to be a logged-in user with an account.
- Google Search and Ads, or another search engine
Google’s PageRank is one of the most famous web algorithms, primarily for the way it goes about ‘optimising’ search engine results. You might use Sandvig’s example above to document your own play with Google Instant, or you could explore the difference in search results depending on whether you are logged in to your Google account (if you have one) when you search. You might also explore how your search terms affect the adverts presented to you within Google, but also within other sites that you subsequently visit. - Facebook
Following Sandvig’s example above, Facebook’s news feed presents an interesting site for your algorithmic play. Can you ‘reverse engineer’ the EdgeRank algorithm, and speculate about why certain posts are being hidden from view? What other algorithms might be operating underneath the Facebook interface? - YouTube comments and recommended videos
During this block you might explore YouTube recommended videos, which appear as a list on the right of the YouTube page. How do your viewing habits influence these recommendations? How has your participation in EDC influenced your YouTube recommendations? What happens if others – perhaps friends or family – watch YouTube with your account? - Spotify, Netflix, or similar music/video streaming services.
Exploring the algorithms within these services will probably work best if you already have accounts and some history of use. What kind of entertainment is offered to you in the recommended lists? What happens if you select something you wouldn’t normally choose? - What other algorithms are recommending content to you or optimising your communications?
As well as documenting the results of your algorithmic play, we also want you to be critical. Rather than simply assuming that algorithms automatically give us the best answers, we want you to interrogate and question the results you see. Think about the following questions when you develop and comment on your algorithmic play:
- How has the algorithm affected the options you are given or what you can see?
- How have your actions changed what the algorithm has done?
- How have other people been involved in shaping results?
- Do results feel personal or limiting? Is this optimisation, or a ‘you loop’?
- What are the ethical issues at stake with your chosen algorithm? Is there data here that should be private?
- And of course, what are the implications for digital education implied by your chosen algorithm?
When it comes to sharing your findings you can present in any format that helps to convey your ideas: a visualisation, a set of slides with spoken commentary, a piece that juxtaposes screenshots with written explanation, or something else entirely. Bear in mind, though, that it needs to be added to your gallery space within the Algorithmic culture exhibition.