Upvotes. Likes. Digs. Internet points are common across the web, and used to signal agreement without needing to comment that. But with the massive growth of the internet, unforeseen consequences of using points to signal agreement have made many wonder if they were a good idea at all.
Youtube
Youtube removed it’s dislike button in an attempt to prevent brigades from happening. Allegedly. Before the button was removed, there was a common thought in Youtube’s creator community that the like button did nothing to actually boost the video. Youtube’s algorithm is pretty opaque outside of a few hard rules, so this rumor spread to the point where PewDiePie, one of Youtube’s largest creators, posted a video specifically requesting people dislike his video, and it seemed he was proven right – that video got nearly four times as many views as other videos he made that month did. That felt horribly counterintuitive: “are we supposed to hit the dislike button on creators we like so more people will see them?” but it’s actually a result of including the dislike button as a positive point in their website’s calculations for interaction.
PewDiePie’s video got so much traction because it asked for something new. Fans watching always hit the like button, right? And people who felt neutrally about the video or just had it on as background noise didn’t interact with it at all. However, by asking people to break their routine, he accumulated several times as many button presses, both likes and dislikes, as he normally did, thus telling Youtube that tons of people were actively engaging with his content and convincing the algorithm to share it far and wide.
While funny, it shows how so many conspiracy videos got shunted to the front of the Youtube Recommended page before and during the Covid-19 pandemic. Likes, dislikes, and comments are all interaction, right? So a ton of dislikes and a ton of arguing in the comments are equal to a ton of likes and a ton of praise for the creator, right? Obviously not. But the website, which was programmed to understand any interaction as good interaction, couldn’t tell them apart.
Reddit treats downvotes as interaction, but it doesn’t do so in a way that rewards bad content… at least not directly like Youtube accidentally did.
If you go to Reddit, and you see an Ask Reddit post asking something like ‘What is Your Most Controversial Food Opinion?’, the top result is likely to be something pretty innocent, or even well-liked – maybe pineapple on pizza is at the top, goofy, but not an abomination. The comments then inform you that you should sort by controversial if you want to see the weirdos eating pickles and sardines together recreationally. Those people answered the prompt, and their food choice is probably the most controversial in the thread, but many people downvoted them anyway, leaving pineapple on pizza the winner.
Reddit ranks comments and posts with an up- and down-vote system. Upvotes carry comments to the top, downvotes carry comments to the bottom, and if a comment gets enough downvotes, the comment can even be automatically hidden. Reddit insists that you use the upvote button on comments that add to the conversation, not as an agree button, but only smaller and strictly moderated subreddits can really make that happen. Everywhere else, the ‘upvote’ may as well be a ‘like’ to the population of Reddit who barely skimmed the rules before making an account. Therefore, even though sardine man answered the question with something unusual and new, he loses to pineapple on pizza, an incredibly milquetoast combination that’s not offensive unless you’re from New York or Chicago.
This system means that the first few people on a post determine what comments appear at the top, because those are the comments that have time to accumulate upvotes.
Reddit defaults to showing either the new comments or the top-voted ones to users, and while you can sort by other metrics, the path of least resistance is to browse what’s already the default. This sometimes leads to nice-sounding but incorrect information appearing at the top of the comments under a post, and anything contradicting that information may be downvoted because it doesn’t sound as good or isn’t written as well. That’s if the correction gets spotted at all! The top comment on a front page post may have hundreds if not thousands of threads and subthreads trailing off of it, so if someone spots some incorrect information and tries to correct it, the odds that they’re successful in doing so rely on them A) writing well, B) being spotted by enough people scrolling down, and C) commenting in the right spot, off a subthread on the problem comment, not the general ocean of comments where it will float to the bottom.
This system creates a culture where being likeable and confident over text is more important than being right. The now infamous jackdaw argument is a great example of this coming to a head. For context, a user with the username ‘Unidan’ would pop in with fun science facts on Reddit posts. Anywhere he posted, he’d get floated to the top of the comments, because he came across as a sort of ‘Bill Nye’ type, fun, educated, and cool. And then he got into an argument over what a jackdaw really is (in which he was pretty condescending, but hey, it’s the internet, and people found the condescension really funny when he was right) and got banned because it was discovered he’d been manipulating votes. Specifically, he’d been downvoting everyone who posted at the same time as him and upvoting himself with a number of throwaway accounts. The result was that his comment was on top first, granting him more visibility and more internet points.
