Searching for insight

No, really, I am trying to find it, and am offering bribes.

“Every block of stone has a statue inside it and it is the task of the sculptor to discover it.” –Michelangelo, allegedly, but kinda seems fake.

Fifteen years ago, I worked at a think tank in Washington, DC. I spent most of my time writing a weekly newsletter about the global economy, and each article followed the same audacious formula: a few charts, 1,400 words, and a couple concluding recommendations for how to solve some major worldwide crisis.¹

In theory, our work was affecting real change—we were talking about meaningful problems, not trying to get people to click more ads. In reality, the newsletter, which went out to a few thousand Hill staffers, government agency officials, other think tank mill hands, and my parents, only affected that change through a long series of highly dubious dependencies:

1. some of the people who got our emails might read what we wrote;

2. some of those people might find it compelling;

3. they might then include it in their own research briefs and office newsletters;²

4. that email might get read by a chief of staff or department director;

5. it might influence their thinking about some issue;

6. they might bring up that issue to an elected official or policy maker;

7. that person might then develop a passing awareness of our idea;

8. that idea might imperceptibly alter the complex political calculus behind some decision that they make or bill that they write;

9. that decision might trickle out of Washington, into the global economy, into someone’s pocket, and affect real change.

If nothing we wrote ever made it past step three—and I doubt anything did—was our work a waste of money? On one hand, yes, obviously. Our job was to help policy makers make better decisions, and we couldn’t do that if the most we ever did was fill some unread “Daily Pulse” brief, the unplugged controller that a mid-level State Department staffer gave to a nepotistic intern to keep them busy for a summer.

On the other hand, the articles, and the hundreds just like them that dozens of other think tanks dumped out every day, served a greater good. They were the set behind the real actors. Our work—the incessant churn of “expert” commentary and analysis—was necessary to backstop the influential stars who carry every think tank’s brand: The former Fed chairman, the presidential candidate, the senior global policy maker, the Nobel prize winner. Those people skip all the steps that doomed our newsletter, and instead just have dinner with senators and White House advisors. But they need job titles to maintain their positions of influence; no CNN producer wants their chyron to introduce Anderson Cooper’s next guest as “unemployed.” So they work at think tanks, the think tanks buoy their legitimacy with dozens of newsletters and research papers and afternoon policy luncheons, and they give think tanks their famous headliners.

So did our work matter? Sort of. It had to be a convincing imitation. We were background characters, costumes, b-roll. The theatrics of the stars wouldn’t work without us, but nobody was really paying attention to us either.

The world is probably full of jobs like this. They aren’t useless, because a movie needs props. But they aren’t the things that they pretend to be. They are the bricks in a false front: Collectively convincing from a distance, but bearing little meaningful weight of their own.

Lord Business

There are other examples. Almost exactly a year ago, Futurism accused Sports Illustrated, a former premier print magazine and now an online brand that hot potatoes itself between minor media conglomerates and second-tier PE firms,³ of publishing AI-generated articles under AI-generated bylines. Sports Illustrated’s publisher, The Arena Group, gave a lukewarm defense: They said that the articles were written by a third-party contractor called AdVon Commerce, and that AdVon “assured” them that the articles were written by humans, but some “writers use a pen or pseudo name in certain articles to protect author privacy.” Notably, neither Sports Illustrated nor The Arena Group defended the content of the articles, which were, you know, weird:

Play Like A Pro With The Best Full-Size Volleyballs
By Drew Ortiz

Volleyball is one of the most popular sports in the world, and for good reason. It’s fast-paced, has a high skill ceiling, and is generally an exciting sport to both play and watch. Even people who don’t watch sports can easily understand the intensity and skill required to play volleyball whenever they watch clips. There’s a reason why it’s been such a mainstay in modern sports to this day. Volleyball can be a little tricky to get into, especially without an actual ball to practice with. You’ll have to drill in the fundamentals in your head before you can really play the game the way it was meant to be played, and for that, you’ll need a dedicated space to practice and a full-sized volleyball. However, finding a good volleyball can be tough. Luckily for you, we’ve found some of the best full-sized volleyballs in 2022 that are perfect for practicing your serves and sets every day.

One way to read this sort of scandal, which has been repeated by other publications, is that the publishers are short-sighted Lords of Business who rather glue together cheap prefab SEO chum than pay real writers for their creativity and culture. Which, yes, clearly; you do not outsource supposed journalism to a company called AdVon Commerce unless you only care about advertisements and commerce.

But another way to read the scandal—and in particular, The Arena Group’s unwillingness to even bother defending its unsettling sales pitches for “full-sized volleyballs”—is that the artists are overvaluing their art. Surely, Sports Illustrated knew articles like these are terrible; their bet, it seems, is that it doesn’t matter if they are. Beat writing and daily commentary about locker room drama is, like the articles I wrote for the think tank, theater; it is the artificial hustle and bustle necessary to make a sports website feel alive enough that its central stars—a few prime time talking heads and influential podcasters—appear to be in the middle of something popular. The leads’ art matters, and the rest is just set design: a façade that only needs to be peripherally believable.⁴

For The Arena Group and publishers like it, the point may not be to use AI to replace journalists, nor is AI up to the ideal version of the job that journalists are supposed to do; the point is that the journalists’ work doesn’t matter to their business all that much anyway. All the publisher needs is someone to churn out superficially useful busywork and SEO bait. Historically, you couldn’t create that without paying people to do proper journalism and to create interesting art; now, publishers might be able to buy the former without being burdening themselves with the latter.

In other words, if AI replaces all our jobs, it might not because because what we do is easy, or because AI is so good that it can do as well as we do. It may be because it never mattered if we were particularly good at it in the first place.

