Disney erased FiveThirtyEight

Nothing on the internet lasts forever. But Disney’s 10-year mismanagement of FiveThirtyEight is its own story.

Fivey Fox by Joey Mills — Silver Bulletin illustration

Last Thursday night, I was working late, trying to put some of the finishing touches on our forthcoming World Cup model — and actually looking up an article I’d written for FiveThirtyEight in 2014 about my previous soccer model, SPI. Although the quality of the archive has gradually deteriorated since Disney shut down the site in 2025 (I left two years earlier in 2023), at least our text-based articles were mostly still there, or so I thought. Instead, I was auto-redirected to ABC News’s home page, which looked something like this:

Sometimes weird things happen on the internet late at night, so I resisted the temptation to tweet something about it. But one of my former colleagues noticed the same thing on Friday. ABC News hasn’t made any public comment that I’m aware of — they declined to make a statement to the New York Times, which wrote about FiveThirtyEight’s disappearance. It’s possible that they have something up their sleeve, I suppose. But presumably, this was either intentional or willfully neglectful. All of the former FiveThirtyEight site from my nearly decade-long tenure at ESPN/Disney/ABC is gone.¹

It’s common to read things like: “what happens on the Internet stays on the Internet”, the notion being that you can never escape your digital past. But this isn’t really true. A Pew study of a random sample of Internet links conducted in October 2023 found significant “link rot”: almost 40 percent of links that had been active 10 years earlier were broken. And that’s probably an underestimate: the study was based on the Common Crawl web archive (the same one that AI labs use to train their models), which is quite comprehensive but probably contains some bias toward more prominent sites. Another study by ahrefs found a two-thirds attrition rate for web links after 11 years.

Yes, you can still access (for now) Disney-era FiveThirtyEight content via the invaluable Internet Archive, and pre-Disney-era content from The New York Times (which I partnered with from 2010 through 2013). And obviously, we’re trying to recreate some of the most popular parts of FiveThirtyEight at Silver Bulletin. The election models and polling averages are here, and new-and-improved versions of the sports models (PELE, ELWAY, COOPER) are gradually returning too.² Galen Druke, Clare Malone and I have even been getting the old podcast crew back together for live shows.

To be clear, we’re not trying to create a full-fledged version of FiveThirtyEight. Having a smaller team gives me more time for creative work, such as writing and building models. In fact, post-Disney life is better along pretty much every quantifiable and intangible dimension.

Still, these abstractions about “link rot” don’t quite capture the feeling of seeing so much hard work erased.

200,000 hours of work erased

Here are some numbers roughly in the right ballpark: during the Disney era, which lasted about 10 years, FiveThirtyEight published about 20 stories a week. Let’s say that each story took about 20 hours to produce between research, writing, graphics and editing.³ Do the math, and that works out to about 200,000 person-hours of work that ABC News just deleted.

What’s probably harder to see from the outside is that none of this was ever smooth sailing: that content was produced through a lot of blood, sweat, and tears (sometimes literally on certain election nights).

Maybe the deletion of the site has me feeling wistful, but I thought I’d write about the various phases of FiveThirtyEight, including some inside-baseball stuff that I’ve never really talked about publicly.

I’m just going to speak from my vantage point, not on behalf of the several dozen people who worked on the site in its various iterations. I’m not going to talk about competing editorial visions or personality clashes or any of that, though we had our fair share of newsroom drama. Instead, this is intended as more of a business school case study: a study of a large corporation, The Walt Disney Company, behaving in an incredibly neglectful way toward a smaller brand it acquired, with dozens of employees who worked exceptionally hard despite constant existential uncertainty.

I want to state one thing up front: I think FiveThirtyEight could have been a highly valuable business if it had been managed more carefully. I know the New York Times considered FiveThirtyEight a valuable part of its subscription offering. I know some of the sharper minds behind subscription-based businesses tried to acquire FiveThirtyEight at various times. And I know the economics of Silver Bulletin, and they’re good. There are challenges: our traffic is highly cyclical, increasing severalfold around major elections and sporting events. Still, my guess is that FiveThirtyEight could have had north of 100,000 paying subscribers by this point: in the same general ballpark as The Free Press, which recently sold for $150 million.

The thinking at Disney is presumably that they invested a lot of money in FiveThirtyEight and were left with nothing to show for it. But to my mind, however much they spent on FiveThirtyEight, they never invested a dollar in it. There was never really any effort, or even any pretense of trying, to make it a profitable unit of the company. At one point, other senior staffers and I basically begged Disney to turn on a paywall, figuring this could provide some security, and were told, essentially, that it just wasn’t worth Disney’s bandwidth to figure out the mechanics of one. Essentially, we were treated like an unused gym membership: you don’t want to cancel because you think you ought to be hitting the gym, but every month a charge hits your credit card statement and you aren’t getting any fitter.

