Small, Crucial OSS Teams Could Use a Suit

Yesterday, Phil Dini asked, “What if all modern open source was a ZIRP [(Zero Interest Rate Phenomenon)¹]?”

It’s an interesting question. Cheap cash and less scrupulous investors helped fuel giant R&D budgets at FAANG companies and beyond (Uber, Square, and Stripe also come to mind), which led to entire teams at these organizations driving open source projects. Following rising rates and a vibe-shift in investor expectations, these companies started to cut projects and people that couldn’t easily be tied to revenue. Including, in many cases, OSS teams.

Did these cuts doom contemporary open source?

One rebuttal is that the technical corners of the internet are always worried about OSS funding. A major conversation about OSS funding occurred in 2014 – smack amid ZIRP – during the Heartbleed fiasco, when a major bug was found within OpenSSL.

“The Internet Is Being Protected By Two Guys Named Steve”, BuzzFeed News famously wrote, as the OSS funding conversation briefly bubbled up towards the mainstream. Crucial yet understaffed OSS projects stumbling and causing chaos is a depressingly common story, regardless of rates.

Considering Phil’s question, I remembered a 2016 post by the researcher Nadia Asparouhova. In her post, Nadia writes that open source infrastructure – specifically, “all the tools that help developers build software” – is “the internet’s biggest blind spot,” before touching on stories like the ones above. Nadia enumerated the state of OSS funding in 2016 rather well:

1. Red Hat is doing great. (But nobody believes there will ever be another Red Hat.)
2. Projects that are effectively “sponsored” by a company, like Go/Google, or React/Facebook, are doing fine. (But many projects are not so lucky.)
3. A lot of companies make their software open source as a “loss leader” to kill a competitor, drive an audience to paid products, or build brand and community. (But these aren’t infrastructure projects.)
4. VCs have poured money into a couple of open source infrastructure companies, like Docker or Meteor. (But these are the exception, not the rule, as Sam Gerstenzang recently explained.)
5. A couple of really big projects, like Linux, are well funded even without a business model. (But Linux is as much of an outlier as Red Hat.)

Fast-forward to now, with the effective rate ~5% higher, and not much has changed! Sure, we’ve seen cuts in the 2nd category above and VC investment in general has slowed – but it’s easy to think about examples in both buckets, especially related to AI and data infrastructure. Meta did make big cuts, but it’s funding open source AI projects with comical budgets. OSS data tooling is being funded decently too, with DataBrick’s Unity Catalog and MotherDuck being examples of Nadia’s 2nd and 4th categories.

This is a 2nd potential answer to Phil’s question: attention is now on data and AI infrastructure, not the web, and the OSS funding has followed.

I’m still chewing on the question, but not limiting it to ZIRP. Revisiting Nadia’s 2016 piece reminds me this is a constant discussion, one we haven’t cracked. Small, crucial, but unsexy projects are critical foundations enabling so many products – and yet they remain unfunded. They lack the marketing, business development, and legal resources needed to hunt down funding from sponsors.

More and more I believe we need to pursue the agency model. We need a CAA or Endeavor for nerds. Teams who can hunt down speaking opportunities, training, publishing, and consulting opportunities that fit within OSS devs’ schedules. Teams that understand why OSS devs do what they do, and find ways to keep them focused on the problems they love. Behind the scenes, this is a big chunk of The Linux Foundation’s function, but for giant projects. We need an Ari Emanuel for individual developers.

One of the reasons I keep coming back to the agency model when people talk about OSS budgets is because I think small OSS development is like writing, painting, playing music, or other fields where lots of people want to do it. Coding is a form of human expression. It is thoughts made solid and communicable to others. A developer’s opinions and values are embedded in their code. Like other forms of human expression, the act is compulsory for many.

Every evening, somewhere, a developer bored with her day job starts a new CMS.

And like the rest of the art-based businesses, we need people to handle the logistics that enable the most essential projects to continue and thrive. And that’s probably an agency model.

Last week, both The New York Times and The New Yorker endorsed Kamala Harris’s presidential campaign. Earlier, Scientific American did so as well.

If you read Slow Boring, you can probably guess who I’m voting for in November. But I haven’t written a formal “endorsement” in the presidential election because I don’t think the persuasive power of such a piece would be high. I write articles about policy all the time, based on what I think is important and whether I have something interesting to say. I haven’t written a lot about abortion rights, which is probably Harris’s best issue in the campaign, because it’s not a policy issue with a lot of complexity to suss out, or one where I think my personal perspective is of much interest. In terms of endorsements, an effective use of Slow Boring’s pixels would be something more like this post about which candidates need your money.

I’m told we raised over $500,000 through this landing page, which is a lot (thank you).