As a result of his banning, people watching finally understood the problem. Unidan, who was well-liked, couldn’t be challenged by other less charismatic scientists, even if he was wrong or not precisely accurate, because people would dogpile any criticism of him. This is another issue with the upvote system! Users see comments with upvotes, they upvote. They see comments with downvotes on them, they downvote. Now, other users can see who’s ‘losing’ an argument by who’s being downvoted, and people like to side with winners even if the winner is, technically speaking, wrong, or at least oversimplifying. Unidan as a biologist is not qualified to be giving speeches on physics, the same way Neil Degrasse Tyson as a career physicist shouldn’t be talking about biology. People are willing to call Tyson out on Twitter, but they weren’t willing to call out Unidan on Reddit, partially because the anonymity made people absolutely vicious. The other person arguing with Unidan over the jackdaw thing got death threats.
Facebook likes have a storied history. Facebook’s origin as a ranking site for college students is not the most graceful or morally upright a website has ever had, and it shed a lot of its original flavor and features to reach its current size and social stature, for good or bad. Sure, it’s a horrible monolith determined to spy on you and sell you things based on the info it gathers… but Farmville was fun, right?
The like system was incredibly straightforward, and worked quite a bit like Youtube does – comments and likes are interaction, and interaction is good, so more people should see a liked, commented post because it is good content. Of course, some people take ‘like’ to mean ‘I like this’ and some take it to mean ‘I’m interacting with this so I can find it again on my timeline’. When Facebook started sorting content feeds algorithmically instead of chronologically (meaning your more popular friends with more interaction would pop up first in your feed instead of whoever posted most recently) finding posts you wanted to share when you were back in front of your computer was unnecessarily annoying without it.
Unfortunately, as a social network, conflating the two meanings of the ‘like’ button could spark arguments, and so Facebook added other reactions. The conundrum of showing support for the passing of a loved one by ‘liking’ the post was memed on for years before Facebook decided to add sad and angry reactions to the mix. Facebook has no dislike button – every option is an input of emotion, not a vote or a simple ‘dislike’.
Comments, as interaction, boost a post the way likes do even if the content is atrocious or dangerous and the comments are simply calling the poster out on it, which is obviously not ideal and mirrors the Youtube issue from before. Conspiracies with lots of vitriolic arguments are better for engagement and so they’re what get spread. The quality of the content on Facebook is suffering because longer, worse content (looking at the people who are mixing drinks in a toilet or dumping food all over a counter) gets more angry reactions than good content does with likes. The same goes for Instagram, even though it only has likes and comment counts – to argue, to clarify, to warn, etc. counts the same as to compliment or praise in the algorithm’s eyes. Speech filters designed to detect angry language are in the works, but it might be too little, too late. A culture is established, and arguments are good for interaction.
Tumblr
Tumblr’s like and reblog system is perhaps the best system you can make on a chronological website. The few algorithmic systems in place cater to you based off of previous interactions like any site does, but no other major social media website is willing to offer a purely chronological option like Tumblr does. You can turn every suggestion off. All of them. While the chronological system still rewards interaction and engaging content, it allows users a ton of freedom to see what they want to see, not what the website thinks will get them to stay. I give all this backstory because Tumblr has had likes and reblogs for forever, with a very recent update that includes the ability to comment on posts without requiring that reblog like it did before.
Tumblr’s like system is fairly unique – likes work like they do on other sites, with a separate page to come back to so you can find them again, but you can turn off the option for anyone else (except for the original poster) to see your likes. Some functions like the ‘in your orbit’ function allow your followers to see what you’ve liked, but only if you have those likes visible and only if they have the ‘in your orbit’ turned on. While all internet points systems have their flaws, this is probably the most capable and least manipulative out of all of them, at least on the website’s side.
That doesn’t mean information can’t be spread on numbers alone. Most famously, a post suggesting that it was possible to get infinite chocolate out of a chocolate bar by just cutting it a certain way made the rounds. People tried it, and got the ‘free’ square suggested by the GIF, freaked out, and then posted about it, further spreading the rumor. The reblog system, while less manipulative than any algorithmic feed could ever be, still has a pretty sizeable flaw in that corrections also have to be reblogged for you to see them. If you follow Blog A, and Blog A is an aesthetic blog focusing on sweets, they might have reblogged the post about the infinite chocolate. When the correction comes out alongside a criticism of everyone’s internet literacy (Tumblr was the first web site a lot of preteens used, so ‘people can lie on the internet’ wasn’t immediately obvious to them) Blog A might not reblog the correction because it didn’t fit their aesthetic… or because it just straight up didn’t come across their dashboard in the first place.
At least users have to build their own echo chambers out of other blogs and blocked tags on Tumblr and Reddit – Twitter and Facebook do that for you, and you might not even be aware it’s happening until you log out.