Sorry, we could not find the combination you entered »
Please enter your email and we will send you an email where you can pick a new password.
Reset password:


By Thomas Baekdal - February 2016

How We Lost Social Media To Algorithms

As you may have heard, Twitter now features an algorithmic timeline where they will show you the 'best tweets first', just like on Facebook. So... we just lost Twitter.

Mind you, at the moment this new 'feature' is opt-in and only activates when you have been away for a while, but we all know that this won't last. The main reason why they are doing this is to make it easiest to use for newcomers and low-end Twitter users, but opting in is only something power-users would do (or rather not do, because we all hate the idea). So it's only a matter of time before the algorithmic timeline will become the default.

But I'm not annoyed just because of my opinion of it. My real concern is how much damage it is causing to our future ability to reach, distribute and influence our audiences. Over the past five years, we have seen a growing problem with 'social' in how it has forced us all into the same rat race.

In many ways, social media has changed. At first it was a social platform where we could connect and interact with whatever we cared about. But as they started to optimize for social scale, the value of social has increasingly turned into a smaller and smaller niche in which social media today is really only relevant for a specific type of content linked to a specific type of behavior.

This behavior is what we call a micro moment for people on a break. And anyone who doesn't make that type of content or isn't designing for that specific behavior has experienced a drop in social engagement and reach.

This is a problem, because our world isn't defined around a single type of content or a single type of behavior. The real future of media is much more varied.

But let's talk about why this is happening, which leads us to the algorithms.

Why social algorithms fail

In theory, algorithms are great because they will look at all the options available and try to match that to us as individuals. And in many cases they work quite well. But there is an inherent problem with all algorithms, which I will now illustrate with food.

Imagine that you had a page with four different types of fruit. You have mangos, apples, strawberries, and a banana.

Now imagine that you ask three people to tell you which two fruits they like the best, by rank.

The first person will say that she likes mangos the most and bananas second. The second person says she prefers strawberries first, and, again, bananas second. The third person says that he likes bananas the most, and apples second.

You then put those into a spreadsheet and rank them based on how much 'engagement' each got, and you end up with this:

As you can see, the bananas were the type of fruit that generated the most amount of engagement, which means they created the most amount of traffic... or what Twitter says:

We've already seen that people who use this new feature tend to Retweet and Tweet more, creating more live commentary and conversations, which is great for everyone.

No Twitter, this is not great for everyone. It's only great for the bananas. One-third of the audience actually prefers mangos over bananas, but those tweets no longer show up on the timeline.

We are not getting the best fruit. We are getting the best average fruit. That's a big difference!

A similar problem is based on how we spend our time.

Let me ask you a simple question. What did you have for breakfast today? Well, in my case it was cornflakes.

Is this the best breakfast in the entire world? No, of course not. Not even close. It's not bad, but it's certainly not great.

This is what we are seeing on all the social channels. When we just look at what people engage with, most of the time we are engaging with things that aren't that important to begin with. It's not because people are shallow, but more that they don't really care.

This is the reason why you eat cornflakes. It's not the best breakfast by miles, but at that moment in time you don't care that it's not the best as long as it is simple. In other words, you have no intent when you engage with cornflakes.

These two patterns define pretty much every social algorithm. First, they only identify the best average instead of the actual best choice. Secondly, that most of the choices we make aren't very good to begin with. And when the social channels then start to rank the content based on these signals, we end up with content that drives a lot of engagement and activity but which means very little.

Obviously, if you can create a product that matches this perfectly, you will end up getting a lot of traffic. This is what Kellogg's is doing. Kellogg's is very good at creating simple, cost-efficient, high-traffic cereals for people who have very little intent.

But of course, the social channels are constantly trying to target people even better, so you constantly need to optimize for it. For instance, imagine that you learn that you could target those people who liked mango; now what would you do?

The answer is simple. You create cornflakes with 'real mango', just like Kellogg's is doing.

You see what is happening here. They have optimized for the latest tactic, which gives them a slight edge over all the other cornflakes producers, but it's still cornflakes.