I think the odds that I can persuade a Slow Boring reader that the North Carolina Supreme Court races are more important than they realize, or that Rebecca Cooke in Wisconsin is a promising young moderate with a real chance to win, are relatively high. In that vein, I want to add a recommendation. Dan Osborn is running an independent campaign for Senate in Nebraska that has been very smart about staying far away from any hint of “I’m secretly a Democrat.” That means he’s not going to get any party committee money and could really use small contributions from moderate pragmatists.

Another one that I’m really proud of is that back in February of 2021, I wrote about Kathryn Garcia, a then-obscure mayoral candidate in New York City:

Right now Garcia is stuck in a kind of low-name-recognition bubble where she’s low in the polls, so she doesn’t get coverage, so nobody knows she exists, so she stays low in the polls. But if you live in New York, consider telling a pollster you’ll vote for her. If you’re enthusiastic about urban reform, consider throwing her a little money and maybe getting a story about her fundraising written. Post something on Facebook or Twitter or forward this note to a friend. The low name ID trap is very real but also very escapable, and I think it’d be huge for east coast housing politics for someone like this to get buzz.

Obviously, the Slow Boring endorsement was not as influential as the New York Times endorsement she got three months later. But that itself was an example of the dynamic that I was talking about. A little bit of attention to a strong candidate who was being wrongly ignored generated more attention, and eventually an upward spiral of support. She ultimately lost very narrowly to Eric Adams, who had similar positions on the issues but also a history of low-grade corruption rather than technocratic competence, and that does not seem to have turned out so well.

Subscribe now

Which is why I think it’s crazy that earlier this summer, the Times announced that they’re going to stop endorsing candidates in local races. It’s great news that Ben Smith and some other NYC-based journalists are planning to step up and fill the void, though of course exactly how great it ends up being depends on the wisdom of their endorsements. The point, though, is that I think all these publications should think harder about what they are doing. A lot of political engagement is primarily expressive — people saying and doing things to feel good. But we ought to try to be instrumental and spend our time and resources doing things that have beneficial effects.

Persuasion is mostly about new facts

Rather than writing a detailed summary of a bunch of different papers, I’m going to gesture broadly at the collected works of David Brockman (often with co-authors) which demonstrate things like media consumption is politically influential (again here), professionals are bad at predicting which messages will be persuasive, some messages are very persuasive, and in-person conversations can be persuasive.

And I think the big takeaway from this work is that facts matter.

Most people don’t know very much about most things, and they also mostly don’t care that much. But everyone knows some information about some things, and has some issues that they care about, often that impact them directly. And providing people with new information can change their minds. Part of Fox News’ propaganda function is that it avoids whole topic areas where the discussion is inherently favorable to Democrats. Most conservatives think that Joe Biden’s efforts to reduce senior citizens’ prescription drug costs are bad on the merits. But most voters like this stuff. Even a pretty right-slanted treatment of the issue would be objectively good for Democrats, as long as it treated the topic as a big deal, conveyed which changes are in the works, and stated that Republicans want to reverse it. Fox mostly doesn’t do that, though. They don’t air segments like, “Here’s why it’s good that Trump is going to slash Medicaid.”

Whether you’re doing endorsements or not, conveying factual information is a very important part of what media institutions do.

And the problem with a New Yorker endorsement of Harris is that it’s really hard for this kind of article to convey anything beyond the fact that most journalists are liberals. It’s difficult in large part because when writing an endorsement, it’s natural to front-load the most important points (Trump is a threat to democracy) rather than the points people are most likely to be unaware of (Trump’s budget submissions would cut urban police force size). But people already know liberal journalists think Trump is a threat to democracy. I also think this, so I respect pointing it out, but nobody is going to pick up the New Yorker and be convinced by this argument.

By contrast, I think that when a prestigious magazine like the New Yorker runs Jay Caspian Kang’s nihilism about the political significance of issue-positioning, that really does help persuade the New Yorker’s liberal audience that there is no cost to Democratic Party politicians saying and doing stuff that New Yorker readers agree with. If Kang is right about this, then of course it’s good for the New Yorker to be equipping its readers with that information. But I don’t think he’s right, and the aggregate impact of an editorial saying “Trump Is A Threat to Democracy” and an article saying “Issue-Positioning Doesn’t Matter” is just to exacerbate the threat to democracy.

Looking back on it, in my donation recommendations post, I didn’t really argue anything at all. The article just presumes that you think the MAGA version of the GOP is a threat to democracy and that abortion rights and fair maps are important. We have readers who don’t agree with those ideas and probably found the post annoying. But I like to think that whether or not you agree with those presumptions, you might find the factual information about state Supreme Court and critical House races interesting. And if you do agree with those presumptions, the article inspires action because it tells you something you don’t know and tells you about something you can do instead of doomscrolling.

