Every year, I write one navel-gazing post about what I do for a living. Past instalments include “Noah Smith’s writing advice”, “How to write a successful Substack”, “How to be an opinion writer”, and “The biggest advantage of op-ed writers”. I expect that the next post will be about how to incorporate AI into your writing, since new “reasoning models” are getting good enough to fundamentally change what writers do. (In fact, I may end up writing that post in less than a year.)

But today’s instalment is about why Substack is starting to cannibalise the legacy print media, and how the newspapers can respond.

Today’s post is inspired by a recent chat I had with Paul Krugman. Paul, who just left the New York Times and started blogging full-time at his Substack, explained why he abandoned the country’s most prestigious newspaper:

“Columns are a very restrictive form. You get about 800 words, and you have to write for an audience that’s coming at you with no prior knowledge. At this point, there are now three levels of editing at the Times. Before it even goes to the copy editor, there’s an editor who makes much more substantial edits than ever before. There’s a lot more editing, and then a layer of editing above that, which became very intrusive in the last year. So, the column became quite an unpleasant experience. And I said, ‘Well, I need some way to have longer-form thoughts and an occasional chart.’ So, I had a newsletter at the Times, which was far more constrained than blogging, but it kind of kept me going. And then, in September of last year, they eliminated the newsletter. So, you would hate the job I had this past year. I came to hate it as well. You know, maybe what they were doing made sense in terms of the Times’ business objectives. I don’t know. But it was intolerable from my point of view.”

This really reminded of my post from last year, about why Substack writers have an advantage over op-ed writers who work for major publications. In that post, I argued that editors subtract more value than they add to the op-ed writing process, by preventing op-ed writers from developing their own distinctive voice, their own loyal audience, and their own organisational skills.

Paul brings up another important constraint that editors place on op-ed writers — they add cumbersome time and effort to the process of writing articles. One of the reasons I left Bloomberg Opinion was that they added an extra layer to the approval process for each article, which greatly increased the time between when I would pitch an article and when the article would get published. This cumbersome process was intended to make sure that only relevant op-eds got published, but it ended up having the exact opposite effect — by the time I or any other Bloomberg Opinion writer was able to publish a post about some current event, the topic had usually already gotten stale, and the discussion had moved on.

But in fact, I think the problem with how legacy publications handle op-eds goes much deeper than the editorial process. Thinking about it more, I’m now dissatisfied with the post I wrote last year — it’s not just ‘editors vs. no editors.’ The problem is about how legacy publications think about ‘op-ed writing’ in general.

Opinion vs. analysis

In the world of newspapers — and of online publications like Bloomberg that evolved from newspapers, and which hire many of their employees from newspapers — the typical distinction is between fact and opinion. This is a conceptual distinction that every child learns about in school — facts are things you can prove with evidence, while opinions are value judgements that aren’t necessarily based on facts. “Taylor Swift is a popular singer” is a fact, while “Taylor Swift is the world’s best pop star” is an opinion.

At newspapers or online legacy publications, reporters are supposed to give you the facts, while op-ed columnists give you their opinions about the facts. In fact, as Wikipedia notes, the New York Times created this system in the 1970s, building on earlier forms of writing:

The “Page Op.”, created in 1921 by Herbert Bayard Swope of The New York Evening World, is a possible precursor to the modern op-ed, Swope explained:

“It occurred to me that nothing is more interesting than opinion when opinion is interesting…and thereon I decided to print opinions, ignoring facts.”

The modern op-ed page was formally developed in 1970 under the direction of The New York Times editor John B. Oakes. The newspaper’s executives developed a place for outside contributors.

Newspapers draw a stark distinction between news and opinion. These are typically managed separately, by different sets of editors (called “desks”), and writers that do one kind of writing rarely do the other kind. At Bloomberg, news and opinion are two different divisions of the company.

But the hoary distinction between fact and opinion, enshrined in the organisation of legacy publications, leaves out a crucial third kind of writing that’s in very high demand: analysis.

Consider a forecast about the future, such as “Democrats will win the 2026 midterm elections.” This isn’t a fact, because it can’t be proven or disproven with currently available data. And although people might colloquially call it an “opinion”, it’s not a subjective value judgement either — it’s based on facts, even though it’s not a fact itself. A forecast is a third kind of thing.