We are still not close to achieving the 'best breakfast ever', because optimizations like these start out by focusing on people with no intent.

In comparison, take a look at this breakfast. It is unbelievably good!


The problem here, of course, is that this type of breakfast will only appeal to a smaller group of people, because like with the fruit example above, many people will prefer something else.

Also, this type of breakfast requires you to make a decision, to have an intent, so it won't be your everyday breakfast. But you see the difference in quality?

This is the problem with social media. It's optimizing for a moment when people care the least, trying to surface the 'best content' within that very low-end baseline.

This is also why I love YouTube so much. On YouTube, we have all these creators who don't care about the algorithms at all, but are instead focusing entirely on building up momentum through their value and passion. They can do that because they are encouraging people to subscribe to them directly and because the 'subscribe' tab in the YouTube app isn't trying to rank the content.

In other words, the chronological subscriber timeline means that our individual choices define the value, not those of other people. It's a really important difference.

In fact, we clearly see the failure of social on YouTube as well. YouTube is trying, like all the other social channels, to rank the 'best content' to build traffic. It does this on its front page and with the 'trending' tab.

But, once they do this, they make the same mistake as all the other social channels. Here, for instance, are the videos YouTube thinks are 'trending' for me. They are all crap!

Most of these aren't even close to my demographics. Most of them have the same shallow crappiness of all other algorithm-optimized social videos (focusing on jackass-type of fun), and some are outright obnoxious.

The only video that I'm even slightly interested in is the one about Bourne Identity, not because it's 'the best video ever' but because it's slightly above average.

So I never turn to the 'trending' tab on YouTube. I always go directly to either a specific YouTuber that I'm interested in, or to the 'subscriber' tab where I can see a chronological list of only the YouTubers I have specifically chosen to follow.

The image above illustrates what YouTube's algorithms think is 'the most engaging content' (and by looking at the traffic numbers, they might be right), whereas the image below illustrates my chronological subscription stream.

These are all things I have chosen to follow because they are actually valuable to me (for different reasons and intents). We have podcasts that I love, singers that I admire, awesome science videos, good Let's Players, some new videos from Disney Animations, tech news and some super delicious recipes.

You see the difference? The algorithms suck, while the chronological lists don't.

This is the reality of all social algorithms. They are not even close to being good, on any channel. All they do is to surface the type of content that people engage with when they have very low intent. This is a systemic problem with the very concept of how social media works.

The result is, as I tweeted earlier this week:

For those who haven't realized it yet, Social has become a niche form of media, useful only for a certain type of content/behavior.

Finding signals in the noise

At this point, you may be thinking, wait a minute, don't we need algorithms to filter our streams considering that most people follow so many things?

As Twitter puts it:

You follow hundreds of people on Twitter - maybe thousands - and when you open Twitter, it can feel like you've missed some of their most important Tweets. Today, we're excited to share a new timeline feature that helps you catch up on the best Tweets from people you follow.

Isn't this valuable? Well... yes and no.

The first thing we have to realize here is that we aren't solving the problem. The real problem is that people shouldn't be following thousands of people to begin with.

By encouraging people to 'engage' with as many things as possible, social sites are encouraging a behavior where people would follow someone without a specific intent. This is not a good thing.

This whole problem is basically a self-inflicted wound. Instead of creating social signals, we have created social noise. And the best way to solve this is to stop following people you don't care about. Encourage people to form real connections based on real choices instead. Don't solve this by removing intent.

It sounds good when Twitter says:

When you open Twitter after being away for a while, the Tweets you're most likely to care about will appear at the top of your timeline

In reality though, what they are actually saying is:

Don't bother thinking about what you really want. Just sit back and we will do all the thinking for you.

Remember, intent is driven by people's deliberate choices. If you ask people to not make any choices, you get less intent. And less intent means more 'cornflakes'.

But there is another point to be made here, which is that we have been here before.

Think about the old print newspapers, or our old linear cable TV packages. They are doing the same thing. They are saying: "Don't bother thinking about what you need. Just sit back and and let us decide for you."