Journalists shouldn’t undermine their credibility

The scientific journal Nature endorsed Joe Biden in the 2020 election. This was an unusual move that — according to a Floyd Jiuyun Zhang study published in “Nature” — had perverse and counterproductive consequences, reducing Trump voters’ confidence in the publication and making Biden voters like the publication more, without changing minds about the election:

High-profile political endorsements by scientific publications have become common in recent years, raising concerns about backlash against the endorsing organizations and scientific expertise. In a preregistered large-sample controlled experiment, I randomly assigned participants to receive information about the endorsement of Joe Biden by the scientific journal Nature during the COVID-19 pandemic. The endorsement message caused large reductions in stated trust in Nature among Trump supporters. This distrust lowered the demand for COVID-related information provided by Nature, as evidenced by substantially reduced requests for Nature articles on vaccine efficacy when offered. The endorsement also reduced Trump supporters’ trust in scientists in general. The estimated effects on Biden supporters’ trust in Nature and scientists were positive, small and mostly statistically insignificant. I found little evidence that the endorsement changed views about Biden and Trump. These results suggest that political endorsement by scientific journals can undermine and polarize public confidence in the endorsing journals and the scientific community.

At the time of Nature’s endorsement, I guessed something like this would happen. But it’s hard to know until you try it! Scientific American, by contrast, really has no excuse.

Note, though, that the study which explains why this kind of endorsement is a bad idea also explains why it happens: Democrats liked Nature more after the endorsement, and most of the staff of Scientific American are Democrats, so it made them happy to endorse Kamala Harris. That said, even though I find this understandable, that doesn’t make it forgivable. “I did this irresponsible thing because of the short-term psychological high” is a reason, but it’s not a good reason.

I wouldn’t even say that a science magazine needs to strictly “stick to sports” here. But they should think about what they’re doing. A well-reported and fair article about Trump versus Harris on funding for scientific research could be politically impactful. Or, if you want to find one particular member of Congress in a tough race to single out as a science hero, someone who people in the overlapping circles of “interested in science” and “interested in liberal politics” might want to support, that would be useful information. But just reiterating standard liberal criticisms of Trump does not provide any new information to swing voters about Trump; what it does is inform people that the staff of the magazine are all liberals, and thus non-liberals should perhaps ignore them.

Subscribe now

Local endorsements matter

This brings us back to the question of local endorsements.

There’s evidence from the 1960s and 1970s that newspaper endorsements made a meaningful difference in presidential elections by pushing more people to vote Republican. Obviously, one would not want to generalize that result to today’s media climate, where almost no newspapers are endorsing Republican presidential candidates. That said, when you think about it, it is probably is true that if the New York Times had endorsed Donald Trump, that would have made a difference. If they were like, “Look, this threat to democracy business is bullshit, you can trust his promises on abortion rights, and it’s true that Kamala Harris wants to flood the country with illegal immigrants,” that would make waves. The reason the kind of endorsements I’ve been talking about are so inefficacious is because they are so unsurprising.

And, indeed, a study of the 2000 and 2004 presidential elections finds a modest but real electoral impact of “surprising” endorsements.

So it’s not that newspaper endorsements are pointless. It’s just become pointless for newspapers in today’s highly polarized era to repeat that the Orange Man Is Bad, even though I absolutely agree that he is very bad.

In local races, though, there are more opportunities to surprise. Big cities like New York don’t have meaningful party competition, so they get a lot of primaries between candidates who are not clearly differentiated. A basic signal that just says “the kinds of people who enjoy our op-ed columnists are all voting for Garcia” is extremely meaningful, because it conveys actual information that people don’t have.

A group of urban reformers in San Francisco created an organization called Grow SF that is almost nothing but an endorsement machine. They publish various reports and bits of content, but the main purpose of the enterprise is to identify a set of common sense, pro-housing Democratic candidates for local office and draw attention to them. They’re even able to do things like say three of the five candidates for mayor all constitute acceptable choices, and you should rank them 1/2/3 in some order on your ballot. It’s a great project, but the difficulty for Grow SF is they’ve needed to generate an audience from scratch. If you’re the NYT, and you already have an audience, it’s crazy to throw it away. And, of course, endorsements can be even more important in cities that (unlike SF and NYC) don’t have ranked-choice voting. When you have a big field and first-past-the-post voting, the coordination function of prominent endorsements is very influential.

I do sort of get that today’s New York Times isn’t really a local New York newspaper, so they feel weird about making local endorsements. But if anything, they should go in the opposite direction and start endorsing local races in more markets. This is the field where endorsements stand the highest chance of swaying opinions and, in particular, of offering some kind of principled reform politics as an alternative to the status quo urban machine.

Share