“Previously unimagined insight”

One of the themes of this blog is that a lot of the analytics industry may be unwittingly playing the same part as Sports Illustrated’s beat writers or as the think tank content farm. We are a stage behind the main show—and in our case, we are a piece of the edifice of executive competence. We give certain CEOs a way to convince people of something they want to do anyway; we give uncertain CEOs a way to say a bad decision isn’t their fault. We give companies a way to market themselves as having some winning competitive edge. But a lot of that work is spectacle—the data isn’t that valuable; the results are tilted; the decisions are emotional; the job is a fad. AI may replace us, not because a bot can do real analysis, but because we never could do real analysis either, and it never mattered all that much if we did.

There are exceptions, of course; some companies are quite good at analysis and building operational engines on top of data organizations. But those are not the norm. Instead, the core narrative that motivates our existence—give us data, and we will find you insight—is storycraft, written by us, and perpetuated by the executives and software vendors who have made being oh-so-data-driven a central part of their own brand. And perhaps it is time to get rid of the whole thing.

But it is Thanksgiving, and today seems like a rude time to write yet another one of those rants. Maybe my cynicism about our usefulness is a projection. Maybe analysis isn’t all that hard; maybe it’s just hard for me. Maybe there are thousands of examples of analytics delivering on its promise of finding previously unimagined insights. Maybe I am the exception, and those stories are, in fact, the norm.

What truth is out there?

A few months after ThoughtSpot bought Mode, ThoughtSpot’s marketing team proposed that I start a podcast. They had been running a successful one for a few years, and, after I joined the company, suddenly found themselves needing to babysit a new employee who, once again, had a job and no useful skills. In 2013, those who can’t, start a blog; in 2023, those who can’t, start a podcast.

The early idea—which never materialized; oops⁵—was an extension of the only data podcast I’ve ever really wanted: One in which people tell stories about the interesting things they learned about how the world works while working with data. Most data podcasts or interview programs are centered around their guests' resumes. People are invited on because they have big titles at important companies; they’re asked about their careers and accomplishments; they are mined for actionable advice and hot takes.

I wanted (want?) a show that talks about interesting problems. A lot of data people are drawn to the industry because they started looking at the numbers behind some hobby or interest—sports, food, politics, fashion, movies, science, whatever. It was the domain that drew them in, not the career path. But once we’re here, we end up talking about the axes that we use, rather than the gold that we were initially eager to find.⁶

So, in the spirit of that podcast, in the spirit of general holiday cheer, gratitude, and hearts growing three sizes some days, and, most importantly, in the spirit of giving this whole analytics and insights thing one more real shot, I would love to know, from you: Is analytics real? Are there examples of times when the narrative of what an analyst is “supposed” to do—ask a question, analyze data, poke and prod and slice and dice and drill, until something profound emerges from the noise, when the angel is set free from the stone. I’m less interested in stories about generic impact—”we ran some tests and boosted signup rates by four percent!”—or of operational refinement—”we used to call people at random and now we score them with red, yellow, and green dots!”—and more interested in true discoveries—”we used to evaluate players based on their batting average, and we built a championship team by using on-base percentage instead.”

We sell this promise all the time. We dream of it as our final destination, once we’re done with building operational infrastructure, cleaning up messy data, and democratizing data exploration. It is our original inspiration and our greatest value. And I’m not convinced it’s real.

But hey, maybe I’m wrong. If I am, please tell me.⁷ For the greater good.

1. As opposed to the current weekly newsletter, which is twice as long, has no charts, and prefers to induce minor crises.

2. Never has a company had an office communication problem—”how can we keep track of everything everyone is working on??!”—and not had someone try to solve it with an internal newsletter, and never has an internal newsletter solved an office communication problem.

3. This is the history of Sports Illustrated’s ownership, I think: Before 2018, it was owned by Time, Inc. When Time sold to Meredith Corporation, a media conglomerate, Sports Illustrated was sold to Authentic Brands Group, a brand management company that was owned by CVC Capital Partners, a private equity firm, and HPS Investment Partners, a private debt investor. The Sports Illustrated brand was licensed to The Arena Group, a brand holding company that is somehow public, despite making forty-five thousand dollars in revenue last quarter. After the Arena Group missed a licensing payment earlier this year—which, uh, yeah—Sports Illustrated was sold to Minute Media, another media brand holding company, which is owned by a former venture capitalist and private equity investor. Fun.

4. Look, I’m not saying I want this to be true. It’s depressing to imagine that half of ESPN’s writers are employed to be extras on Stephen A. Smith’s stage. But it also seems somewhat plausible that half of ESPN’s writers are exactly that.

5. My (rejected, reasonably) proposal was to call the podcast “Vesting Out,” and make the punchline that “I’m only here so I won’t get fired.” It was an ok joke, meant to make fun of the strained enthusiasm that everyone has for every post-acquisition project—”We could not be more excited for this next chapter!” But it would’ve been a very funny joke, by ThoughtSpot, if they had told me to create it, knowing that they were going to get rid of me a month later. I’m not sure I can ever say I’d be proud to start a podcast, but I might’ve been proud to start a podcast called “I’m only here so I won’t get fired” that got cancelled because I got fired.

6. Of course, this stuff is hard to talk about, because it’s not often public information. I’m not sure how you deal with that, though it seems like there would surely be a way. If nothing else, things stay interesting a lot longer than they need to stay secret.

7. The best response wins a pair of L.L.Bean slippers! What is the best one? I don’t know. The most interesting? The most surprising? The funniest? The most insightful? I will get back to all of you with an answer, and one of you with a pair of L.L.Bean slippers.