Another example of a third thing is an assessment — a more complex type of prediction. For example, consider the statement “Russia now has the upper hand in the Ukraine war.” That’s not a fact, because “upper hand” isn’t precisely defined — there’s no universally agreed-upon measure of who has the upper hand in a war, and reasonable people can disagree as to the proper measure. But it’s not an opinion either, because there’s no value judgement involved.

Other examples are theories (e.g. “Wars happen because of scarce economic resources”) and recommendations (e.g. “If you want to win a war, you should secure your supply chains”). These are based on facts, but they themselves are not facts.

Forecasts, assessments, and theories are all part of a category called analysis. Some people casually refer to analysis as “opinion” — if you say “Democrats will win the midterms”, they might say “Well that’s just, like, your opinion, man!”. But discerning people recognise that there is a salient, useful distinction between value judgements and reasoned interpretations of facts. The way people argue about the two things is fundamentally different — to argue about opinions, you typically have to make an emotional appeal, while arguing about analysis can be done using logic and reason alone.

There are plenty of analysts out there in the world, for whom providing analysis — forecasts, assessments, theories, and recommendations — is their primary job. There are stock analysts, business analysts, intelligence analysts, and many more. They all have opinions, and these often motivate them personally and influence their work, but producing opinions is not their primary job — the more dispassionate and objective they are, the better of a job they’ll do.

Now here’s the thing — news readers want analysis. They don’t want to just read the reporting from the Associated Press or some investigative journalists and then have to put it all together and decide what it all means. They want writers to explain those facts, to identify the salient implications, to make predictions, and to recommend courses of action.

Analysis is most of what I do at this blog. As I always say, I think of myself as basically an open-source version of a CIA analyst. Instead of agents in the field, I gather my facts from reporters, research papers, and other publicly available sources on the internet. I then analyse those facts and try to synthesise them into a short, comprehensible story. Instead of writing my reports for my bosses higher up in an organisation, I release them to you, the public.

There’s a huge demand for what I do, and that demand is only increasing. As people’s information diet has become bigger and more complex, they’re in desperate need of people to make sense of it all. Some people come to my blog to hear me rail against totalitarianism or extol the virtues of progress, but I think most of my audience just wants me to explain what’s going on and tell them what they can do about it. They usually resonate with my opinions, but what they really want is my analysis.

Should opinion sections be about opinions?

Currently, at legacy publications like the New York Times and Bloomberg, the newsrooms do a bit of that analysis. Look at any news article, and it won’t simply be a list of quotes and numbers and events — there will be some amount of explanation there. And in recent years, news teams have begun to do longer, more complex features that blend more analysis into their reporting. But most of their analysis comes from their “opinion” sections. Generally speaking, reporters still just try to report the facts; forecasts, assessments, theories, and recommendations are the domain of op-eds.

And yet in my experience at Bloomberg — and from what I see at the New York Times and other legacy publications — the editors who run opinion desks still think the primary job of their writers is to dispense opinions. At Bloomberg I was always pushed to make my writing more subjective and politicised; I had to have a “take”, to distinguish myself from reporters. If you read the New York Times’ stable of regular opinion columnists — Maureen Dowd, Nicholas Kristof, David Brooks, Jamelle Bouie, Thomas Friedman, and so on — their weekly columns are chiefly polemics with a bit of analysis thrown in.

This ends up holding those writers back. It certainly held me back when I was writing for Bloomberg. And more importantly, the focus on the opinion section as being fundamentally about opinion rather than analysis ends up holding legacy publications back and opening the way for platforms like Substack to steal their thunder.

There are several reasons for this. First, opinion becomes tiresome fairly quickly. The modern world of social media is chock full of political shouters, aggressively denouncing each other and shoving every possible ideology into your face. Sure, it’s nice to be able to read some writers who do that more eloquently than the average X poster or TikTok video, but ultimately it just adds to the maelstrom. Right now, when the country is exhaustedly struggling to put a decade of social unrest behind it, the constant flood of opinion seems even more intrusive; dispassionate analysis provides something of a refuge.

