Assembly, a response to the breakdown of trust
The advertising business model is poisoning our discussion-spaces. What are the problems and what are the solutions?
We have an urgent problem with us today. We don’t trust each other. Divisiveness, hate, outrage, fear, civil unrest, disinformation, conspiracy, censorship, fake news. It’s difficult to understand what is going on in the world. I believe many of these issues have roots in the way we “pay” for information.
In this article I will explain the problem of incentivizing media with advertising revenue, and how this problem has evolved alongside new technologies. Then, I will propose a solution that addresses the root of the problem.
What’s the business model?
In 1980 CNN introduced “always-on” 24-hour news coverage for the first time. Other networks soon followed. The idea was simple: more attention (from viewers) equals more revenue (from advertising). Journalism has always been funded by a combination of subscriptions and advertising, but the new technology of television unleashed a huge new audience that marketers could now tap into. If only the station could keep people watching…
Well, “keeping people watching” is a clear-cut goal, and profit is a powerful incentive force. Advertising has been the primary source of revenue for media since the inception of the newspaper industry in the latter half of the 19th century. It was only natural for this pattern to continue with modern mediums of communication.
How the business model would thrive under technological advancements in telecommunication was less obvious. History is messy, and it is difficult for us to anticipate these big slow-moving problems. Even so, there were a few canaries in the coal mine.
In 1895 Joseph Pulitzer ran the New York World and William Randolph Hearst had started up the competing New York Evening Journal. The 32-year-old Hearst arrived in New York City and bought the failing Journal with his family fortune. Hearst set up an office in the same building as Pulitzer’s World and poached the entire Sunday edition staff from his competitor. Needless to say, Pulitzer was not pleased.
This contentious rivalry boiled over when a headless torso was found in the East River. Both newspapers engaged in a vicious tug-of-war over coverage of this latest murder mystery. They used bold headlines, cartoons, gory images, dramatic writing, and intrigue - even offering rewards for readers to provide tips directly to the paper so they could gain the advantage. It was the first tabloid war, news as entertainment, or as historians would later describe it: yellow journalism.
Even as yellow journalism became the norm, things weren’t so bad. The following decades were known as the Progressive Era, a highly activist and reform-focused period of our US history. It was during this period that President Teddy Roosevelt held the first presidential press briefing. A new generation of muckraking reporters followed in the wake of icons like Nellie Bly, who had gone undercover in a mental asylum to expose mistreatment and neglect of patients. Corruption was being exposed, and crimes were being busted.
The effects of the business model of attention became more noticeable only as technology advanced. The next canary sounded off right after the Great Depression. Advertising revenues had plummeted, and the established newspaper industry was determined to hold out against a new technology that threatened everything they’ve built - radio broadcasting.
President Franklin Delano Roosevelt was looking for a way to get the word out about a “new deal” he was working on, but he didn’t have much support in the beleaguered press. This resulted in the fireside chats, a series of 30 informal evening broadcasts that went directly to the people. In many ways, it was America’s first podcast.
Up until this point, the newspapers had held a monopoly. Now, a growing audience was turning to radio, and it represented an existential threat to the industry. FDR’s chats were very popular, attracting up to 70% of all radio listeners by the time they concluded in the early ’40s. The tension boiled over (in what was dubbed as the Press-Radio War) when the papers tried to stifle radio and keep it out of the reporting business. It was a noisy and embarrassing chapter for the press, as well as a complete and utter failure.
What brought low the high-minded ideals of the newspapers? The same thing that threatens any industry: self-preservation. Advertising wasn’t working anymore, papers were failing, and radio was a threat to the big players that remained.
The news adapted more gracefully with the advent of television, but the consequences of the business model were the same. It was a dog-eat-dog world, and advertising was paying to keep the lights on for everyone. CNN’s innovative “always-on” news coverage meant television networks now had a strong incentive to steal away as much human attention as possible, starting a sort of arms race that would continue steadily even to the modern-day. Newspaper circulation soon peaked for the last time, in 1984.
There was another canary in the coal mine with the invention of the internet, this one louder than any that had chirped before. Journalism was once again caught completely off guard by new technology. There was excitement over the possibilities, but little consensus about how it all should work. What would the business model be? How would reporters tap into the wide reach and distribution potential of the internet?
