With advertising and event revenues plunging in the wake of the COVID-19 pandemic, media are betting the house on reader revenues. That means creating and engaging communities has become the top priority.
This is incredibly timely. Rapidly replicating the sense of community, solidarity and belonging online is critical under the current crisis, and might just end up redefining the whole experience of consuming the news.
The Fix discussed these issues with Lior Zalmanson, a writer, lecturer, and researcher in digital culture and the information society. His focus is on social media, pricing of information products, consumer engagement, and user generated content.
Also an artist, Zalmanson explores the consequences of replacing human-driven decisions with algorithms and AI. His recent “Image May Contain” work is a part of an exhibition (recently opened and even more recently closed) at the New York New School of Design.
We met in person at a noisy Tel Aviv cafe shortly before the borders closed and our offline social lives were locked down.
This interview has been edited and condensed.
TF: You wrote a paper about turning content viewers into subscribers. Could you tell me more about that?
LZ: Subscription fees are an increasingly prominent revenue source for news organisations. But when most online news remains available for free, publishers need to find new ways to persuade readers to pay.
In a more recent study, together with Efrat Nechushtai from the Columbia Journalism School, we looked at newspapers in America and how they appeal to readers with “please subscribe” messages. We examined how the 55 most-read US newspapers frame their value proposition when asking readers to subscribe and analysed the informational, social, and normative elements mentioned in ‘subscribe now’ webpages.
TF: What did you find?
LZ: Firstly, we found out that every major newspaper in America basically belongs to one of four groups. So there are basically four decision-makers and they usually decide marketing campaigns for more than one paper.
They usually copy the same appeal. Our data shows newspaper subscriptions are often promoted as a relationship with social and normative dimensions, rather than merely an information-based transaction – with some differences in the language used by newspapers of different sizes, ownership types, US regions, and local political orientations.
Newspapers use a very social – community language, the idea that you are not subscribing to a service, you are joining a community, and that is something else. Of course, the question remains of whether this is just branding or something more substantial.
TF: How can it become more substantial?
LZ: It’s not enough to just add participatory options and hope that users will click and interact. To see any real economic benefit from user engagement, a website has to have a clear strategy in place.
This is based on research I’ve done with Gal Oestreicher-Singer from Tel Aviv University that introduces the concept of the ladder of participation — a framework for strategic thinking about using site engagement to improve conversion. The notion of the ladder is based on the idea that, as users become increasingly engaged with a website, they become more willing to pay for its services — and the website must take an active approach to engage and interact with its users, guiding them “up the ladder.”
To do this, website managers must understand the current user community, identify how they would like that community to evolve, and select the participatory features they offer the order in which to introduce them.
For example, a study by Dikla Perez, Gal Oestreicher Singer, Matthew Rubin and myself, shows that the use of feedback surveys may lead to consumer loyalty and study whether simple ‘calls to action’ — prompts that require the user to rate the content or service — encourage monetary conversion.
There are also chat boxes which are half-communal, I would say they help you personalise and allow you to get answers to very specific news-related queries.
TF: That is the social part. What about “the normative” dimension?
LZ: We found that people who subscribed were not necessarily the most avid users, it was people who mainly consumed the “community elements.” We understood it to be, at least partially, that you would subscribe to show your loyalty and pay respects to the community and reassure your status within. So a normative commitment is when you feel that you have to, you have to be a part of it because all of your friends are or you are being expected to showcase your loyalty in some way.
In other words, people subscribe to a newspaper because It is a part of their identity. It is about a status symbol, it is a commitment.
AP: Is this community-focused rhetoric good or bad, and is it really about “community”?
LZ: The idea of the rhetoric appeals came from a larger question that I had in mind regarding the nature of the relationships between users/consumers and news organisations or content organisations (because I am also interested as well in video,d arts, and music websites).
I do see that with user generated content and the idea of engaging users in a meaningful way, you also bring about a lot of challenges. I am talking about things such as content moderation (if you get user generated content and suddenly you need to employ all these people who will moderate the content and make judgment like decisions about what stays and what gets deleted) and having very different norms, that originate from users’ activity versus firms’ guidelines, colliding.
Moreover, you start to almost expect your audiences to work for you in a certain way, when asking them to contribute content. And the line between “using” and “abusing” is very thin. What you are saying is that you want to build a community but what you really want is money, people’s time and effort, and you are not going to pay them for it. In the end, It can result in social features that are there just as a revenue model and then community is merely-token participation.
TF: Are you optimistic about online communities replacing physical meeting places?
