The Economics of the Modern Web: The Three Percent Problem

Thunderbird, the open source email client, recently showed a donation appeal to its users. One line stood out:

"All of the work we do is funded by less than 3% of our users."

No ads. No corporate backing (sort of). No selling your data. Just: we built a thing that millions of people rely on, and 97% of those people have quietly decided that someone else will pay for it. And that 97% might be low for the industry.

This isn't really a Thunderbird problem. It's the operating condition of most things worth having on the internet.

---

The economics of "free" content have a hidden cost structure that generates some conversation but not a lot of action. When something is free and ad-supported, the real product is your attention, and the incentive - the thing that actually drives revenue - is engagement. Not quality, not accuracy, not usefulness. Engagement. Time on page, since ads can reload and deliver more impressions. Comments, because they create more impressions and because the best comments garner a lot of reactions and thus shares and reloads and reposts. It all comes down to "how do I engineer more ad deliveries, and the way to do that is RAGE."

This is why ragebait is everywhere. This is why headlines stopped being descriptions of articles and became traps. This is why journalists started using adverbs everywhere. This is why CNN is unwatchable (especially when CNN leans into adverbs. And CNN is hardly unique.)

It isn't that the people running these organizations are villains, or stupid, or just bad writers; it's that the revenue model requires them to optimize for the things that make you angry enough to click, comment, and come back. Every reload: another ad impression. Outrage is just conversion rate optimization with worse consequences. Time spent writing is time wasted when an AI can generate milquetoast-quality content loaded with dopamine triggers.

The alternative - people directly funding things they value - keeps almost working. Patreon exists. Ko-fi exists. Substack exists. Thunderbird has a donate button. And yet: Thunderbird claims a whopping 3% of its users actually help the project survive.

The collective action problem here is brutally simple. People assume that if a project is worth using, someone else will fund it - so they don't have to. If a writer is worth reading, some other reader will reward them. The individual cost of opting out is zero, and the individual contribution feels too small to matter, so everyone opts out and waits for the "someone else" who isn't coming.

---

The free-rider problem is only part of the story, and probably not even the most interesting part. The harder problem is that the ask itself is complicated.

Thunderbird's donation appeal is on Hacker News. It has a pretty busy comment thread - and the "busy" wasn't people going "Oh yeah, I should keep this thing going." People who'd donated before were questioning whether they'd do it again.

The organizational structure around Thunderbird can be genuinely confusing: MZLA Technologies Corporation is a for-profit subsidiary of the Mozilla Foundation, which also controls the Mozilla Corporation, which brings in around $500 million annually - mostly from Google - and spends a remarkable share of it on manager salaries and travel. Someone worked out the math in the thread: Mozilla's manager salaries alone dwarf what Thunderbird receives in total donations. Meanwhile, Thunderbird's own donation page, when you try to leave it, redirects you to a page asking for your email address so they can remind you to donate later. That's... not the back button's job.

Thunderbird itself makes the ask intrusive, and that's both understandable and faintly repellent.

And then there are the people who simply don't trust the product anymore. Long-time users have mentioned regressions and confusing workflows; I don't use Thunderbird myself because it does not work well for me. One commenter summed up his relationship with Thunderbird by noting he had it pinned to the current version specifically to prevent it from updating, and was planning to shop for a replacement when his next OS upgrade forced the issue.

This is the part of the 3% problem that doesn't get discussed enough. The math of collective action is bad enough on its own. It gets worse when trust is eroding - in the product, in the organization, in where the money goes. You can't ask people to fund something they're not sure deserves funding, and "please keep this going" is a harder pitch when the honest follow-up question is "keep it going in which direction?"

---

This plays out in open source in ways the Java ecosystem knows intimately, even if we don't always name it directly. Spring exists today because Pivotal (then VMware, then Broadcom) decided it was worth funding, after the consultancy fees paled in comparison to corporate sponsorship. The JDK exists today because Oracle and a consortium of corporate actors decided it was worth funding, and Sun had open sourced it before it could be locked behind a paid model. Jakarta EE exists because Eclipse and IBM and Red Hat decided it, too, was still worth funding. The pattern is: a thing becomes valuable, corporations realize they need it, corporations start paying for it - not out of altruism, but because the alternative is worse.

That works until it doesn't. Broadcom's acquisition of VMware has been an object lesson in what happens when the corporate calculus changes. The open source project you depend on is always one strategic pivot away from becoming someone else's leverage.

The developers who maintain the libraries you use every day, the ones without Fortune 500 backing, are mostly working on your behalf for free.

Some of them burn out quietly and walk away.

Some of them get acquired and the project goes dark or goes away entirely. (Anyone remember GLUE and EXML, still some of the best examples of web service implementation and XML parsing at the times they existed? What about projects like XStream, or Sitemesh? Can you name any other projects you used to love but somehow went away?)

Some of them flip to closed source - which, honestly, is just them deciding that the 3% problem is someone else's problem now, and can you really blame them? (That's what happened to GLUE¹, and EXML went away entirely.)

---

It'd be nice if there was a clean solution. The micropayment dream - "ooo I like this, here's a dollar" - has never materialized in any usable form outside of gaming. If it existed for developers, the collection action math still isn't likely to add up to much; there'd be an occasional donation, and those are incredibly rewarding - mostly because of how rare they are, and most such donations might buy part of an inexpensive dinner. There are people who reward effort, deliberately, but most don't, and most won't.

When we choose ad-supported content, we are choosing to fund engagement optimization and all its downstream consequences. When we demand that software be open source without asking how its authors will eat, we are making a choice about who absorbs the cost. When we reload the outrage machine one more time, we are voting for more of it.

And we fight even that process, by running adblockers, which means there's a race against providers: providers end up investing effort that they could use to create good content in creating more effective ads, which means we end up with even more complex ad evasion, and more complex ad delivery schemes... effort that could go into creating good content. Because we don't want to pay for it. We think to ourselves, "It's out there, I can get it for free, it should be free."

Thunderbird is asking 3% of its users to keep the lights on. That's a sustainable enough number, if you squint. But it's also a confession: the other 97% have decided, without deciding, that this is someone else's problem.

It doesn't have to stay that way. But it probably will.