Another reason is that readers are inherently distrustful of opinion. A polemicist is selling you something and usually is trying to recruit you for an ideological team; that bias makes them a little bit suspicious as a source of ideas.

And opinion is in some sense cheap — after all, everyone has an opinion. In my experience, people at legacy publications tend to look at op-ed writers with a mix of envy and contempt — envy because it’s a much more glamorous job than reporting, and contempt because there’s a sense that the glamour is undeserved. After all, what makes Tom Friedman’s opinions, or Maureen Dowd’s, more valuable than those of the average reporter?

Within legacy publications, there’s a tendency to treat op-ed writers as tawdry entertainers rather than as respected analysts. Inevitably, they are marketed as such and encouraged to fill that role. I always felt that Bloomberg was pushing me to be a rhetorical gladiator rather than a wonk; the NYT certainly pushed Krugman in that direction, especially in more recent years. This didn’t do either of us any favours.

The constraints of the legacy op-ed format

But the biggest drawback of the “opinion” concept is that it limits the format that writers can use for analysis. Essentially, editors at legacy publications think of op-eds as just news articles where the writer says what they think instead of quoting sources. In fact, the editors who work at opinion desks tend to have gotten their start as reporters or as editors at the news desk and typically try to port their editorial approach over from one desk to the other.

As Krugman pointed out, his op-eds were limited to somewhere around 800 words. That was the length my editors at Bloomberg also wanted me to aim for in every column; if I wanted to write 2500 words, I had to get special permission to “go long”, which took days of editorial wrangling to secure. Now, as an independent Substacker, I can write as much as I want — my previous post was 3200 words long, and the one before that was 2600. Do most people read all of that? No! But some do, and more importantly, different people read different sections, zeroing in on their area of interest.

The longer format allows for a greater diversity of information presentation. I tended to put in a lot of charts at Bloomberg, but Krugman was barred from using any — a horrible and pointless constraint for an economics teacher. I can post open-source images and infographics, figures and tables from academic papers, long block-quotes, diagrams I make myself, or even equations. I can also take long side-tracks in the middle of a post, or in the footnotes, to explain concepts for people who don’t have the relevant background. I just couldn’t do that at Bloomberg.

The longer, richer format of Substack is perfect for analysis. You usually don’t need thousands of words to express an opinion, which is one reason why legacy publications think it’s OK to limit op-ed writers to a few hundred. But good analysis often has to be long in order to be thorough, which is why research reports tend to run into the dozens of pages. And a more varied presentation format tends to facilitate explanation, which is why teachers use blackboards and visual aids. Legacy publications, meanwhile, stunt their analysis by forcing it into a short bland format optimized for conveying opinions.

Now, to be fair, in recent years there has been some improvement! The NYT, Bloomberg, the Washington Post, the WSJ, and others have focused more on “data journalism”, and have introduced longer features where some of their op-ed writers are free to do in-depth analysis. But this is still the exception rather than the rule, and while these features tend to have beautiful graphics, their analysis is often pretty hit or miss, simply because they’re not used to the format.

The short format is also related to yet another problem with the “opinion” focus — the tendency to hire too many op-ed writers. Legacy publications try to get a diversity of opinions and perspectives for their op-ed pages, which in practice means hiring tons of columnists and soliciting tons of guest contributions. That means that each op-ed writer only gets to have a small space in the publication, forcing them to keep their analysis short and shallow.

If you think about it, the business model of Substack shouldn’t work. Why would people pay $10 a month for a single writer, when they could pay $25 a month for the entire New York Times? The answer isn’t that writers like me are just much better than all the people at the New York Times or Bloomberg¹ — it’s that a lot of people want to read a few in-depth analyses instead of a ton of short punchy polemics.

Essentially, legacy publications are constrained by an ingrained culture, that treats op-ed writing as a weird, flashy, slightly disreputable adjunct to their main business of news reporting, instead of a valuable layer of analysis that sits on top of that reporting. There are lots of people doing great work at the op-ed sections of the New York Times and Bloomberg and so on, but somehow, whenever we go independent, we suddenly get much better.

Legacy publications should learn a lesson from the success of Substack and unleash their op-ed writers to do more analysis. I don’t think this will be enough by itself to arrest the decline of traditional print media, but I’m sure it would help.