There’s a lot of public discourse about media neutrality (or the lack of it). People keep telling journalists, “Just state the facts and spare the opinion.”
According to me, unbiased media is a myth.
As Hunter S. Thompson said 50 years ago, objective journalism is impossible barring a few exceptions like box scores for sports (to which I’ll add stock price, forex rates, and other quantitative data).
"With the possible exception of things like box scores, race results, and stock market tabulations, there is no such thing as Objective Journalism." – Hunter S. Thompson, 1973.
At last I found a great quote that reinforces my long-held belief that unbiased journalism is a myth.
— Ketharaman Swaminathan (@s_ketharaman) April 23, 2020
But media bias is not caused by the usual suspect “paid media”.
Media has always been supported by ads. Ad agencies are agents of media outlets and get ~15% commission from them for selling ad space, which is why they don’t charge any fees to their clients (i.e. advertisers).
When they learn about this revenue model of the media industry, the common man or woman (henceforth J6P) jumps to the conclusion that media will write favorably about whichever advertiser pays them. While I won’t deny that there are rags that print whatever advertisers want them to, reputed publications have a clear demarcation between the editorial side that decides what gets published and the business side that sells ads. Blatant “pay to play” is rare. (Media does print edit-looking content created by the advertiser but they’re marked specifically as SPONSORED CONTENT or such – they’re called advertorials or native ads.)
It’s easy to be cynical about the existence of the touted Chinese Wall between edit and business, but let me ask naysayers the following question: Media takes money from two advertisers who compete with each other to publish their ads, so which advertiser’s point of view will it shill? Neither? Both?? Either way, what gets published eventually will be two-sided aka unbiased.
So, the fact that media is paid does not cause the fundamental bias problem.
What does?
I attribute five reasons for media bias:
- Limited context
- Nuance doesn’t fly
- Contradictory expectation
- Subscription business model
- Vested interest
Let’s get on with it.
1. Limited context
The Economic Times article titled Flexi Office Cos Bet On Big Revenue Boost dated 20 November 2023 lists the top five players (by revenues) in coworking space in India as follows:
- WeWork India, 6.5M sqft.
- Table Space, 8M sqft.
- Awfis, 7.5M sqft.
- SmartWorks, 8M sqft.
- IndiQube, 7M sqft.
I’ve heard of all of these companies except Table Space, and I was surprised to see it occupy the second spot (by revenues).
But I was more surprised by the conspicuous absence of Regus, the largest flexible office space company in the world (Disclosure: My company’s office is in a Regus center).
At first blush, this is a major omission and adequate ammunition with which to allege that Economic Times is biased against the pioneer and the global market leader of the coworking space industry (and a few more conspiracy theories.)
But, if you scratch the surface, it might turn out that this omission is a canonical example of the limited context problem.
To write this piece, the reporter must have reached out to all leading coworking companies in India seeking information about their network and future plans. I’m sure his list included Regus – because no reporter can be so negligent as to miss out the OG of coworking. I speculate this is what happened subsequently.
On the back of WeWork Inc.’s bankruptcy in North America (USA and Canada), WeWork India has gone on an overdrive to assure its customers and the general public that it’s in good health and will continue to be in business.
@WeWorkIndia: We will continue to operate and serve our members, landlords, and partners as usual, as we are committed to the growth and success of our business.
Since WeWork is the bellwether of the industry, I’m sure the other coworking companies like Awfis and SmartWorks were also eager to put the word out and must have provided the details sought by the reporter.
Let’s come to Regus.
Regus has been in coworking business before WeWork was a glint in Adam Neumann’s eyes. It didn’t react to WeWork at its inception nor when WeWork was at its peak. Regus has probably forgotten more about coworking business than all its competitors know put together. Knowing how this business works as it does, Regus probably saw well ahead of time the train wreck that WeWork and its lead investor SoftBank were hurtling towards and let them get there without any prodding from its side. I’m guessing Regus India wasn’t interested in commenting about WeWork’s bankruptcy or didn’t feel the need to justify its robust business model. Also, Regus is a foreign company (IWG – International Workplace Group), and MNCs have very strict rules about which, if any, of their local employees can talk to the media.