That's what linear TV and print newspapers do. They algorithmically replace your personal choices with the convenience of not having to decide. Of course, they do it with people rather than a computer, but the concept is the same.

And look at what these human algorithms did to old media. Look at traditional newspapers. Are they filled with only the best and most relevant content just for you?... or are they filled with the type of articles that aren't really that amazing nor really that bad? In other words, are they merely the 'best average'?

It's the same with traditional TV channels. Are the shows on those channels the best you have ever seen? Well, sure, a few are. We can all mention really good TV shows that we absolutely love. But, for the TV channel as a whole, 95% of the programming isn't really what you wanted. And worst of all, many shows that you really liked are often cut to 'make room' for shows that are optimized for a more general audience. It's the same concept as what we see on social media channels today.

And then look at the trend of what is happening to them. Print newspapers are in an existential crisis and cable TV is losing out to on-demand. On-demand meaning that you let people decide for themselves. Meanwhile, we are seeing a tremendous growth in individual media by YouTubers and other creators who aren't defining themselves around what the algorithm wants them to do.

Ask yourself, why would a model that has already failed once (algorithmic based programming) suddenly be the future of social media? We are not solving the future. We are repeating the mistakes of the past.

I do understand why people think algorithms are what we need to solve social abundance. I get it. For instance, Marshall Kirkpatrick tweeted this shortly after Twitter had announced the changes to its timeline:

But when we then look at Marshall's account data, we see that he is following 16,800 people.

Can you imagine having a Twitter timeline with tweets from 16,800 people? If these people only tweet three times per day on average, that's one new tweet every 2 seconds... and most of them are from people you have no personal relation or interest in?

What a nightmare!

So, I get it when people like Marshall don't see this change as a 'big deal'. To him, his Twitter stream is already so filled with noise that it's impossible to use, and he is probably already using excellent tools like Nuzzel to surface only the best content.

To him, having Twitter tell him that they will surface "the Tweets you're most likely to care about" sound like a pretty good deal. Compared to what he had before, it is.

But it's not the right solution. The right solution would be to have three streams:

We need a focused (chronological) stream of tweets from a smaller group of people that we truly care about. The people we have chosen specifically to follow on an individual basis because of what they mean to us.

This is your default view.

We need an algorithmic stream of tweets from what we would call your 'circle of interest'. This is a second stream that you can choose to look at (as an extra) whenever you like to get a summary of 'the best since your last view'.

Twitter is already doing this with their Discovery tab, and it's not bad. Although its one tweet per screen 'card design' makes navigating them a painfully slow experience. You will never 'discover' that many things before you get bored with it.

Of course, Twitter is also doing 'Moments', which I don't even want to talk about. It's as bad as YouTube's 'top trending' tab.

Finally, we need an intent based stream in which you can query the best content in relation to a specific interest. For instance, if a person you follow links to a post about Disney's annual results, you can click on that and ask 'what have other people said about this?'

This is what you see over at Mediagazer where they are expanding the 'discussion' about each topic, and it's a brilliant way to expand your views.

The first stream is based on our specific choices. The second stream is based on discovery. And the third stream is based on a need.

Can you imagine how much better our social channels would be if this was how they worked?

The problem isn't that Twitter has created a service that summarises just the best Tweets. The problem is that they are trying to make it the default. The problem is that the social algorithms almost always give us sub-standard results (the bananas or the cornflakes) and that they are doing it as a replacement for making real choices.

They are using algorithms to replace our intent. That's not what we want.

From a trend perspective, the result is pretty obvious. Social channels are increasingly becoming a niche form of media. In traditional media terms, social media channels have turned into what we usually define as a tabloid journalism channel.

That's what Facebook has become, and that is what Twitter is becoming.

Look at how non-tabloid type publishers perform on Facebook, and you will notice that almost all of them are doing pretty poorly. Look at the high-value niche verticals, and you will see that Facebook is almost meaningless to them. Look even at the successful YouTube creators, and you will find that they too don't really get much out of Facebook.

