Infogami flopped—there’s no other way to say it. Swartz had no experience building something this ambitious from the ground up. He had trouble attracting investors, as he struggled to articulate the specific problem that Infogami was supposed to solve. And Swartz and his roommate/collaborator, a young Danish programmer named Simon Carstensen, did not make great partners. When Carstensen arrived at the small, stuffy MIT dorm room that they’d share that summer, it was the first time the two men had ever met. “I think my memory is sitting in this dormitory and coding and it being very hot,” Carstensen recalls.
Swartz rewrote a lot of Carstensen’s code, and at the end of the summer, the Dane returned to Europe and didn’t work on Infogami again. Swartz stuck around Boston, but he started to lose patience with his startup. He was alone, staying at Paul Graham’s house, and he was miserable:
One Sunday I decided I'd finally had enough of it [Infogami]. I went to talk to Paul Graham, the only person who had kept me going through these months. "This is it," I told him. "If I don't get either funding, a partner, or an apartment by the end of this week, I'm giving up." Paul did his best to talk me out of it and come up with solutions, but I still couldn't see any way out.
The next night I had dinner with Paul and his friends. They noted my birthday was tomorrow and asked me what I wanted. I thought for a moment about what I wanted most. "A cofounder," I finally said. We all laughed.
Around the same time, two other Y Combinator participants desperately needed help for their own startup, a social news site called Reddit. Graham’s solution was simple: Infogami would merge with Reddit, creating a new umbrella company called Not a Bug. Swartz moved into Alexis Ohanian and Steve Huffman’s Davis Square apartment and got to work.
The story of Swartz’s time at Reddit is a complicated one. Swartz wasn’t involved in conceptualizing Reddit; when he came aboard, the project was already on its feet. While Ohanian and Huffman didn’t consider him a co-founder, Swartz often employed the term.
What can be said definitively is that Swartz was unhappy at Reddit, and the Reddit crew was unhappy with him. The original plan was for Huffman to help Swartz with Infogami while Swartz helped with Reddit, and for both projects to run off of a common back end that the two of them would build together. Huffman was excited to work with Swartz, whom he considered a talented programmer. “There was a time for a couple of months when we were like, OK, we're gonna start this new thing and it's gonna be bigger than Reddit, bigger than Infogami,” says Huffman. “And then it became pretty clear a few months in that that was not gonna be the case.”
The Infogami launch didn’t go well, and Swartz became demoralized. “He got sick of working on Reddit, and I never wanted to work on Infogami at all,” says Huffman. “That was basically the beginning of the end.”
For months, Swartz did no work on Reddit, though he continued to live with Huffman and Ohanian. His fitful mind focused on other things, including an unsuccessful run for a position on the Wikimedia Foundation’s board of directors. And, as he’d done during his lowest periods in high school and college, he vented his frustrations on his blog. “I don’t want to be a programmer,” he wrote in May 2006, when he was 19. “When I look at programming books, I am more tempted to mock them than to read them. When I go to programmer conferences, I’d rather skip out and talk politics than programming. And writing code, although it can be enjoyable, is hardly something I want to spend my life doing.”
Swartz’s quest for personal fulfillment made for an unpleasant work environment. With Huffman serving as Reddit’s sole full-time engineer, it was unclear whether the site would succeed. But in October 2006, 16 months after Reddit was founded, Condé Nast purchased the company for an undisclosed sum, reportedly somewhere between $10 million and $20 million. Some of that money went to Graham and some went to Chris Slowe, a part-time programmer who had equity in the company. Swartz, Huffman, and Ohanian split the rest three ways. (Swartz gave some of his money to Simon Carstensen, as a reward for his early work on Infogami.)
As part of the deal, the Reddit team moved from Boston to San Francisco; the openly unhappy Swartz was not expected to go with them. But for some reason—be it masochism, a sense of duty, or a belief that a change of scenery might make things better—he decided to move back west.
The Reddit staff worked out of the Wired office, in a corner that had been cleared especially for them. This was a physical environment very similar to the idealized “modernly-designed loft” the teenage Swartz had dreamed about in 2001—a big, open space with a personal chef who made everyone breakfast. But Swartz was predictably miserable, and not at all suited to corporate life. Every meeting, every banality must have seemed like a moral maze. He stopped showing up at the office, wrote blog posts critical of his co-workers, took impromptu trips to Europe, and did very little work. Once again, he moved to separate himself from an environment that failed to live up to his ideals. This time, though, he didn’t get to leave on his own terms. Less than three months after the Condé Nast acquisition went through, Ohanian and Huffman asked Swartz to leave.
Around the same time he was fired from Reddit, Swartz wrote a long story on his blog about a man who starved himself, then committed suicide.
The day Alex killed himself, he was awoken by pains, worse than ever. He rolled back-and-forth in bed as the sun came up, the light streaming through the windows eliminating the chance for any further sleep. At 9, he was startled by a phone call. The pains subsided, as if quieting down to better hear what the phone might say.
It was his boss. He had not been to work all week. He had been fired. Alex tried to explain himself, but couldn’t find the words. He hung up the phone instead.
The day Alex killed himself, he wandered his apartment in a daze.
“Alex” was originally named “Aaron.” (Swartz took the post down for a time, then changed the character’s name when he reposted it.) Alexis Ohanian, whose then-girlfriend had attempted suicide less than a year earlier, called the police and had them check on Swartz, who was fine. He later explained away the blog post as a misunderstood attempt at fiction: “I was deathly ill when I came back from Europe; I spent a week basically lying in bed clutching my stomach. I wrote a morose blog post in an attempt to cheer myself up about a guy who died. (Writing cheers me up and the only thing I could write in that frame of mind was going to be morose.)”