Accordingly, Regus didn’t respond to the reporter. Ergo he skipped Regus from his league table.
J6P might expect the reporter to carry out more research to dig up Regus’ footprint and rank. I could not find any mention of square feet on Regus India website – it only says 108 offices. So the reporter couldn’t either.
Since PR is one of our company’s offerings, we work closely with journalists, and know that most reporters are trained to get to the bottom of most things. However, they might always not be able to put their training to work in actual practice.
Because more digging requires more time and money. In an ideal world, there would be an unlimited supply of both. But, in the real world, resources are always finite, and journalists would have to stop somewhere and file their story. So there’s a limit to how much context they can glean and how much narrative they can coherently weave around all the collected facts within tight deadlines. At least a print newspaper has a full day to work with whereas online journalism works on hourly deadlines. Ergo the common man’s expectation to get the “complete context” will never be fulfilled in the real world of scoops and 24 hour newscycle.
As Astral Codex Ten highlights in their article titled The Media Very Rarely Lies:
The media rarely lies explicitly and directly. When the media misinforms people, it does so by excluding context, or signal-boosting some events while ignoring others. Context matters.
The problem with context is that nobody will state upfront how much of it is necessary. In the absence of prior agreement on that crucial aspect of reporting, the reporter will need to make an intelligent guess about it, and hope that he or she has gathered enough context for the story.
The utmost I’d have expected from the aforementioned reporter is to have mentioned that he contacted Regus and did not get any response before article went to press, due to which he didn’t have enough data on Regus to know whether it belonged to the Top 5 or not, due to which he had to skip the #1 global coworking space brand from his league table.
2. Nuance doesn’t fly
Whenever I’ve shared my take with the media, I’ve covered both sides of the story. As you’d expect, I’d qualify many things. According to media professionals, readers don’t appreciate nuances, so they strip away all the qualifications, and what finally gets published comes across as one-sided.
This has happened to me several times during my decades of interactions with media. Let me give a recent example vis-a-vis my take in the American Banker article entitled Lessons from Canada’s Struggle With Faster Payments. The article was about A2A RTP rails in various countries such as RTR in Canada, FedNow and TCH-RTP in USA, FPS in UK and IMPS / UPI in India. It included a quote by someone about Open Banking and PSD2. I pointed out to the reporter that Open Banking and PSD2 have nothing to do with A2A RTP rails:
Your article is about RTR, which is a rail, not frontend app. Rod Reef’s remark about Open Banking / PSD2 is more relevant to frontend payments apps but irrelevant to the article’s context of a backend rail. UK FPS went live in May 2008. It was triggered by Office of Fair Trade (OFT) diktat in circa 2005 to cut down retail payments processing time from the then five days it took to process cheque payments. UK FPS had nothing to do with Reg PSD2, which probably didn’t even exist in 2005.
She reverted back thanking me for the clarification but regretted inability to publish it since, according to her editor, the difference between rail and app is too nuanced for the average reader to comprehend.
3. Contradictory expectation
On the one hand, readers say reporters should state facts and spare their opinion. On the other hand, they don’t like it when reporters do that.
Let me illustrate this contradiction with the following example: Suppose the reporter is writing about an incident involving two people, John Doe and Jane Doe. Suppose further that John said “It rained”, and Jane said “It didn’t rain”. A fact-only article would read “John Doe said it rained. Jane Doe said it didn’t rain”.
When they read that sentence, most readers would exhort the reporter to look out the window and see whether it’s raining or not. But, if the reporter wrote that it rained or that it didn’t rain, depending on their observation, they’d be sharing their opinion and not reporting facts, which is what John Doe and Jane Doe said.