It was a Big Problem. Unfortunately, it never got solved. The news industry gave away their stories for free on the internet and turned to ads to foot the bill. The incentives of advertising revenue once again lead to sensationalism. Shorter was better, clickbait was king, viral was hitting the jackpot. The quality of stories diminished until we were just calling it “news content.” Long-form, nuanced, hard-hitting journalism wasn’t necessarily aligned with the bottom line; it took way too long and was too expensive. It could still get done, but it was harder now.
There are experts that have talked about this problem (ex. 1, 2, 3), so I won’t wade too far out of my depth here. What I can say is that, as a consumer, I am not satisfied with the state of journalism. In fact, my default reaction is to avoid the news, and the data show that I’m not alone.
The problem is, there’s never been a worse time for our large-scale sense-making machinery to fall apart. Humanity is progressing exponentially at this point in our history. We have new technologies that have changed the shape of life on our little blue planet. We have more power than ever before. We can do things now our ancestors would never have imagined, and we will continue to add more to our toolbox of tricks in the future. We are living through a transformational time in history that will require us to make very important choices about the future we are creating.
We can’t make good choices if we don’t have good information. There aren’t any shortcuts. It requires quality education that teaches critical thinking skills. It requires freedom of speech that allows for an open exchange of ideas without fear of retaliation. And before any of that can even happen… it requires trust.
The most recent canary, and the final nail in our trust coffin, was the advent of social media. On this topic, I highly recommend checking out a documentary on Netflix, The Social Dilemma. The doc is a bone-chilling account of how our technology has moved beyond simply showing us ads for stuff we want to buy. At first, the goal was to keep people engaged - engineering the most addictive user experience possible to keep people coming back to the platform. Then the nature of the game changed to predicting and influencing our behavior directly.
Early on, we discovered the miraculous power of recommender systems. You may be familiar with these systems through the video recommendations on YouTube and the newsfeed on Facebook. Here’s how it generally works:
Everything you do on a tech platform is collected in big databases. This includes what you read, what you watch, what you like, where you click, how long you hover over something with your mouse, or how long something remains in view on the smartphone screen.
This data is fed into AI models that are tasked with figuring out what kind of person you are. You’re classified into a group of people that have similar interests and behaviors.
Content is then recommended to you to keep you (a) in the app longer, and (b) coming back.
The algorithm tweaks the balance until your online behavior is highly predictable - meaning that the AI can essentially predict what your future behavior will be.
This information is used for targeted advertising, showing you only ads that are likely to be highly relevant and highly effective.
This doesn’t sound so terrible, right? If you have to see ads, why not have them be highly-relevant ads?
The problem is the fact that this has been so damn effective. The systems built for social media have successfully tapped into our psychology, and humans are now predictable. Social media allows people to disappear into worlds of their own creation, full of content cherry-picked by algorithms to reinforce wildly different worldviews. The tech has inadvertently amplified powerful and ancient aspects of our human nature.
Imagine you’re Russia or China and you’re seeing the United States start-up these huge tech platforms. You take a closer look and find out that Facebook provides very sophisticated marketing tools. You’d be able to advertise directly to very specific groups of people, down to the zip code. You can even target people based on gender, age, work, relationship status, purchasing patterns, device usage, activities, and interests. And the whole process is 100% automated, with practically no human oversight.
Is it any wonder that Russia’s Internet Research Agency exists, and actively meddles in the United States using memes, groups, and fake accounts? Is it any wonder that Tik Tok is spying on us for the Chinese government? In 2019 Facebook removed over 5.4 billion fake accounts from its platform. That’s quite a number considering there are only 7.8 billion people on the planet.
Renee Diresta is a researcher that has studied these issues extensively. Here is an excerpt from a statement she prepared for the Senate Select Committee on Intelligence, “Open Hearing: Foreign Influence Operations’ Use of Social Media Platforms.” (The entire statement is a must-read.)
