We all have an audience
To talk about what we say, we have to talk about who we say it for.
This Substack has about three dozen editors. Not literally, of course—the only true editor is Google autocomplete—but every post is guided by the anticipated reactions of a handful of real people who occasionally read this blog. They are the influencers in the industry, the people who run companies or control capital, and the people who have audiences far bigger than mine. I don’t write for them, but it’s difficult to get their imagined voices out of my head when I choose how decisively to take a stand or how far to go in some bit of criticism or praise.
I didn’t start this way. In this blog’s early days, I wandered through different topics, ranging from gossip on Clubhouse, complaints about Paul Graham, charts of Twitter profile pictures and baseball seasons without bats, and whatever this was.1 Over time, however, I saw what got reactions and likes—posts about data tools—and what vanished into the abyss of the internet—just about everything else. I liked the attention, and my editorial calendar and I chased after it. What started as a blog for a relatively general audience became, in fairly short order, a clubby publication about itself.
Moreover, as I saw the names of those who signed up or commented, I found myself overweighting the reactions of the people I either respected or who, cynically, “mattered.” More than wanting everyone to like what I had to say, I wanted them to like what I had to say. If a hundred people hate what I write but Andrew Gelman calls it sharp, Gia Tolentino says it’s well written, or Michael Lewis applauds the story, I’ll be proud of it. And if everyone loves it, but Michelle Obama tells me she’s disappointed in me, I’m shutting this whole thing down.2
Admitting this goes against the ethos of a Free Thinker—a label I wouldn’t have applied to myself before, and something I certainly can’t claim now, not after torpedoing the illusion that I somehow operate above the influence of the rich, the powerful, and the popular. But like anyone, I’m corrupted by the same basic motivations that plague middle schoolers and former presidents alike: I want the people I respect to respect me. To paraphrase David Foster Wallace’s oft-cited graduation speech, we all worship something, even those of us whose entire brand is our supposed unbiased, data-driven atheism.3 For people who talk on the internet, our religion is the opinions of other people who talk on the internet—particularly those who are more successful, who make more money, who are more widely followed, or who simply talk and write better than we do.
Once you get a bit of attention, this insular dynamic feeds off itself. You cater to the influencers’ club; you’re invited further in; your views become more aligned with theirs, either intentionally or because of the gradual conjoining of contexts and desires. Even for me—a gnat compared to people with real reach—I’ve noticed how easily you become the center of your own conversational universe. People ask you questions about what you think; they engage with what you write; the opinions you read are increasingly reactions to you. Your window into the world used to be your feed; eventually, your window becomes your notifications.
A brand, not an antidote
As professional data people, we often act as though we're above this impulse. Our day jobs, analyzing data to find The Truth, compel us to position ourselves outside of irrational religions and emotional desires. For many of us, this posture is both a professional brand and personal identity: We have conquered our feelings and balanced our scales with nothing but math and reason, and have the Deming quotes in our slide decks to prove it.
But we haven’t. We haven’t done any of these things. We’re just humans—as emotional, corruptible, and in need of therapy as much as anyone else. We respond to the roar of an online crowd,4 the praise from our bosses, and the disappointed looks of our peers just as everyone else does. The type of affection we’re chasing may be different—we want people to think we’re smart; we want people to believe us to be coolly detached and unshakably empirical—but we’re still chasing it, all the same.
As Tristan wrote a few months ago, this affects how we operate within our companies. An analyst, he says, is “not actually a disinterested observer. You are compromised by your very existence as a self-interested human on the team, with stock options, a track on a career ladder, and human relationships you care about. You are deeply, deeply compromised.”
This, however, doesn’t stop at our employer’s door. It can also infect us en masse.
Inside the Beltway
Data conferences used to be staid affairs. Conference headliners, scattered among technical discussions and command line demos, were typically drawn from adjacent industries. They served, indirectly at least, as reminders of who we worked for, and what the point of our work was.
As our industry has grown—as it’s created enormous companies, minted a number of billionaires, elevated community celebrities, and stoked an entire ecosystem of well-respected and well-funded venture capitalists—we no longer need this external validation. An entire conference lineup can be populated with homegrown influencers and entrepreneurs, without any need to look beyond our own rolodexes.
Last week in Austin, at Data Council, the first major in-person data conference in two years, this is exactly what happened. The data community—and just the data community—came together for two days of conversations about our tools, our jobs, and the companies we founded or funded. Those from adjacent industries were nowhere to be found.