The origins of FiveThirtyEight.com

I founded FiveThirtyEight.com under the pseudonym “poblano” in March 2008 as a spin-off from a popular series of posts at the progressive blog Daily Kos. For the previous several years, after quitting my corporate consulting job in 2004, I’d had two main sources of income. One of them was building statistical models like PECOTA for Baseball Prospectus, an early adopter of a subscription-based business model that still exists today and has graduated dozens of staffers into Major League front offices. BP had good timing going for it — this was the dawn of the Moneyball era in sports. But the subscription-based offerings were relatively rare at the time. BP churned out enough revenue to support a middle-class income for around a dozen co-owners/stat nerds/writers but not much more than that.

However, I didn’t really care because my real source of income was playing online poker. Yes, there really was a time when you could click buttons for a living and make a pretty good income by working the late shift against what I imagined to be drunk Scandinavians. It seemed too good to be true, and ultimately it was. The passage of the UIGEA in late 2006 essentially cut off new money flowing into the game, the remaining players were getting sharper, and the format I specialized in at the time (limit hold ‘em) was gradually losing ground to no-limit hold ‘em, which I wouldn’t become proficient at until years later.

The UIGEA, passed on the last day before Congress adjourned for the 2006 midterm elections, did pique my interest in American politics, however. I wanted the bastards who had deprived me of a living to be voted out of office, and they largely were, including the chief sponsor of the bill, the 15-term Iowa Congressman Jim Leach.

“Moneyball, but for elections” was a logical enough pitch, but I didn’t anticipate the degree to which FiveThirtyEight.com would become a viral hit in 2008. It helped that the election featured a number of compelling personalities, including a guy whose name I’d vaguely remembered from my time at the University of Chicago: Barack Obama. I ran ads on FiveThirtyEight, and the money wasn’t fantastic by any means, but digital advertising rates were comparatively healthier back then, so it was enough to pay the rent.

FiveThirtyEight, de-anonymized midway through the 2008 cycle, drew a lot of media attention, particularly after the model’s highly confident “call” that Obama would defeat John McCain proved correct. On a train platform on my way back from the MIT Sloan Sports Analytics Conference conference in 2009, I encountered a senior editor at the New York Times who had been in attendance. I’ve long been an admirer of the NYT — my parents in East Lansing, Michigan even had a ritual of walking to the bookstore every day to buy the print edition. Long story short, the NYT made a good-enough offer. The finances weren’t fantastic, but it was an extremely clean deal: I was a contractor, not an employee, so I was free to pursue other sources of income, and I regained full ownership of the models and the other IP after I left the Times.

Honestly, I expected to renew with the Times. At the risk of being immodest, the 2012 election forecast had been a smashing success, with the FiveThirtyEight model famously “calling” all 50 states correctly that year in a stroke of luck that later felt like a bit of a curse. (There was only about a 3 percent chance this would happen, according to the model’s internal logic.) The Times had launched a digital paywall midway through my time there, and its subscription revenues had grown significantly in the 4Q of 2012, coinciding with the Obama-Romney election. How much FiveThirtyEight was responsible for this was hard to say — the Times has had a lot of success without us, obviously — but I had a lot of leverage.

“We” (my attorney and I) basically decided to give the Times an exclusive negotiating window before exploring the market. My keen sense at the time was that the NYT would not be the highest bidder, but I really did like working there, and they deployed me creatively on everything from the Magazine to the book review. The Times was a deeply factional place in those days, though, and the FiveThirtyEight product had both internal champions and internal critics.⁴ The Times was also in the midst of a leadership transition, and new management tends to want to move on from the old regime’s pet projects, even if they were successful. Although I’d moved to an apartment within walking distance of the NYT office in anticipation of a new deal, the Times dragged its feet to the point where we eventually felt like we had no choice but to test the market.

We wound up signing up with ESPN. It doesn’t seem like the most obvious fit now, but this came during an era when ESPN basically thought of itself as the best business in the world, guaranteed an annuity based on cable rights fees. Its then-president, John Skipper, had highbrow ambitions for premium products, notably Grantland. Grantland, built around another spiky founder, was quite explicitly a precedent for FiveThirtyEight @ Disney. I don’t catch up much with Bill Simmons these days, but he was a helpful consigliere during the negotiations, offering reliable advice on navigating any and all things Disney-related.