Propaganda and malign narratives have existed for a very long time, but today’s influence operations, which co-opt popular social platforms, are materially different – the propaganda is shared by our friends, often in the form of highly effective, shareable, immediately graspable memes. It is efficiently amplified by algorithms, and the campaigns achieve unprecedented scale. To conduct an operation, adversaries leverage the entire media ecosystem to push a narrative and manufacture the appearance of popular consensus. The operation is planned on one platform, such as a messaging or chat board. Content is created, tested, and hosted on others, such as Reddit, Pinterest and YouTube. It’s then pushed to platforms like Twitter and Facebook, with standing audiences of hundreds of millions of people, and targeted at those most likely to be receptive to it. The platform’s trending algorithms are gamed to make the content go viral - this often delivers the added benefit of mainstream media coverage, increasing attention via traditional media channels including television. If an operation is successful and the content gets wide distribution, or a manipulative Page or Group gains enough followers, the recommendation engine and search engine will continue to serve up the content on an ongoing basis.
The result of our technological connection is clear now. Polarization is profitable. Facebook was founded in 2004, and the following year saw the founding of YouTube and Twitter. In the past 16 years, the United States has become more divided and more fearful.
Our technology has only amplified the problems in our media. Last month, the Institute for Advanced Studies in Culture released their 2020 report. The report surveyed 2,205 adults in a nationally representative sample. Their findings focused on the growing partisan divide, but the survey also asked some tough questions about the news media. This chart is taken from that report, titled Democracy in Dark Times.
The data is somewhat validating of the evidence we have all been seeing with our own eyes. A vast proportion of Americans, democrats and republicans alike, are concerned about the problems in our news media. We’ve come to a consensus on this issue. Now is the time to roll up our sleeves and get to work.
Big Problems, Bold Solutions
Our incentive structure for sharing information at scale is married to the advertising business model. More attention equals more profit. This incentive was workable at first, but we’ve recently gotten really really good at it. Human nature has been hijacked for power and profit, and the business model of attention has become too dangerous to tolerate.
To solve this problem, we need to flip the business model of sense-making on a large scale. It’s certainly a challenge, but it’s not hopeless. In fact, there are many positive projects on the problem-solving team. An early example is the peer review system used in academic research, where scientists review each other’s research papers before being accepted for publication. Wikipedia is a crowd-powered system that allows people to collaborate on creating a shared understanding of the world. The success of podcasts has shown that millions of people still crave long-form nuanced conversation. The platforms of Patreon and Substack have shown that people are willing (and happy) to pay if they see creators making content that feels authentic and trustworthy.
That’s the kind of solution I’d like to propose for journalism: a system with better incentives than the current business model.
Large scale changes only happen as a result of individual choices. Tesla CEO Elon Musk knew people weren’t going to buy electric vehicles if the car wasn’t worth the money, so he focused on making it a consumer choice that makes sense for people to choose. The result is a car that is reliable, fast, fun, and safest in the industry.
What does an information media landscape that “makes sense” look like for a consumer?
Private - We don’t want our data collected and weaponized against us for profit.
Free - We expect news to be free, though many of us are willing to pay for quality.
Convenient - Staying informed should not require an excessive investment of time and effort.
Relevant - We want to support quality local journalism that is relevant to us.
Engaging - We want to understand what other people are saying about the story, how the discussion has changed over time, and what the current consensus is.
Trustworthy - We want to follow people rather than institutions because people feel more trustworthy.
This is a tall order to achieve, and for a long time, it hasn’t been possible to create such a system. This has changed recently, thanks to advances in several areas of technology.
In the following sections, I dive deeper into a few ideas that make it possible for us to transform our media landscape. What does real, trustworthy news look like without advertising? We have an opportunity now to find out.
the role of AI
Machine learning and artificial intelligence are areas of intense research in recent years. These topics cause quite a stir - equal parts excitement and dread. One thing is certain: these are powerful tools that will continue to change our lives in the future.
I believe machine learning is the key to building a healthy information ecosystem, one in which we would be able to hold meaningful discussions on a large scale. More specifically, we can apply AI to information tasks that would be boring and slow for humans to do themselves.
Topic models are machine learning models that lend themselves to the problems of our information ecosystem well. We can use topic models to summarize and rearrange large amounts of text, in order to get a sense of what topics are being discussed without having to read every word.