It’s impossible to say whether this post was a melodramatic story, a cry for help, or a bit of morbid foreshadowing. At the time, though, it served as a dramatic and definitive end to an unsatisfying chapter in his life. Swartz was 20 and jobless. But now, at least, he felt like he had control of his life again.
* * *
Sooner or later, every true believer gets the opportunity to sell out—to put money or comfort or expediency ahead of what he claims to value most. In the late 1970s, the MIT AI Lab was falling apart. Several programmers had left to form their own company, Symbolics, which built and sold Lisp machines and the software that was used to run them. This was the same computer language that, up until then, the hackers at the AI Lab had been developing for free. Stallman saw this as a tremendous betrayal and fashioned himself into the last bulwark of idealism in an environment where it once flourished. The last of the true hackers performed incredible feats of programming to match—by himself—every Symbolics software update, ensuring that the free version would be just as good as the one you had to pay for. Soon thereafter, Stallman left the AI Lab to start the Free Software Foundation, and make idealism his life’s work.
Swartz’s separation from Reddit, though not as drawn out and openly adversarial as the Stallman/Symbolics fight, was similarly ideological. Though he didn’t articulate it at the time, it’s clear in hindsight that Swartz hated the feeling of doing something for the money. After the Condé Nast sale went through, he described his guilt over being paid so much for a project that he believed was so insignificant:
Once I went far outside the city to have lunch with an author I respected. He asked about what I did, wanted me to explain it in great detail. He asked how many visitors we had. I told him and he sputtered. “I’ve spent fifteen years building an audience, and you’re telling me in a year you have a million visitors?” I assented.
Puzzled, he insisted I show him the site on his own computer, but he found it was just as simple as I described. (Simpler, even.) “So it’s just a list of links?” he said. “And you don’t even write them yourselves?” I nodded. “But there’s nothing to it!” he insisted. “Why is it so popular?”
Inside the bubble, nobody asks this inconvenient question. We just mumble things like “democratic news” or “social bookmarking” and everybody just assumes it all makes sense. But looking at this guy, I realized I had no actual justification. It was just a list of links. And we didn’t even write them ourselves.
Leaving Reddit led Swartz to reassess his life. Any single project might turn out to be a slog or a bore or a gigantic pain in the ass. The solution to that problem was to sign up for “all the interesting projects that came my way.” He worked on another startup, Jottit, with his old Infogami partner Simon Carstensen. He also helped the Internet Archive’s Brewster Kahle launch Open Library, an ambitious effort to create a landing page for every book in existence.
Increasingly, the projects that interested him were at the intersection of data and governance. Swartz had grown up in a family where activism was valued, and his early work with Creative Commons gave him a cause of his own. In October 2002, the 15-year-old Swartz went to the Supreme Court as Lawrence Lessig’s guest. The law professor was serving as plaintiff’s counsel in Eldred v. Ashcroft, a key case in modern copyright law. Lessig argued that the court should overturn the Sonny Bono Copyright Term Extension Act, which extended copyright protections significantly. Though Lessig lost the case, the experience was formative for Swartz—he was an ardent copyright reformist for the rest of his life.
In the intervening years, both Lessig and Swartz became more explicitly political. In 2008, the law professor co-founded an organization called Change Congress, which urged citizens to do just that. Swartz eagerly signed on, joining the group’s board. The project sparked an interest in electoral politics. Swartz reached out to friends in Washington, polling them about how the system worked, and how it could work better. “If you talk to people in Silicon Valley, they often say that people in politics are stupid,” notes Matt Stoller, one of the people Swartz contacted. “Aaron didn’t think that way. Aaron thought people in politics are people, and they operate in a system, just like Silicon Valley is a system. And you have to learn these systems if you want to manipulate the process.”
Swartz soon realized that his talent for accessing and synthesizing information could have political value—that he was a better programmer and data gatherer than most other activists, and more ardent in his activism than most software aces. Swartz helped found a group called the Progressive Change Campaign Committee, which was devoted to electing progressive candidates to Congress. He got a Sunlight Network grant to start a site called Watchdog.net, which gathered information about voting records and campaign finance and gave its users tools to manipulate and present that data themselves.
But Swartz soon had his eyes on a larger prize: PACER, the electronic storehouse for federal court records. PACER’s contents are all public records, and anyone can access them via the Web for a small fee (now 10 cents per page). An open-government activist named Carl Malamud argued that these documents shouldn’t cost a dime, as they’re government products and thus non-copyrightable. In 2008, Malamud put out a call for like-minded folk to grab as much from PACER as possible during a trial period in which documents would be downloadable for free at select libraries.
This call for action held a natural appeal to Swartz, an avowed enemy of copyright restrictions and supporter of open data. In September 2008, he went to a library in Chicago and installed a Perl script that pulled a new document from PACER every three seconds. Before the library caught on and shut him down, he’d downloaded 19,856,160 pages, which he donated to Malamud’s open-government site Public.Resource.Org.
For Swartz, the decision to install that Perl script represented a subtle but dramatic shift in his ideology and methods. He had always believed that information wants to be free. Now, he had acted as its liberator. He’d also chosen to tweak the federal government, actively mocking a system that was resistant to change.
Though they’d done nothing illegal, the PACER stunt earned Swartz and Malamud some FBI attention. Swartz later acquired his FBI file, which indicated that agents had surveilled his parents’ Highland Park home. That FBI file, Swartz said, was “truly delightful.” At the time, it all seemed funny—the feds getting so upset over something so minor. But Malamud now believes the PACER downloads contributed to the government’s subsequent fervor in prosecuting the JSTOR case. In their eyes, Swartz was a repeat offender, a data vigilante. This was no small thing.
* * *