{"id":21386,"date":"2016-07-01T10:01:35","date_gmt":"2016-07-01T14:01:35","guid":{"rendered":"https:\/\/thesocietypages.org\/cyborgology\/?p=21386"},"modified":"2016-07-01T10:28:21","modified_gmt":"2016-07-01T14:28:21","slug":"meet-the-new-algorithm-same-as-the-old-algorithm","status":"publish","type":"post","link":"https:\/\/thesocietypages.org\/cyborgology\/2016\/07\/01\/meet-the-new-algorithm-same-as-the-old-algorithm\/","title":{"rendered":"Meet The New Algorithm, Same As The Old Algorithm"},"content":{"rendered":"

\"23507877733_a602f0f6ac_z\"<\/a><\/p>\n

Image credit: Kevin Dooley on Flickr<\/a><\/em><\/p>\n

This is a follow-up post to this essay<\/a> on accusations of censorship on Reddit and the unpredictable consequences of algorithmic quirks.<\/em><\/p>\n

Reddit is the self-described \u201cfront page of the internet.\u201d Millions of users rely on Reddit to keep them informed on a wide range of topics from world news to gaming developments to the latest in pictures of cute dogs (or, often as not, reposts of pictures of cute dogs). But what happens when the front page fails us, and how do Reddit administrators respond?<\/p>\n

In the aftermath of the Reddit debacle<\/a> surrounding \/r\/news deletions of posts about the Orlando shooting, Reddit has rolled out new changes to the algorithm that ranks posts on \/r\/all. \/r\/all is the frontest of Reddit front pages; it is the algorithmic ranking of popular posts from all subreddits that 1) choose to be featured on the front page and 2) haven\u2019t been quarantined based on questionable (read: bigoted) content. The exact details of Reddit\u2019s ranking algorithm are complicated and unnecessary for this brief discussion, but it\u2019s essentially a combination of when something was posted, how many upvotes it has received, and how recently it\u2019s been upvoted. Time + Attention = Rank. You can read more about it in this (slightly dated) explainer<\/a>.<\/p>\n

According to Reddit CEO Steve Huffman, \/u\/spez, the new algorithm has been in the works for a while but was implemented earlier than intended given recent events. The quick and dirty of the incident is: \/r\/news mods deleted a large number of posts and comments on the Orlando shooting, leaving a huge chasm on the front page that was filled with reports from the Donald Trump subreddit \/r\/The_Donald. Redditors were (rightfully) outraged that the incident was only being reported on by a widely-despised, non news-related subreddit.<\/p>\n

Huffman announced the changes shortly after the \/r\/The_Donald incident in a post<\/a> that he called a \u201ctown hall\u201d about \/r\/all. In it, he argued that \u201c\/r\/all is a reflection of what is happening online in general. It is culturally important and drives many conversations around the world.\u201d Leaving aside the grandiosity of this statement, Reddit is undoubtedly an important source for the millions of users who depend on the site to stay up to date on new, popular online content.<\/p>\n

The changes to the algorithm are, to hear Huffman describe them, relatively simple. The goal is to \u201cprevent any one community from dominating the listing\u2026 as a community is represented more and more often in the listing, the hotness of its posts will be increasingly lessened. This results in more variety in \/r\/all.\u201d In other words, \/r\/The_Donald, or any other subreddit, can\u2019t completely take over the front page.<\/p>\n

For most users, Reddit\u2019s algorithm is a black box. Content goes in, content comes out, and what happens in-between is neither comprehensible nor relevant. However, when the input and output of Reddit\u2019s black box changes significantly, everyone notices.<\/p>\n

Initial reactions in Huffman\u2019s \u201ctown hall\u201d thread were largely positive and hopeful. But in the following days and weeks, some redditors began noticing a major, frustrating change: not enough new content. An undoubtedly unintended consequence of the change was greater stability on \/r\/all, with posts hanging around longer and new content rising to the top much more slowly.<\/p>\n