Allow me to provide an example to help illustrate how this might work. At the beginning of the COVID-19 pandemic, Bill Gates was one of the first to hold a large-scale discussion on the Reddit platform. This was an AMA post (ask-me-anything), which is the Reddit equivalent of a question-and-answer session. In the following hours over 11,000 comments were submitted. Mr. Gates responded 34 times to various questions and comments.
It’s impractical for any single person to sift through such a large number of comments. Instead, we rely on sorting options (Best, Top, New, Controversial, Old, Q&A) to get a glimpse. Even then, we’re only seeing a sliver of the discussion, and it’s difficult to get a sense of how representative that sliver is.
This is a challenge that machine learning is particularly suited for: large amounts of data. We can use topic modeling to create a picture of (a) what topics people are talking about, and (b) how many people are talking about each topic.
The nice thing about this is we get a sense of something that’s normally difficult to achieve in our current online discussions: scale. Instead of thinking to ourselves, “it seems like a lot of people are worried about their kids going to school during the pandemic” we can actually see the proportion of people discussing that topic.
We can reimagine the COVID-19 discussion on a platform that gives us these insights live, as they occur. We would see which topics are growing and which topics are shrinking as time goes on. Instead of sifting through a mass of text, Bill and his co-hosts could discuss the questions live on video.
It would be just as if we were watching an interview on TV, but with the added benefit of knowing what all of the other viewers are saying about it. You could even take out your phone and type out a question directly to the hosts who are running the program. They might not see it, at least not right away, but there’s a chance you aren’t the only one asking that question. There’s always the chance that it bubbles up onto their radar.
Unlike the algorithms in Facebook, Google, and YouTube, the AI doing the topic modeling here is looking only at the comments, not at humans. It isn’t controlling the discussion in order to maximize profit or attention, and it isn’t manipulating the topics in order to produce more text in the form of new comments. It’s simply organizing information so we can get a sense of the discussion at a glance.
putting the news on stage
Understanding what the audience is thinking while the discussion is fresh is a game-changing prospect. Why? Because of trust.
Think about the last conversation you’ve had face-to-face with someone. This experience is the pinnacle of our ability to communicate with one another. In an in-person conversation, you are not only aware of the words that are being spoken, but also of body language, microexpressions, tone, and even chemical signals like pheromones.
You may not necessarily trust or like the other person, but you do trust yourself to figure out what is happening.
This is something that is missing from our modern asynchronous information environment. There is information everywhere, and there’s a lot of it. It’s noisy and disembodied. We have no way of evaluating the freshness and ‘truthiness’ of it, at least not at the moment. Not live.
The ancient Greeks had an interesting concept of communication, perhaps best exemplified by Socrates. He believed that when information is written down, it inevitably loses some ability for change and growth. It is removed that much further from its original form, that of the spoken word.
Oral storytelling was celebrated in these ancient societies. The Greeks built huge amphitheaters to accommodate performances of poetry, music, dance, and theater. They were hubs of culture - places for political assemblies, philosophical debates, speeches, and for sharing information from far-away lands.
One of the major technological breakthroughs of the time made this possible: acoustics. The ancient Greeks discovered how to build structures that amplified the higher-frequency bands of the speaker’s voice on stage while simultaneously filtering out the low-frequency sounds of background noise from the audience. We have the ability to use our technology now in a similar way, to reimagine a virtual amphitheater for the digital age.
There’s a movement growing in the world today: putting the news back on stage. The idea is simple and inspiring. It is summed up well in this short excerpt, pulled from an article by former BBC journalist Catherine Adams.
The rising “techlash” has revealed a desire for experience and liveness. Audiences want proximity, as Walter Benjamin pointed out, and perhaps more than ever, humanity. Physical reality can serve as an antidote to the virtual epidemic.
This article was written in January of 2020, in a pre-COVID world. Adams discusses how newsrooms in Finland, France, Spain, Romania, South Africa, Syria, UK, and the US are experimenting more and more with “live news” events in physical theatres and auditoriums. Adams writes,
Hearing from journalists in the flesh humanizes both the stories and the writers and lifts the veil over newsroom practices. Attendees at these events were glad to have the chance to ask questions, participate in a discussion, and potentially influence editorial strategy.