LZ: I used to be more optimistic about the online realm as a social meeting place a decade ago and then I saw that people basically still crave, for many reasons, being out and physical interactions.
I now see the second wave, which is mainly a VR wave. Facebook’s vision for VR is a social VR and they believe it could be used for virtual meetings which was something talked a lot about in the time of Second Life, but now it is different because you are supposed to log-in with your real-life identity. It is not a fantasy world, it is not just for games.
I am also kind of sad that from the early 21st century the internet is one business sphere where I need to keep my consistent identity across the platforms. Even give our identity politics reality where everything is supposedly more fluid. The internet is now this strictly business entity where you make money from things which is why you have to keep a very clear sense of identity and a very clear sense of who you are talking to.
TF: Where do you think this need moved, this need to be anonymous?
LZ: There are some groups still which are more about anonymity, like reddit which is still pseudonym-based, even though if you showed this platform to a designer at the beginning times of social media it would look so outdated to them. Think about the UX! So reddit is like the 90s internet, it is what is left of it in a way. And it is still very successful and I guess the meaning is that people still want to converse in an environment where they can protect their identity…
TF: Your installation is a part of “The Question of Intelligence — AI and the Future of Humanity” exhibition at the New School in New York. What is it about?
LZ: In 2017 I have begun researching the notion of artificial intelligence and accessibility, and how artificial intelligence developed on the biggest platforms, to help people who are disabled. I aim to show that this development, in many ways, created new forms of inaccessibility. The work I am exhibiting in New York is called “image may contain” and it is a work where I took famous photos, mostly from the 20th century,, and ran it through an image recognition algorithm that was put in place by facebook to describe photos for the blind. I wanted to see how it flattens their meaning by taking them out of their social context.
My installation juxtaposes some of the images (such as e.g. Kennedy’s assassination) with other images that the algorithm describes using the same signifiers (in this case: “10 people, car”). By looking at all those photos together we are hopefully also understanding something about algorithm as a curator or as a social curator, as lenses to a reality that we are facing, because it is not just about what a blind person imagines after hearing those words, it is about other choices made by these type of algorithms, based on the same flattened meaning which will influence what we are exposed to and what we will see or not see online.
TF: The phenomenon of algorithms as social curators has created a new type of audience that all content creators cater for — an “algorithmic audience”. How do you think the existence of an algorithmic audience impacts the news world?
LZ: Algorithmic audiences learn from us — humans and prioritise the content we are most likely to click on serving us more and more of the same. Also, as we know, the more clickable content tends to be the more outrageous one. Algorithmic audiences have also forced us to introduce a new metric imposed on us by the advertisers – in order to stay on our website, our content consumers need to be “engaged”, and so our content needs to “engage” them. What it means differs from content provider to content provider. The effect it had on the music played by Spotify is that, for example, more songs start with a chorus – no more psychedelic rock like long and slow intros. We need to be engaged from the very beginning…
TF: How do algorithms help news organizations optimise user engagement?
LZ: Since most of the business models of current media websites are built upon an assumption that you want to have a lot of user engagement – you want it in the millions, but you only want to employ dozens of actual news people. You then really need the algorithm as a sort of a middle layer between the audience and your staff. So it could be a chatbot, or it could have to do with moderation, personalisation.
In many cases algorithms are now sort of like the agents between journalists and audience: they can restructure or re-engineer the website to maximise or optimise for engagement, they can design specialised versions of your website in order to make specific user personas click more… But they can also raise many issues because — again — since they are built to optimise certain metrics, they usually disregard some other things completely, so you need to supervise them.
Again, if you are looking at your user as not just readers but you expect them to actually exert effort and time, and contribute in many ways. They, meaning the users, could be really, really frustrated and angry sometimes if they think they are only speaking with bots and the algorithms, and so that the specific brand or entity doesn’t really care about them.
TF: So there is no replacing the human factor for now?
LZ: We have recently discovered that humans are biased, and so what we call “the cultural canon” is also biased which sometimes means racist and non egalitarian. On top of that, behavioural economics tells us that humans are not very rational beings. We are making decisions that are not very optimal, so in many situations, people have begun thinking, well…maybe we should let an algorithm decide for us.
On the other hand… algorithms are also socially constructed and they are also biased, and they don’t understand everything. So I think what we are grappling now with is the combination of algorithms and people. Because sometimes you need a human curator to be able to find something that is rare and unique, and spotlight the obscure, unknown, and find a true genius in that. If you, as a CEO of a newspaper, are now employing algorithms to manage your audiences, you need to make sure you are not treating the audiences themselves like robots. That is, I think, the biggest challenge, for the next – I think – few decades.