Welcome back to the Baekdal Plus newsletter. Today, we are going to talk about changes. First in relation to AI and publishers, and then about social media and publishers.
I know that there have already been about a million articles published about ChatGTP and natural-language AIs. But I want to show you what happens if you go into the specifics. So in my latest Plus article, I decided to show what it can do if you ask it to do my work as a media analyst.
The result was both very impressive, but also very problematic. But, it illustrates the challenges that we now face (not just me, but the entire media industry). Not just in terms of it doing our work, but also how it misrepresents that work.
So take a look at: How scared should we be of AIs taking our jobs?
As we all know, the concept of social media has changed tremendously over the past fifteen years, and with every change, it's like we are losing yet another part of the "social element".
One of my friends, Avinash Kaushik, recently wrote a very good article about this, called "Bye Bye Social Graph". He talks about how the algorithms work now, and how it is no longer about the people we interact with (the social graph) but about broadcasting content to people with a specific interest (the content graph).
As he writes:
Seemingly overnight, we went:
Before: Content's relevance and value being almost exclusively driven from the closed environment of our individual social graph (people we knew, and the people they knew).
After: Content's relevance being derived exclusively from the type of content I might like, regardless of who published it, regardless of where in the world they published it.
This is simplifying it a bit, but...
Before: If I published a sexy and amazing video of how to use Excel to do algorithmic sorts (yes, that can be sexy!), it would primarily be shown to my social graph.
After: My sexy and amazing video would be in front of all like-minded Excel lovers.
The compelling benefit of unshackling content from the social graph:
Five people in my network to, literally, infinity. (IF the content deserves attention.)
This has a lot of implications for creators, publishers, the audience, and even something like privacy, and Avinash also explains that well (you should read and subscribe to him ;)). But, I want to give you my take on this from a publisher's perspective.
First of all, the biggest change this is creating is the destruction of the direct relationship. The content graph might be very good at creating views across a massive amount of people on a platform (like on TikTok), these views are random from people that you likely never had a relationship with before.
Fundamentally, this is a problem, but it could be fine if two other things happened. First, that traffic needs to then also come to you, and secondly the traffic you then get needs to be monetizable.
To put it simply, if you post a video on TikTok and it gets a million views, this literally means nothing if people then just stay on TikTok. A video like that is not helping you as a publisher, it's only helping TikTok.
(Also remember, as Hootsuite points out, a TikTok view is a heavily inflated metric. It's measured like this: "Different social media platforms measure "views" in different ways, but on TikTok, it's super simple: the very second your video starts to play, it's counted as a view. If the video autoplays or loops, or a viewer comes back to watch it multiple times, those all count as new views." So when you see that something has a million views, that's not the number of actual people.)
You need to somehow get people to come directly to you. But since TikTok does nothing to help you do that, the only way you can accomplish this is via second-level referrals.
What is a second-level referral? Well, a first-level referral is when people come to you directly from your posts. So, if you post something to Twitter, with a link to your story, and people click on that link to go directly to you, that's a first level referral.
A second-level referral is when people have to, independently, visit you on their own.
This, of course, is nothing new. This is how outdoor advertising works. For instance, if you walk down the street and you see an ad on a bus, it too might be shown to a large number of random people. And then you hope that at least some of these people decide to later come and visit your shop.
This is second-level referrals. It's when people (independently of the first interaction) have to take another separate step in order for you to get any benefit from it.
So, how many do this? From bus ads? ... or because you posted something on TikTok? Well, I have no idea. I have never seen any reliable metric for either.
But let's say that some people do, which is just going to be a miniscule amount, like a fraction of a percent, now we have to monetize them. We can do this in two ways. Either as advertising, or by converting them into subscribers.
Well, advertising doesn't work as well as it used to, so even if you get a million views on TikTok, that's not going to amount to much in terms of ad views on your own site. And, in terms of subscribers, we have a new problem.
Remember, the social graph is about the people who follow you, whereas the content graph is about random people who probably have no idea who you are. The result is that this audience is far less likely to convert.
I mean, think about it. The social graph allowed us to connect with people directly, and then use that connection to nurture and influence people over time, gradually building up momentum that could be used to convert them.
The content graph is like starting over from scratch with every single person. It's totally random, and so there is no momentum being built up.
