Marko Zovko

I am a programmer currently interested in the intersection of systems-level programming, cybersecurity and AI.

I don't think social media is a good place for humans anymore

images/i-dont-think-social-media-is-a-good-place-for-humans-anymore.png
March 27, 2024

Today is the perfect time to delete your social media accounts. I know it’s 2024, and this isn’t a new topic to anyone, but there are two main reasons I think this is especially important right now. First, 2024 is a super electoral year in the world, quarter of the world population is having elections 1. The second reason is that generative AI is here, and it’s not going away. As social media, I define anything that has an infinite feed.

Don’t worry, this isn’t some thinly veiled argument to get you to vote for someone. The people who will be elected this year will control a big chunk of the world’s money flow. Since a lot of money is at stake, there is a direct financial incentive to deploy huge budgets to influence the situation. That’s on top of already deployed budgets that convince us to buy stuff we don’t really need. Add to that formula cheap text generation, voice cloning, and image generation, and it starts to get interesting. I think video deepfakes are still not here cheaply, but they do exist. It seems like a perfect opportunity for startups to fix that this year, maybe someone is releasing it as I write this.

Both of those reasons are tied together in Dead Internet Theory. It’s an idea that started somewhere in the depths of the internet, and it states that most of the activity online is perpetrated by bots. We cannot know for sure if the content we are consuming was created and posted by actual humans. All content systems on the Internet can be simplified into three parts. Agents who create and post content, a way to deliver the content, and agents who consume the content. No, I didn’t type agent instead of people randomly. Since this post is focused on social media, in an ideal world, at least in my imagining of it, that would translate to people creating interesting content. Then, The Algorithm shows that content to other people who find it interesting, and they consume it.

Social media algorithm

In the ideal system above, The Algorithm controls 1/3 of the system and people the other 2/3. But are we in that ideal system? The obvious question with the definition above is: what do we consider interesting versus what The Algorithm thinks it’s interesting? I initially thought that I would write here something clever about how The Algorithm thinks. But I would be a liar if I said I knew. It’s not even one single algorithm. What I can claim is that The Algorithm has to measure something to decide what interesting is. That’s the thing I do know. Even if we don’t understand what it does because it’s hidden behind a complicated neural network, it still has to follow some instructions. What we do know is what The Algorithm measures. Views, clicks, read time, watch time, comments, likes, votes, reports, followers, subscribers, and other things that we call engagement.

Which means that the graph above is not correct, because it’s missing a piece. From all those aggregated measurements, it has to provide something for consumers to consume. The feed cannot be empty, or someone would get fired. As long as the creative agent’s interest in engagement is not zero, it will either create a positive or negative feedback loop about that type of content. If we start caring even a little about engagement, the more we use social media, the more we are invested in the feedback we get.

Social media algorithm 2

The question then is: what do you think The Algorithm rewards? By this time, I assume people reading are either thinking that I’m stating the incredibly obvious or that I’m a tin-foil enjoyer. I would bet a decent amount of money that there is no in between. That is what I think The Algorithm rewards—polarization. That is just my opinion, and I jumped to that conclusion. Whether you agree with me on that or not is not that important. The question I consider far more important is: is this useful? Technology should provide a better quality of life and solve actual problems; is The Algorithm doing that?

I cannot help but wonder, with cheap and easily available generative AI and monetary incentives, how many creative agents will be there in the months to follow? Anecdotally, Youtube is already filled with AI-generated stuff since every part of a generic Youtube voice-over video can be easily automated. Proof of Twitter and Reddit bots is one search away. LinkedIn is full of ghostwriters that will grow your business. It doesn’t seem too farfetched that AI will become part of their process. I don’t see any obvious reason that is stopping people from doing the same on other social media sites.

That’s the creative side, but what about the consumer side? We know that Instagram and Twitter follow-for-follow bots have existed for a long time. I technically have a Twitter/X account; I use it once in a blue moon, and I’ve never posted anything. Somehow, I have 36 followers. All random names, random images, weird descriptions, and 0 actual posts. Buying YouTube views and Spotify streams is a thing that is documented as well. As I was writing this, campaigns for elections in Croatia started, and we directly saw posts engaged by accounts from Vietnam. Maybe the good people of Vietnam are deeply invested in the Croatian elections, but somehow I doubt it. We can get infinitely creative in how we can apply these automation tools, but the point is that they exist on both sides of the equation.

Social media algorithm 3

Which leads me to the second important question: what is the ratio of bots to people on both sides of the equation? Things in life never fit neatly into the binary options, but I do think they provide useful mental shortcuts to visualize the situation. Sort of axes on where we are and where we are going.

  1. People are the majority, and The Algorithm rewards useful content.
  2. People are the majority, but The Algorithm rewards not useful content.
  3. Bots are the majority, and The Algorithm rewards useful content.
  4. Bots are the majority, but The Algorithm rewards not useful content.

Still, the picture feels incomplete. Other questions that pop into my mind are: should we really consume content? Is that what we are - consumers of content? How do you feel after using something with an infinite scroll? Should I care about all the problems in the world? How much time do I spend on this? Is that time better spent elsewhere? Do you let anyone into your house randomly? Why then let other people’s thoughts enter our minds? Is this too much stimulation for our caveman brains? How is your view of, well, everything, shaped by the frame of reality The Algorithm portrays? Should we experience boredom more often?