The critics were swift. Rather than gathering to talk about business problems and the “end users” of our work, those of us in attendance, who were mostly, it seemed, vendors and venture capitalists, talked about ourselves. We lost ourselves in our own navels, delighted in the same indulgent debates that we have on Substack and Twitter, and ignored the questions about business impact that we’re all ostensibly paid to answer. Relative to the prior versions of Data Council, which promoted technical talks and stories from the trenches of the tech industry’s most advanced users of data, this year’s edition was, as Drew put it, very different.
Though I agree with Drew that there’s nothing wrong with some conferences evolving into trade shows built around product demos and branded happy hours,5 the thrust of the criticism, which extends well beyond Austin and into the mash of Substacks and Twitter conversations that dominate today’s discourse, is fair. We are more inward-focused; we do talk a lot about insider industry news; we do write about the stacks we’re building and not what we’re using them for. Speaking as someone who’s contributed their fair share to this corpus, it’s worth asking the uncomfortable question: How did we lose our way?
The answer, I think, is simple. We stopped talking to other people. The success of the industry (or at least, the rush of venture capital into it) paved paths to success that don’t require engaging with end users or business problems. We can build companies, careers, and reputations on each other alone. Just as this blog tumbled down a self-referential rabbit hole, so too did the community.
The solution, then, is both easy and incredibly difficult: We have to bring other people back into our conversations. The challenge, though, is we can’t just evoke their needs or remind ourselves to think about them—we have to make them our audience. We have to write for them, and give talks to them. We have read their blogs, reference their tweets, and chase their affection. Without that, the same self-interested demons that lurk in this blog and in our daily decisions at work—we want to be respected, and we inevitably cater what we say to those whose opinions we care about—will keep us focused on ourselves, and not those who we’re supposed to be helping.
There’s a useful parallel between this and political journalism in the United States. Every election cycle, political writers file countless stories about the opinions of “real Americans” eating at diners in Kansas. Putting aside the obvious point that people who live in cities are just as real as those who live in Topeka,6 these stories are mostly a charade. Axios writers don’t write about Nebraska because they care about Nebraskans; they write about Nebraska because they think other journalists will view it as astute. In this way, dispatches from Denny’s are no different than Beltway gossip stories: They are judged, not by how Nebraskans view them, but by how elite readers view them.
So long as we only talk to one another, conversations about the “end user” will come to serve the same purpose. We’ll throw around the phrase to grandstand to each other how wise we are. But the imaginary editors on our shoulders will remain the same; we’ll write to impress the same readers; we’ll tweet to get retweets from the same people. End users will become our slang de jour—another way to celebrate our savvy and show off for our friends—but nothing else will change.
Changing this is easier said than done. Stepping out of our comfortable community is difficult. And speaking for myself, my drafts folder is still full of titles like “Model-view-controller design patterns in the modern data stack.” I'm also, like most people, still a teenager at heart, wanting to be liked by the cool kids. But, to add George’s, Juan’s, and Sarah’s voices to the chorus of critics in my head, I think it’s worth all of us remembering that there are lots of cool kids—and, of course, “business value”—beyond the halls of the data industry.
April 4 – After posting this on Friday, there's been a swirl of conversation about tools, conferences, and vendor content, and their role in community discourse. Though I’ve struggled a bit to follow the thread of the discussion (and, this being the internet, threads are always frayed anyway), the consensus seems to be settling on three points: Vendor content is undesirable; we talk too much about tools; and we should talk more about practical stuff, like the problems we’re solving and how we’re building the teams that solve them.
All three of these points could potentially be extrapolated from this post as well. That’s not quite what I meant here; if it was read that way, consider the addendum below and extended correction and clarification.
First, on the role of vendors in the community: I don’t have any objection to vendors participating or making content; that’s how businesses do business these days. I’ll gladly let LeBron and Leo sell me tequila on Instagram if that keeps the internet free, and I’ll gladly read content hosted on vendors’ blogs or watch talks at vendors’ conferences if it saves us from courtesy calls and being visited by traveling salespeople.
That said, I would make a distinction between vendor participation—which is usually transparent in its motivations—and community astroturfing, which is more common among VCs and angel investors. If you're an investor and you believe in an idea or market, say that you believe in that idea and you're proudly backing such-and-such company to make it a reality. Instead, some investors have a habit of parroting the talking points of their portfolio companies, but not disclosing exactly who they’ve invested in. (You see this a lot in crypto; investors will talk about how they believe there’s a big need for decentralized financial services for the middle class in southeast Asia without explicitly saying they’ve invested in a company that provides decentralized financial services for the middle class in southeast Asia.) Generously, I think investors want to be seen as serious thinkers, and they don't want people to assume their intellectual viewpoints are biased. Cynically, they do it because the best ads are the ones that don’t disclose that they’re ads. Either way, this type of content, unlike run-of-the-mill vendor content, does have a corrosive effect on the health of a community.