This fundamentally changes our relationship with social platforms.
This is also true for new tech startups. For instance, Instagram's co-founders, Kevin Systrom and Mike Krieger, are working on a new text-based social app called Artifact, "a personalized news feed that uses machine learning to understand your interests and will soon let you discuss those articles with friends."
What is it based on? Well, the TikTok content-graph model. As the Verge put it:
The simplest way to understand Artifact is as a kind of TikTok for text, though you might also call it Google Reader reborn as a mobile app or maybe even a surprise attack on Twitter. The app opens to a feed of popular articles chosen from a curated list of publishers ranging from leading news organizations like The New York Times to small-scale blogs about niche topics. Tap on articles that interest you, and Artifact will serve you similar posts and stories in the future, just as watching videos on TikTok's For You page tunes its algorithm over time.
This sounds fine until you realize what this means. The posts that people see are random. As a publisher, you are not connecting with anyone. Your posts are just part of the stream (if the AI picks them), but there are no followers, no connections, no nothing. As the publisher who creates the content people see, we are being reduced to "random bits for an AI to pick out".
Now, as Avinash wrote in his article, the potential they promise is more traffic. But, as a media analyst, I have heard that promise before... from countless tech platforms over the years. I mean, remember Oatmeal's comic from 2017?
Mind you, for other parties, it's much more promising. For a brand just seeking more exposure, the content graph has a bigger potential reach than the social graph. So, if you just want an advertising effect, it's fine.
What is also interesting is how it changes the way we define relevance. As Avinash also wrote: "Why does the home page of the NY Times still have 97% irrelevant content after they have [seen] my behavior, as a paid subscriber no less, for five years?"
This is an excellent point. Why indeed!
So, inside the newsroom, the content graph is important to think about.
Also think about advertising. For decades we have been told by the ad tech companies that there was only one way to create relevant advertising (aka, by giving them all our data so that they could create a social graph). So, isn't it interesting that all the big tech companies are now themselves focusing on this other model.
Think about what this potentially means for the future of ad targeting and programmatic data sharing. Maybe what the ad tech companies have been telling us all these years is not the best way. Maybe there is another way...
If you haven't seen them already, don't miss out on the 'known to work' series, where I talk about the things we know work for publishers.
I want to mention one more thing before I sign off today. It's about how publishers sometimes shoot themselves in the foot.
As a media analyst, I come across this quite often. Sometimes newspapers are called out about something, but instead of listening and addressing the problem, publishers often double down instead and just make things worse for themselves.
The latest example of this is from the New York Times, and... I mean... this screams of a lack of situational awareness.
What happened was that yesterday, quite a very large group of journalists signed an open letter to the New York Times, pointing out the problem they have with the "newspaper's reporting on transgender, non-binary, and gender nonconforming people." I'm not going to repeat the letter here, you can see it in full here.
What happened next was that the NYT responded, and they basically said that there was nothing to talk about and that they were proud of their "both sides" focus.
And then... and I mean... they published this:
I mean... seriously? This is basically giving the finger to every single person who signed that letter.
As a media analyst, my very simple advice to you is this. Please, for the love of kittens, stop behaving like this. Not only is this directly undermining the work relationships and culture within the New York Times. But it also massively polarized the readers into either people who now hate and might cancel their subscription to the NYT because of their lack of understanding, or those who now appreciate how the NYT is helping them promote anti-trans sentiments.
I'm not making this up. Go over to Twitter Search and just look up what people are tweeting about the NYT right now. You will see exactly these groups in action. So instead of creating balance (which is not even a thing in this context), they actually ended up creating even more conflict.
And, this is just one example of many.
We can have a long discussion about the underlying problem, but I'm not going to do that here. But please, stop shooting yourself in the foot.
Also, remember that while this newsletter is free for anyone to read, it's paid for by my subscribers to Baekdal Plus. So if you want to support this type of analysis and advice, subscribe to Baekdal Plus, which will also give you access to all my Plus reports (more than 300), and all the new ones (about 25 reports per year).
The more we use automated tools, the more important it becomes to also create 'originals'
Young people will cancel and come back later... if you let them
Everyone is talking about ChatGPT and MidJourney, but their size is also their downside.
Why BuzzFeed News failed, and the coming increase in news avoidance
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