For the record, other offers we seriously considered came from NBC News, Bloomberg and the Times. The latter two would probably have been better fits, because they’re basically in the same business as Silver Bulletin/FiveThirtyEight: selling premium subscription products.⁵ ESPN and NBC are primarily television networks, by contrast. I’ve never really liked going on TV, and it’s probably my worst medium despite some genuinely good-faith efforts by ABC News to make it work.

The early Disney years: mistakes were made (mostly by me)

In hindsight, I chose poorly. Skipper had been fairly explicit that he didn’t really care whether FiveThirtyEight made money: like Grantland, we were essentially a hood ornament on ESPN’s oversized SUV and a “rounding error” relative to Disney’s gigantic P&L.

That might seem like a fantastic situation — Disney’s pitch was basically “we’ll cut you a nice check to be maximally creative” — and I consider myself privileged to have been given the opportunity. But there were several problems with the arrangement.

The b-school way to put this is that there was never brand or incentive alignment — with Disney being very “macro” (it takes huge swings: theme parks, Marvel movies, NFL rights deals) and FiveThirtyEight being very “niche”. Furthermore, the Disney attitude basically seemed to be that “creatives” were sensitive artist types who didn’t care about money so long as they made enough of it, but I’ve always had a fairly entrepreneurial spirit.

But most of all, this sort of arrangement makes you extremely dependent on your bosses’ goodwill, and Skipper abruptly left ESPN one morning in 2017 because of an extortion scandal.

FiveThirtyEight relaunched @ Disney/ESPN in March 2014 to “mixed” reviews. And by “mixed” reviews, I mean mostly bad ones. The launch was kind of a disaster, really. I’ve written before about some of the Mistakes That Were Made, most of which were predictable mistakes that I made. Among other problems, I did too much bragging in the media and didn’t anticipate the extent to which public opinion toward FiveThirtyEight would shift once we became a corporate-backed incumbent rather than an eccentric upstart. We added too many staffers too quickly, perhaps anticipating that we’d never have more leverage to add “headcount” than when we were a new, shiny object for Disney. Relatedly, we put too much emphasis on quantity and not enough on core products.

The core mistake, though, was that almost nobody was thinking about how to make FiveThirtyEight a viable business. We had essentially no dedicated business people or “product” people or anyone else whose job description depended on the site being economically viable. We never developed the muscle memory or the infrastructure to be a commercial product.

Despite all of this, FiveThirtyEight eventually rounded into a very good and popular website, or so I’d like to think. Our 2016 election forecast was literally the most “engaging” feature on the internet, according to Chartbeat, and our podcast received hundreds of thousands of downloads per episode.

But it never quite felt that way. Whenever we felt like we finally had some open runway, we’d encounter an unwelcome hurdle.

In 2015, I was taken aside while walking with my boss one warm spring afternoon and told that Bill Simmons would be dismissed from Grantland. Grantland had a brief and tumultuous post-Bill period but was shuttered completely by late 2015. This was a very ominous precedent: Grantland had literally been in the same sub-unit of ESPN as FiveThirtyEight, with much of the same senior management.

Then in 2016, there was the whole … Trump thing. Being dismissive of Trump’s chances early on in the Republican primary was the biggest analytical mistake of my career, and I think I deserved a lot of blame for that.⁶ I will insist that our general election forecast did an excellent job that year, though. Our model famously/infamously gave Trump about a 30 percent chance of winning, much higher than prediction markets, other models or the conventional wisdom at the time. That’s not how the rest of the internet saw it, however. It was a difficult election night and a difficult aftermath period. Everyone had their own way of coping; I remember playing an inordinate amount of the EA Sports NHL game and trying to take the Rangers to the Stanley Cup.

We wanted to leave Disney sooner, and we probably should have

Honestly, ESPN took the Trump stuff relatively well. FiveThirtyEight had produced a truly extraordinary amount of traffic in 2016, at least — even if none of it was being monetized. But at some point in early/mid 2017, with my initial contract with ESPN set to expire in early 2018, Skipper called me up to his office one morning and told me basically that FiveThirtyEight could no longer be a thing at ESPN, but he’d work his connections to find a good landing place for it. I don’t know what the impetus was for this — although it came amidst the whole “stick to sports” period at ESPN, and FiveThirtyEight obviously mostly wasn’t sports. For what it’s worth, though, I thought Skipper was relatively sincere, and we did take a lot of meetings.

