My intention is not to distract from the more "explicitly" material and infrastructural issues raised in the readings for the upcoming week, but there's an article I read a little while ago that made me think of another type of digital waste: hardly-watched, random online videos. Just go to YouTube and search any keyword, then filter results to the most recently uploaded. There will be lots of videos with little to no views, and many of them will stay that way forever.
 |
| (This is just an example of a random user I found by searching "grass" in YouTube. This 22-second video is literally of 2 ducks in the grass. The user has a prolific number of videos from ducks to pinball machines and concerts... and hardly any views. It's actually really interesting. Are these "waste"? What gives me/anyone the right to label digital content like this as "waste"?) |
But if everything and anything online becomes part of a real, material archive, taking up real, material space and energy in a building in some cold place––unwatched, unengaged-with online media (especially the ones that do not get deleted) are also material waste. I can't say I'm totally convinced of the urgency to study unwatched digital media, deleted or not, but it's an interesting thought: that these sorts of "wasteful" videos––literally wasting the energy of its place in a data center––may have an aesthetic, cultural, even counter-algorithmic significance of their own. They are ghostly digital flotsam that no one cares about, no one sees, but might nonetheless have material footprints and potential meanings of their own?
(The article: https://online.ucpress.edu/fmh/article-abstract/8/2/219/169788/Towards-a-Methodology-of-Unwatched-Digital-Media)
This is fascinating to think about. It brings to mind Jani Scandura's notion of "depressive modernity" from Down in the Dumps (https://www.dukeupress.edu/down-in-the-dumps/) -- though Scandura's sites were all about the materiality and place-based nature of discards. In general, I wonder if "critical discard studies" as articulated by Max Liboiron and Robin Nagle might be helpful for thinking about such "immaterial" unwatched content: https://maxliboiron.com/2014/07/31/discard-studies/. Also makes me wonder about the relationship between content-as-waste and waste-of-time.
ReplyDeleteThis is a really interesting article and provocation about unwatched media. It reminds me of when I worked at a digital media startup that curated online content (Upworthy). We were famous for making content "go viral", but our strategy wasn't driven by algorithms or a secret formula, but rather by iteratively testing content with users. Our editorial team was incentivized to publish as much content as possible so that we could display each piece to 100 users at a time. Its fate was determined by whether or not those 100 users engaged positively with it. If it did, it was sent to an audience of 500 users, then 1,000, then 5,000, and so forth; the audience kept expanding as long as it outperformed other content published at the same time. In other words, the system relied on increasingly large samples to predict how the broader population of our audience would respond.
ReplyDeleteBut precisely *which* 100 users got the privilege of testing that content was not completely random. If an editor published a piece at 2 AM ET, it would likely be tested by our Australian audience; at 6 AM ET, our European audience; at 10 AM ET, our US east coast audience.
I think about this whenever I come across a TikTok on my FYP that has fewer than 100 likes. I feel conflicted about the expectations imposed on me: on one hand, I feel privileged to be entrusted to submit a highly weighted signal of quality (positive or negative) to the machine learning system, but I also feel annoyed that my attention is being exploited as labor.
All this to say: unwatched digital media is likely being tested in a similar way, using tiny audiences (among other sources) to elicit signals of positive quality. Under this perspective, unwatched media is "waste" from a user's perspective, but actually very fundamental to ML systems that rely on vast samples of diverse content in order to sort and filter with high efficiency. In other words, our waste is the algorithm's sustenance.
Blog post 10/10 (Hyejoo)
ReplyDeleteHamsini and Rohan, thanks for your intriguing comments. It's my first time hearing the term "critical discard studies" but it's cool to think about engaging with the "waste-of-time" as a critical practice. Maybe even as protest against the hyper-efficient, fast-paced time of our contemporary neoliberal hustle culture, which we increasingly see on our social media feeds as well (like the quickness of ads that pop up right after you watch someone's "story" on Instagram, or simply the short lengths of TikTok videos). Taking seriously the waste-of-time theorizes nicely with something like Jack Halberstam's concept of failure. But like you said, Hamsini, factoring in the question of waste complicates an overly optimistic view of digital wastes-of-time when they are, also, "content-as-waste."
Aaaaand this gets even more complicated considering what Rohan shared (I really appreciate your firsthand experience, btw): that even if one watches/engages with a waste-of-time as a critical protest, even this is free labor one is giving away for machine learning and thereby (unknowingly) for greater engagement (and probably, ultimately, profit) for the website/platform. No online video can become "popular," I guess, without the unpopular... the unpopular/unwatched/waste-of-time, as Rohan shows, can be extremely crucial for machine learning systems to work at all.