Choose a path

Obviously, Facebook and Twitter aren't crap nor should you just forget about them. That's not what I'm saying. My point is that social media is no longer good for everything, and it hasn't been for a long time.

For brands, this means that we have to think about social channels like Facebook and Twitter as marketing channels with paid-for content strategies (whereas you would think about Instagram, YouTube and Snapchat as momentum channels).

But for publishers it means taking a very hard look at whether this very specific type of behavior fits your editorial focus. If it does, great... go nuts optimizing for it and enjoy the silly amount of traffic you might get.

If not... well... that's when you have start thinking about social as a marketing channel rather than a distribution channel, and whether the ROI of creating things specifically for Facebook is worth it. For some publications it is, but for others it's not.

An extreme example would be Nieman Lab, which has a Facebook page with 50,000 likes. This is quite good, until we look at each post and we see a very low amount of activity. Obviously, the moment and intent of Nieman Lab doesn't match the intent of how people use Facebook.

So it now has three choices:

Each one of these choices could work but for very different reasons, and the monetization options would be differently focused depending on which path they choose (which isn't really a concern for Nieman Lab, but is a huge concern for any other publisher).

Every publisher has to ask whether the social niche is the right one for them, and many are now realizing that it isn't really what they want.

And it's not just the traditional publishers who are saying this. Many digital natives are wondering about this too. Already, many YouTube creators have deemphasized Facebook because it doesn't match their focus.

Many have instead turned to Twitter, YouTube, Instagram or other channels where the chronological timelines allow them to focus on their type of value. We can still do that with Instagram and YouTube, but it seems we are losing Twitter to the same trend.

This leaves us with a vacuum.

We have been so distracted by this amazing social phenomenon that we have forgotten that there is much more to life than just the micro moments. We lack a good platform for vertical macro moments.

There have been several attempts. We have had news aggregators like Zite, which solved the behavior but brought with it the same problem that we have on Facebook: that people are no longer making real choices (and thus we lose the intent). It was interesting, but it was like cable TV... doomed by concept.

We have platforms like Flipboard, which again, solved the behavior. But since it is built on top of the social platforms, which are focusing more and more on the micro moments, Flipboard's baseline is moving away from where we need it to be.

We have tools like Nuzzel, which are absolutely brilliant for summarizing your Twitter stream and turning that into a daily newsletter, but again... it's build on top of a platform that is becoming more and more 'micro moment focused'.

Then we have news readers like Feedly, which are actually quite brilliant. But it's build on top of RSS which is very poorly supported, quite difficult to understand for newcomers, and lacks the much needed integration that allows us to have a two-way conversation.

And, of course, we have email newsletters. These actually work quite well but it's unrealistic from a trend perspective to believe that people will consume media by subscribing to individual newsletters for each site they like in the future.

In other words, we already have so many options for summarizing our streams for when we just want to get the big picture (email, news aggregators). We also have so many options for all the content we just consume when we are on a break but don't really have any specific intent (Facebook etc).

What we don't have is a channel where we can follow and stay up-to-date with all the other content as it happens. The content that we have chosen to follow because of a specific intent. The content that isn't designed to be 'engaging', but is designed to be worth knowing.

Or rather we used to have a channel for that. It was called Twitter.

This is the challenge that we now have to undertake.


The Baekdal Plus Newsletter is the best way to be notified about the latest media reports, but it also comes with extra insights.

Get the newsletter

Thomas Baekdal

Founder, media analyst, author, and publisher. Follow on Twitter

"Thomas Baekdal is one of Scandinavia's most sought-after experts in the digitization of media companies. He has made ​​himself known for his analysis of how digitization has changed the way we consume media."
Swedish business magazine, Resumé


—   strategy   —


A guide to using AI for publishers


How to fix people's perception that climate news is not useful?


A conversion that (never) ends. Mapping publisher funnels


Addressing news avoidance will help every other element of publishing


Managing churn from start to finish


What to consider before switching from subscriptions to memberships