Second, on conversations about tools: I don’t have any objections to these conversations either. Tools are useful, and talking about how to make them better is useful too. Many of us in this industry have devoted significant chunks of our lives to building tools; if we didn’t think they were worth a few blog posts and tweets on an infinitely scrollable internet, we’d have some much bigger decisions to question.
Moreover, even if talking about tools isn’t useful, that’s ok too! When we talk about community content, there’s often an implied assumption that that content has to have some practical application. There should be lessons to be learned, key takeaways to take away, slides to take pictures of and bring back to your team.
Speaking for myself, that’s not why I write this blog or give talks. Bluntly, I don't have that much to teach anyone, and I'm lousy at teaching the things that I do know. I talk about what I talk about because I’m interested in it, and I enjoy talking with other people who are interested in it too. In that way, this blog is, much like data dad jokes on Twitter, entertainment. Nerdy, narrow, niche entertainment for the most boring people in the world, but entertainment nonetheless. I read books like Dataclysm and Everybody Lies because I think they’re fun; I’d read blog posts about the COGS of green onions, not to learn anything practical, but because it’s fascinating. My aspirations here are to provide the same. If people come away from what I write asking themselves interesting questions, or reflecting on something in a different way, great. But most of all, I hope they just enjoy the few minutes they spend reading it.
Which, I suppose, raises the obvious question: What was the original post about, then? Where, if not in talking about tools, did “we lose our way?”
The point I’d hoped to make was that it’s not the content of what we talk about that matters, but who we make that content for. Talking about tools is fine if we recognize that, in most cases, it’s an entertaining indulgence for us, as industry people, to have low-stakes debates about. The problem arises when we make those debates more meaningful than they are by losing sight of how small the room is that’s having them. That room has definitely gotten bigger—and richer—over the last few years, but compared to the corporate environments in which we operate, we’re still in the closet.
More specifically, my concern is that it’s easy for us to assume, given the energy in the space, that our problems, like how do we architect a data stack, are capital-M Meaningful. If we don’t talk to audiences beyond ourselves, we’ll hear that they are, indeed, quite Meaningful. The only way to get perspective on what is and isn’t is to go beyond the community, and talk to people who don’t, by default, care that much about what we do.
This applies to conversations about topics other than tools as well. We can just as easily lose sight of what’s important when talking about organizational structures and business problems as we can when we talk about tools, if those conversations are still just had amongst ourselves. Debating the ideal structure of a data team or the best way to run an experiment can be just as much of an insider indulgence as debating a data stack.
Ultimately, for me, the concrete lesson in all of this—the takeaway, if you will—is not to stop writing for my Andrew Gelmans, Jia Tolentinos, Michael Lewises, and Michelle Obamas (this offer still stands). There may not be business value in some of these conversations, but there’s connective, human value, and that’s still worthwhile. The lesson, instead, is to find people like them, whose professional opinions I respect and whose professional respect I want, outside of the data world, and write for them too.
As far as I know, these people don’t read this blog. Which is for the best, because I don’t think I could handle it. But a few others who I respect just as much do, and their opinions can make or break my feelings about a post.
Many of today’s “independent thinkers” are clearly infected by the same intellectual rot. Writers like Matt Yglesias and Bari Weiss talk about how they’re willing to tell clear-eyed and courageous truths in front of a frothy mob of cancel culture warriors. I’m dubious. They write not for the unwashed masses, but for each other. They have the “courage” to say what they say because they know the others in their club—the people who they actually respect; the hosts of the parties they actually want to go to—will agree with them, or at least applaud them for saying it. This is why, given enormously complex problems, from how to respond to Covid to how to address centuries-long social problems in America, these supposed free thinkers, constrained by no editor and driven by nothing other than their prodigious intellect, always seem to agree with each other. Either they’re just smarter than us, or they’re chasing their own religion: The approval of one another.
I.e., six likes and a quote tweet that says, “I agree with this theory, but don’t think it’s right in practice.”
Unless your happy hour includes an actual, living anaconda—which allegedly happened in Austin—in which case I’ll enthusiastically join a frothy mob of cancel culture warriors to banish you from polite society.