These are all the questions I wish I could provide easy answers here. It would make for a far better post. But I can’t. I only have strong opinions on the first two questions. I firmly believe we are already in option 4 or are approaching it quickly. The creation side of the equation is, in my opinion, less debatable, so I’ll start there. When you post something online, you at least put some thought into what you are posting. I’m sure you would post more stuff, but it takes too much time to prepare your content. So what if you can just type make this sentence witty "..."? What if instead of manually selecting the photos you want to share, you just dump 20 photos into some application that selects the best ones, automatically applies filters, and automatically posts the best ones? What if you just dump 10 video clips and tell the AI to make a montage reel of your vacation? What if you tell an AI tool to create clips from your podcast and it does it instantly?

We are all self-editing our online presence, but when you offload that editing to increasingly available AI tools, how much of your creation remains? What if you want to grow your following online for fame and fortune. Why would you not use a ghostwriter, a team, or a tool? It’s still you that’s deciding on what the lines are, it’s just that someone else does the coloring. Is that wrong? I don’t think it is by itself. The reality is that the cover image for this post was generated by Midjourney. Of course, we all want to put our best foot forward when presenting ourselves online. If you open the newsfeed  and see 20 items, each improved by AI tools by, let’s say, 30%. The question then becomes, since that percentage adds up, are we in aggregate consuming content from people or AI tools? Since there are still not that many generative AI providers, shouldn’t that mean that, in the end, we are consuming inherent biases from those few datasets?

What if I want to be a bad actor and exploit social media to push my narrative dishonestly? Could I not cheaply generate 100 social media posts, then launch them at the same time from 100 accounts, and see what sticks? After some time, pick the ones with the most engagement and use them with other accounts. What if I generate 10000 posts? This is the first primitive idea that popped into my mind, but what happens when teams of people work 40+ hours weekly to solve this creatively? What happens if there is a military conflict and there is a direct military advantage to pushing your narrative? Maybe you later find out the truth, but those stories are still somewhere in your head since the availability heuristic has already done its work. But what if you never find out what is wrong?

Consuming side of the equation scares me more, because all the examples above have a human somewhere in the loop. Bots that create engagement don’t. How would you know why something gets shown in your feed? How would you know if those views and likes were from real people? I hope nobody actually reads comments online, but for the sake of an experiment, pick one social media site that has comments enabled. How many comments can you find that are clearly bots? How many comments are suspicious? How could you ever tell the difference?

The companies owning social media sites cannot detect all bots. It’s simply impossible. At best, it will always be a game of cat and mouse; at worst, the bots will be running freely in the wild. Theoretically, all social media could ask you to verify your profile with a government ID and then segregate verified and non verified people. But do we, as people, want that? Would that cause other problems? Do social media companies even have an incentive to do it? As long as it brings engagement, and engagement leads to spending on ads, why would they fix it? If half of the world population is having elections, I don’t think the ad spend budget is going to be a limiting factor. Surely, at some point, those budgets won’t be justified anymore, and the problem will have to be solved. But what happens until then?

What happens when those two things run in parallel next to each other? One side of bots creates, the other one consumes, and people meander somewhere in there. If you spend a lot of time online, your whole view of reality could be warped to what the bots think people want to see. They are not bad bots, they just do what they are told to do. When people talk about AI catastrophe situations, I don’t think we are heading for a Terminator-style future. And I honestly don’t think that’s even the worst eventuality. But I do consider this the most important challenge we face regarding AI today.

It’s pretty obvious by now, I cannot shake my bias that it’s probably best if we let bots play by themselves on social media, and we go somewhere else. I’ve contributed with enough negativity, so here are what I think are good next steps to try to solve the problem for yourself. These are bits and pieces of actionable steps I found online and use daily to achieve that. But I have to state a disclaimer first, all actions I’m talking about have the undercurrent that we, as individuals, are in control of this situation. We are not. In the grand scheme of things, I don’t think it matters if you or I completely quit social media. Other people will still use it.

This realization is incredibly comforting, at least to me, for two reasons. First, it removes all the childish pretense that not using social media is somehow morally correct. I’ve been guilty of that thought many times. Second, if it’s all the same, why not get rid of something that you don’t want to partake in? Maybe the second and third-order consequences of that decision will bring something better.

The first step in every optimization is always the same: take stock of the current situation. Track what you use the most. Smartphones have screen-time applications that track your usage, and browsers track your history. Every week, review that data, see what bothers you the most, and decide on one social media platform that you want to get rid of. Look up more information on how algorithms work so you know your enemy. Look up scrolling and internet addictions. This is, I guess, the only good time to be a hypochondriac.

The second step is to create friction between you and the thing you want to stop using. The goal is to prevent those moments when you unconsciously use social media on autopilot. Use website blockers, delete apps from your phone, and turn off your internet if need be. Then stop algorithm reinforcement. Use all the possible features to curate your feed: delete, mute, block, unfollow, and unsubscribe from anything and anyone that is even mildly reinforcing your behavior to use it. Turn off history if you can, don’t use recommended feeds, use historical feeds only. There are even interesting addons such as Youtube Unhook, which disable almost everything from the UI that draws you in to click and consume more on Youtube specifically.

The third step is to give it up completely for a short time and then start using it again. Write down how you feel when you use it. Maybe it’s not a good decision to leave it, maybe you really need it. I don’t think I’m going to quit LinkedIn anytime soon, it still has benefits. But you can never know for sure unless you get off it for some time. Fill your new-found time with something interesting. After some time passes, it’s entirely possible that you will look at your time spent using social media as one of those “what the hell was I thinking?” situations.

Maybe the only positive thing that comes out of this is that we return to how the internet used to be. Separate private forums where people gathered based on their interests, but that’s a topic for another day. During the writing of this post, I’ve realized that it’s been six years since I’ve deleted Facebook and Instagram. I wonder how many pointless arguments in Facebook groups I have missed. I don’t even want to think of all the “fun” I missed during the health crisis of 2020. I do wonder how many people I used to know think I’m dead.