Also, porn. Lots of porn. The insurgence of porn on the first few pages of \/r\/all is likely due to two factors: 1) porn is really popular on Reddit and 2) there are so many porn subreddits that the possibility of any one of them overtaking \/r\/all and, subsequently, getting pushed down the listing is less likely.<\/p>\n

Of course, porn is one of those things where even a small uptick is very noticeable. So just how much of \/r\/all is porn, relative to other things? This morning I did a rough analysis of the first five pages (top 100 posts) of \/r\/all. Here\u2019s a general breakdown of the number of posts related to each of these popular topics:<\/p>\n

Pets: 6<\/p>\n

Sports: 6<\/p>\n

Gifs (funny, interesting, reaction): 12<\/p>\n

Gaming: 11<\/p>\n

News: 3<\/p>\n

Politics: 3<\/p>\n

Porn: 12<\/p>\n

It was a lot of porn. Porn\u2019s only real competitors are gaming and gifs, and \u201cgifs\u201d comprises so many different topics it hardly seems fair to count it as a category. The amount of porn content on \/r\/all has gotten so large that some users have asked for a safe-for-work version of \/r\/all so they can browse during office hours.<\/p>\n

So the algorithm change that was supposed to prevent large subreddits like \/r\/the_donald from dominating the front pages of \/r\/all has instead paved the way for a mass of porn and significantly reduced the amount of news content, while also reducing the frequency with which content changes. If most of \/r\/all\u2019s news content comes from the large subreddits like \/r\/news, \/r\/worldnews, and \/r\/politics, then the new algorithm makes it less likely for multiple posts from these large subreddits on various topics to make it to the top pages. Meanwhile, because there are so many porn subreddits, it\u2019s easy for that content to rise to the top.<\/p>\n

In trying to develop some conclusions about this phenomenon, I spoke with my media studies colleague Nick Hanford<\/a>. He was struck by Huffman\u2019s description of the changes to the algorithm, and of Reddit generally, as constructing some sort of Platonic Ideal<\/a> of the internet; a \u201ctrue\u201d representation of what is \u201creally\u201d happening at any given moment online. This is certainly one of the failings of Reddit\u2019s \/r\/all algorithm, both before and after the change; it cannot hope to accomplish the lofty aims set by Huffman because the internet is a many-sided prism, and it cannot be reflected on a flat page of ranked, numbered posts. My internet may have significantly more recipes or cat gifs than yours, and \u201cwhat is happening online in general\u201d as Huffman puts it will vary greatly from region to region, not to mention various subcultural groups. \/r\/all can never really be<\/em> \/r\/all, it is necessarily \/r\/some.<\/p>\n

Perhaps the biggest takeaway from this story is best summed up by Mr. Hanford: \u201cPORN RULES ALL, ESPECIALLY SO-CALLED DEMOCRACIES.\u201d*<\/p>\n

Britney is on Twitter<\/a><\/em><\/p>\n

*Nick Hanford agreed to be quoted for this essay on condition that he be quoted in caps lock. <\/em><\/p>\n","protected":false},"excerpt":{"rendered":"

Image credit: Kevin Dooley on Flickr This is a follow-up post to this essay on accusations of censorship on Reddit and the unpredictable consequences of algorithmic quirks. Reddit is the self-described \u201cfront page of the internet.\u201d Millions of users rely on Reddit to keep them informed on a wide range of topics from world news […]<\/p>\n","protected":false},"author":1931,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":false,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2}},"categories":[9967],"tags":[],"class_list":["post-21386","post","type-post","status-publish","format-standard","hentry","category-commentary"],"jetpack_publicize_connections":[],"jetpack_featured_media_url":"","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/posts\/21386","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/users\/1931"}],"replies":[{"embeddable":true,"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/comments?post=21386"}],"version-history":[{"count":3,"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/posts\/21386\/revisions"}],"predecessor-version":[{"id":21390,"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/posts\/21386\/revisions\/21390"}],"wp:attachment":[{"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/media?parent=21386"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/categories?post=21386"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/tags?post=21386"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}