Lately, the reader’s ask to skip opinion is not fulfilled even by scientific journals, which are traditionally associated with reporting hard facts and nothing else. According to the article entitled A wave of retractions is shaking physics, editors of respected scientific journals like Nature, Physical Review, and Science are pushing back on scientists’ expectations:
Scientists complain that journals tend to “focus on story”, which can lead editors to be biased toward experimental results that match theoretical predictions. But editors say they “feel a responsibility that science serves humanity, and if humanity can’t understand what’s in our journals, then we have a problem”. Ergo they feel they should not stop their focus on the story.
I tend to agree with the editors. Reporting just the facts can make a piece very drab.
After reading all this, if somebody still insists on “facts and nothing but the facts”, they should stick to Associated Press, Reuters, Tass, and other news feeds – they’re not the target audience for journalism.
4. Subscription business model
Media is traditionally supported by advertising and is free for readers. The publisher can publish articles from both sides in an attempt to be neutral. When she doesn’t pay for content, a reader who doesn’t like one article because it doesn’t reflect her worldview can skip that article and move on to another one that does.
However, when publishers adopt the subscription business model, they put content behind paywalls. When a reader pays a fixed monthly subscripton fee for content, every article she skips is a pinch on her pocket, and makes her think twice about renewing the subscription at the end of the month. By increasing the risk of churn, neutrality directly hits the topline of publishers. To mitigate this risk, subscription-fueled media companies need to self-select for the tastes of the majority of their subscriber base. Bias creeps in. Journalists become Activists.
Correlation is not causation but rants against media outlets like New York Times and Washington Post for being biased have skyrocketed in the last couple of years, which coincides with the period these publishers increasingly turned to the subscription business model.
If the aforementioned conspiracy theory is right, it raises two questions:
- Why can’t a media outlet publish one article with one worldview and sell it to people who hold that worldview and another article with the opposing worldview and sell it to people with that opposing worldview? (AI can help in that pursuit – mass personalization is anyway one of the more promising use cases of GenAI.) That way, the publisher will be neutral on an aggregate basis. Unfortunately, that can’t happen because it doesn’t make business sense for publishers to sell individual articles.
- Is there a strong correlation between fencesitters and non-subscribers? I have no personal opinion, anecdata or research to answer that question one way or the other but it’s a topic worth exploring.
5. Vested interest
Media is as much a business as any other and can have vested interests in the topic it covers. When that happens,
- Can it be unbiased?
- Will it be unbiased?
- Should it be unbiased?
Let’s take OpenAI’s negotiations with media companies for licensing their content to train its GPT large language model (LLM).
Jessica Lessin, Founder & Editor-in-Chief, of The Information writes in the 8 January 2024 edition of her newsletter Editor’s Pick:
OpenAI behemoth is in the process of negotiating deals with publishers to license their news articles for use in training its large language models. Our team revealed this week that the negotiations are surprisingly stingy: the company, which has projected it will generate more than $850 million in annualized revenue by the end of 2024, has offered some media firms as little as between $1 million and $5 million annually.
Since annual subscriptions for most of these publications cost in the region of $250 to $500, it can be argued that one to five million dollars is a ridiculously generous offer. However, as far as publishers are concerned, the more they can extract from OpenAI the merrier. To fulfill its MSV fiduciary duty to its shareholders, the media company must push for more, and cannot be unbiased. Which is exactly what The Information has done.
And it has worked: OpenAI has added one or two more zeros to its hitherto “stingy” offer:
Reddit, Axel Springer, and a few other publishers have reported data deals with OpenAI, Google and a string of other Gen AI companies. For example, Reddit stands to earn $200 million a year by letting LLMs train on its data – Ten Revolutionary Things About AI – Part 1
It’s clear that media, in its present shape and form, cannot be unbiased.
However, that doesn’t mean there are no other models where media can be less biased or at least make the direction of its bias evident.
In a follow up post, I’ll explore a few such models. Stay tuned!