Unfortunately, since the rise of the pandemic, these live news events have been severely diminished, but the problems they intend to address remain with us. The time is ripe for some radical problem-solving.
Assembly: putting the pieces together
I propose building a live-streaming platform for crowd-powered local journalism. I’ve been calling this concept Assembly. Here’s how it would work.
Customers pay a monthly subscription, comparable to a Netflix or Spotify subscription.
Journalists tell their stories in live events on the platform. The event could be a talk, speech, Q&A, debate, interview, whatever it takes to tell the story.
Customers watch any event that interests them. They can also tune in to a “channel” of live events that is especially relevant to them, like local news, sports, science, or politics.
During the event, customers can comment and question the journalist. Topic modeling will make sure everyone can see which topics people are talking about, and how many people are talking about those topics.
At the end of the event, customers rate and review the journalist. Journalists make the lion’s share of profits from customers’ subscriptions, and highly-rated journalists make more money – encouraging high standards of quality.
Customers schedule future programming for their local news channel in the online Assembly community. The community can also assign experienced journalists as editors, just as if the local channel was a physical newsroom.
This business model is a two-sided marketplace, like Uber or Airbnb. It serves customers and creates work for journalists. Assembly makes a space for local news that is funded by local customers, while still permitting the global stage we’ve gotten accustomed to in a connected world. There are no algorithms deciding what events people see and when they see them. Everyone sees the same thing at the same time, regardless of who they are and where they are watching from. It’s just actual people telling newsworthy stories, under the scrutiny of their local community.
a final word
I believe we need a place in society for discussion, clarity, and consensus. A place that will not be so easily corrupted by large sums of money and aggressive foreign powers.
Something special happens when large groups of people look at the same thing and come to a shared understanding. When Martin Luther King Jr. gave his 17-minute “I have a dream” speech, it was enough to spark radical change in a decades-long fight for civil rights. When Apollo 11 landed on the moon in 1969, the world was collectively mobilized in an exciting new age of space exploration.
These moments in history are galvanizing and empowering. They reveal in us a shared humanity, an ability to come together and change the world around us. Imagine if these moments came around more often than a few times in our lifetimes.
The future before us has many challenges, but it will be a future that will see many such moments. Will we be arguing amongst ourselves when these moments arrive? Will we wonder whether the source of the information can be trusted? Will we be living in a world where we can trust our eyes, our ears, and our minds?
I believe we can have that future, one in which we do come together to solve really big problems. One in which we use tech as a tool for good, rather than allowing it to rule and control us.
That’s the dream behind the Assembly concept: working towards a future in which we use technology to support human discussions. Creating a space that nurtures sense-making, collaboration, and consensus.
I don’t really care if I end up being the person that brings something like this to market. The problems in our media need solving, and I believe Assembly can help move us in a better direction. Steal it if you like!
Just please please don’t sell out to the advertising industry.
join the conversation
We’ve been working on Assembly in stealth mode for about a year now, and we’re ready to take it to the next level. We’re excited to announce that we’re launching a crowdfunding campaign to raise our first round of funding.
Every day, we see more and more people passionate about our problematic media landscape. If the campaign were to gain steam it would send a strong message that people are fed up and want something different. Help us build Assembly the right way, with forethought and care, and in partnership with local communities.
I hope this article can help spread awareness of the problem. I would love to connect with other people that are working on these issues and work together on finding solutions.
Thank you for reading! Inspired? Here’s how you can help:
Share this article with the link below.
Pass it along to friends and family and get the discussion started.
Send it to your local newspaper or news station, or to your favorite podcaster.
1/29/2021 edit - Our crowdfunding campaign has come to a close. Although we were unsuccessful with the campaign, we still managed to raise over $11,000 from 67 sponsors, largely from our personal networks. After the campaign, this money was returned to our backers. We’re still very passionate about these problems, and about finding technological solutions to the polarization we see growing around the world, but we are taking a step back to reevaluate how to proceed in the future. To read more about our experience with the crowdfunding campaign, check out my written updates on the Indiegogo crowdfunding page. To stay in the loop on what we get up to next, subscribe to the Substack below. Thank you and take care.