Yes. There’s an interesting economic problem here: the paywall may be protecting the subscription business while simultaneously damaging the publication’s ability to acquire readers.
The irritating sequence is particularly counterproductive:
provocative headline → click → perhaps two paragraphs → Subscribe to continue
At that point the reader hasn't been given enough value to decide whether the publication is worth paying for. Instead, the publication has effectively used the reader's curiosity as a lead-generation mechanism and then told him the product is unavailable.
And, as you say, the internet supplies substitutes. If the story concerns something important, there will often be Reuters, AP, the BBC, a newspaper in the country concerned, a specialist publication, or somebody reporting the underlying facts elsewhere. Once a reader develops the habit of saying “Oh, that's a Times paywall; I'll find it somewhere else,” the paywall has trained the customer rather effectively—but in the wrong direction.
Advertising has its own problems, of course. But a relatively restrained model—ordinary display advertising, clearly identified sponsored material, perhaps some unobtrusive contextual advertising—has one enormous advantage: the reader gets the article. The publisher gets an opportunity to establish a relationship rather than putting a tollbooth between itself and the prospective reader.
I particularly agree with your qualification about simple advertising. The web's advertising industry largely went mad by turning ads into tracking systems, auctions, pop-ups, autoplay video, and enormous quantities of JavaScript. That created the demand for ad blockers in the first place. A newspaper saying, in effect, “We'll let you read our journalism for free, and in return we'll show you a few advertisements” is a much more intelligible bargain.
And there is an irony in the Economist and New York Times cases: they have unusually strong brands. They don't necessarily need the gimmickry of a clickbait site to attract attention. Their reputation itself is an asset. A headline that says what the article is actually about, followed by good journalism and modest advertising, might be a better long-term strategy than repeatedly irritating non-subscribers.
The difficult question is whether advertising revenue can replace enough subscription revenue to make that model work for expensive, labor-intensive journalism. That's where the economics get genuinely interesting. A publication employing hundreds of journalists cannot necessarily live on the advertising rates that supported newspapers in the old mass-market era. The web destroyed much of that advertising market by creating vastly more inventory and concentrating digital advertising revenue in enormous platforms.
So I wouldn't say “they ought to advertise instead of charging.” I'd say they have a strategic choice between maximizing revenue per reader and maximizing the number of readers who develop a relationship with them. The present model heavily favors the former.
And there's a third possibility that I suspect would suit someone like you rather well: a free basic article with a subscription for the deeper product—archives, specialized analysis, newsletters, puzzles, podcasts, saved articles, fewer ads, etc. That makes the paywall an upgrade rather than a locked door.
I think your second point is probably the more consequential one.
The economics of journalism may eventually push toward something rather different from today's newsroom. AI can reduce the cost of producing a first draft enormously, while experienced human editors preserve the things that matter most: judgment, verification, context, taste, and responsibility.
A reporter might give an AI a pile of documents, interviews, transcripts, public records, and observations. The AI could produce several possible structures, identify contradictions, summarize the evidence, suggest questions that haven't been answered, and produce a draft. A good editor then becomes something closer to an editorial intelligence director than a traditional copy editor.
That could actually make better journalism cheaper, rather than merely making mediocre journalism cheaper.
There is an important catch, though. AI doesn't eliminate the need for original reporting. If every publication simply feeds the same internet material into AI and asks for a story, you eventually get a hall of mirrors: increasingly polished prose derived from increasingly derivative information. Somebody still has to talk to the Cambodian farmer, attend the city council meeting, examine the court file, cultivate the source, or discover that everybody else has misunderstood something.
So I could imagine a future newsroom with:
- far fewer people doing mechanical writing and research;
- a relatively small number of journalists doing original reporting;
- AI doing transcription, document analysis, preliminary research, drafts and comparisons;
- highly experienced editors doing the crucial human filtering;
- and the publication distributing most of the resulting journalism freely, financed primarily by advertising and perhaps voluntary premium subscriptions.
That last part is where your two arguments reinforce each other. If AI makes producing each additional article much cheaper, the publication has less reason to erect a subscription wall around every additional reader. The marginal value of acquiring another reader becomes much more attractive.
There is also a delicious historical reversal here. Newspapers once had enormous audiences because distribution was cheap relative to the cost of gathering the news. The internet made distribution essentially free but demolished the old advertising model. AI could potentially attack the other side of the equation: making the production of high-quality journalism substantially cheaper without eliminating professional judgment.
I wouldn't bet that the NYT or Economist business models collapse—but I would bet that today's newsroom economics aren't the final form. The combination of abundant alternatives, reader resistance to paywalls, AI-assisted production, and increasingly sophisticated independent publishers puts considerable pressure on them.
Exactly. I think I overstated the "hall of mirrors" problem as though subscriptions somehow solved it. They don't.
If ten honest reporters attend the same press conference, they should produce ten accounts of essentially the same event. That's not a defect; it's the useful consequence of having independent observers. The hall-of-mirrors problem arises when the tenth reporter is simply rewriting the first nine without adding observation or verification.
And subscriptions have almost nothing to do with that.
In fact, subscriptions can arguably intensify the selection problem. A publication needs readers to keep subscribing, so it has an incentive to choose stories and headlines that its particular readership finds compelling. Two publications can therefore report the same day's events quite truthfully while giving radically different impressions of what mattered.
What subscriptions can buy is resources for doing something beyond the common story:
- send a reporter to the place rather than rewrite the wire report;
- spend six months investigating something;
- obtain and analyze a large body of records;
- maintain foreign correspondents;
- develop sources who won't talk to everybody;
- have an editor recognize that the conventional story doesn't make sense.
Those are genuine advantages, but they are advantages of spending money on journalism, not advantages of charging the reader for access to it.
And this gets back to your AI point rather nicely. If AI can cheaply handle the repetitive part—“Here are 47 reports of what happened; give me a coherent factual account and identify discrepancies”—then human resources can be concentrated precisely where journalism is valuable: finding out something that isn't already in the hall of mirrors.
There's an amusing distinction here: more reporters don't necessarily produce more information. More independent observation does. A thousand reporters copying the same press release are less useful than one reporter who discovers that the press release is wrong.
I think your second point is probably the more consequential one.
The economics of journalism may eventually push toward something rather different from today's newsroom. AI can reduce the cost of producing a first draft enormously, while experienced human editors preserve the things that matter most: judgment, verification, context, taste, and responsibility.
A reporter might give an AI a pile of documents, interviews, transcripts, public records, and observations. The AI could produce several possible structures, identify contradictions, summarize the evidence, suggest questions that haven't been answered, and produce a draft. A good editor then becomes something closer to an editorial intelligence director than a traditional copy editor.
That could actually make better journalism cheaper, rather than merely making mediocre journalism cheaper.
There is an important catch, though. AI doesn't eliminate the need for original reporting. If every publication simply feeds the same internet material into AI and asks for a story, you eventually get a hall of mirrors: increasingly polished prose derived from increasingly derivative information. Somebody still has to talk to the Cambodian farmer, attend the city council meeting, examine the court file, cultivate the source, or discover that everybody else has misunderstood something.
So I could imagine a future newsroom with:
-
far fewer people doing mechanical writing and research;
-
a relatively small number of journalists doing original reporting;
-
AI doing transcription, document analysis, preliminary research, drafts and comparisons;
-
highly experienced editors doing the crucial human filtering;
-
and the publication distributing most of the resulting journalism freely, financed primarily by advertising and perhaps voluntary premium subscriptions.
That last part is where your two arguments reinforce each other. If AI makes producing each additional article much cheaper, the publication has less reason to erect a subscription wall around every additional reader. The marginal value of acquiring another reader becomes much more attractive.
There is also a delicious historical reversal here. Newspapers once had enormous audiences because distribution was cheap relative to the cost of gathering the news. The internet made distribution essentially free but demolished the old advertising model. AI could potentially attack the other side of the equation: making the production of high-quality journalism substantially cheaper without eliminating professional judgment.
I wouldn't bet that the NYT or Economist business models collapse—but I would bet that today's newsroom economics aren't the final form. The combination of abundant alternatives, reader resistance to paywalls, AI-assisted production, and increasingly sophisticated independent publishers puts considerable pressure on them.
Exactly. I think I overstated the "hall of mirrors" problem as though subscriptions somehow solved it. They don't.
If ten honest reporters attend the same press conference, they should produce ten accounts of essentially the same event. That's not a defect; it's the useful consequence of having independent observers. The hall-of-mirrors problem arises when the tenth reporter is simply rewriting the first nine without adding observation or verification.
And subscriptions have almost nothing to do with that.
In fact, subscriptions can arguably intensify the selection problem. A publication needs readers to keep subscribing, so it has an incentive to choose stories and headlines that its particular readership finds compelling. Two publications can therefore report the same day's events quite truthfully while giving radically different impressions of what mattered.
What subscriptions can buy is resources for doing something beyond the common story:
-
send a reporter to the place rather than rewrite the wire report;
-
spend six months investigating something;
-
obtain and analyze a large body of records;
-
maintain foreign correspondents;
-
develop sources who won't talk to everybody;
-
have an editor recognize that the conventional story doesn't make sense.
Those are genuine advantages, but they are advantages of spending money on journalism, not advantages of charging the reader for access to it.
And this gets back to your AI point rather nicely. If AI can cheaply handle the repetitive part—“Here are 47 reports of what happened; give me a coherent factual account and identify discrepancies”—then human resources can be concentrated precisely where journalism is valuable: finding out something that isn't already in the hall of mirrors.
There's an amusing distinction here: more reporters don't necessarily produce more information. More independent observation does. A thousand reporters copying the same press release are less useful than one reporter who discovers that the press release is wrong.
That makes sense, and it's a habit many serious news readers have had long before the internet.
Reading three accounts of the same event often produces a richer understanding than reading one account once. Not because one is necessarily lying, but because each reporter notices different things:
-
One includes a revealing quote.
-
Another explains the historical background.
-
A third describes the personalities involved.
-
A fourth notices a statistic everyone else ignored.
The overlap isn't wasted. As you say, repetition aids memory. Teachers have known for centuries that encountering the same information in slightly different forms helps it stick.
I sometimes think there are two very different kinds of news consumers:
-
Information gatherers who want one concise summary and are done.
-
Information synthesizers who enjoy comparing sources and building their own picture.
You strike me as much closer to the second type. When you've read several versions, you're less dependent on any one editor's judgment about what mattered. You start to see the event itself rather than merely one publication's framing of it.
That's also why paywalls can be particularly annoying for people like you. If you're trying to compare five accounts of a story, hitting a paywall isn't just losing one article—it's losing one of the perspectives that helps complete the picture.
In a way, your approach resembles how historians work. A historian doesn't normally ask, "Which source is the truth?" He asks, "What do all these sources, taken together, tell me?" The differences between accounts are often as informative as the similarities.
And for someone who spent years in analytical work, that instinct is very familiar: don't trust a single report when you can examine several independent ones and look for the pattern that emerges.
No comments:
Post a Comment