But the resulting negotiations were fraught. There were plenty of “suitors” interested in a pared-down version of FiveThirtyEight, which would trim staff by perhaps half to two-thirds. For better or worse, I’d chosen to place a big emphasis on staff retention, even though the market consensus was probably right that the business model would be better with a leaner staff. The two main external suitors were The Athletic (later sold to the New York Times) and, confusingly, The Atlantic. I think highly of the leadership at both organizations, and they had the right idea: FiveThirtyEight could be a compelling subscription product even if it wasn’t viable based on web advertising alone.

We came quite close to securing a deal with The Athletic, close enough that the founders came to New York for an entire week of meetings to sort through every detail. I’d expected things to culminate with a handshake agreement and a celebratory lunch before they headed back to California. If you know anything about my organizational skills, they aren’t great, but I thought we (me and the other senior staff) did a reasonably good job of softening the ground for a potential transition to The Athletic and a subscription-based business model. This wasn’t a hard sell because most of the staff would be offered jobs at the new organization.

Long story short: the potential deal with The Athletic hit a last-minute snafu, which there might have been time to work around if there hadn’t been a hard deadline imposed by Disney, but Disney needed a decision from us. The sale process was complex: Disney owned some of the key IP (the trade name and site archive), while I owned some of it (the models). Moreover, Disney was both running the sales process and represented one of the bidders (ABC News). Somewhat bizarrely, ABC, a Disney related party, had entered the bidding at some point midway through the process; apparently, there had been some signals crossed about this in Burbank, and ESPN hadn’t realized that ABC had been interested.

So we put our tail between our legs and signed up for another tenure at Disney, doing our best to make it seem as though this had been the plan all along when it obviously wasn’t. Although ABC News was nominally a better fit than ESPN — they did put me and other staffers on TV more often — I was quite certain at the time that this was going to be our last hurrah at Disney. It became apparent, even before there was any ink on the deal, that while ABC News was happy enough to spend money on FiveThirtyEight with major elections forthcoming, there was the same issue as with ESPN: they weren’t really looking to invest in the property in a way that might make it profitable and sustainable. And even more so than ESPN, ABC News was a dinosaur of a business that attracted mostly older customers and managerial talent.

Nevertheless, the period from roughly 2018 to 2019 was probably the high point of FiveThirtyEight at Disney. It helped that our 2018 midterms forecast basically totally nailed the outcome, offsetting at least some of the stink from 2016. Still, the existential uncertainty about Disney’s loyalty to the site persisted.

This was when we pitched the paywall idea to upper management. Disney batted it around and ultimately turned us down. We never quite got a coherent rationale for why: something something about Disney being busy with its Hulu acquisition and not wanting to launch multiple subscription-based businesses at the same time.

It didn’t make much sense. My content had been paywalled at the New York Times and at Baseball Prospectus before that. Long before Substack came along, it was the business model that works for differentiated, high-quality content. (For “niche sites”, if you will, although elections and sports are large niches.) I don’t know what the paywall would have made, but after a year or two to get settled, it would probably have been comfortably in the seven figures annually: let’s call it $5 million. The cost might be annoying some customers who were used to free content, but we were hardly running any ads on the site anyway.⁷

What sort of business basically turns down a “free” $5 million? Well, apparently, a company like Disney, a company that made $69 billion in revenues in 2019. If you’re a rounding error when you’re losing a couple of million bucks a year, you’re a rounding error if you’re making a few million also.

The most generous interpretation is that Disney only understands things that operate at very large scales to truly mass-market audiences. ESPN.com had once been the exception to this, full of quirky offerings like Page 2 and Grantland. But it has largely been lobotomized, too, with talents like Pablo Torre and Zach Lowe having been let go.

The long, weird, bitter end

Then came the COVID pandemic in 2020. We thought we were smart by telling staff to start working from home a day or two before official Disney guidance, not realizing that the newsroom would never really be the same again.

Honestly, I think people forget how difficult this whole period was. I consider myself lucky never to have gotten seriously sick or to have lost any close friends to COVID. But there were challenges if you were working in the media at the time: the adjustment to “going remote” and the political “reckoning” that basically any non-explicitly-conservative-coded media business experienced, and all of this coming on top of an election year.

Even as the world awkwardly and unsteadily got “back to normal,” FiveThirtyEight never really did. Key staffers like Clare Malone were let go without any real plan to replace them. Slowly, the oxygen tubes were being removed.

There was also the matter of my contract. The new deal I’d signed with ABC News in 2018 was nominally 5 years long but included a mutual opt-out after 3.5 years, originally timed to December 2021. Either I or ABC had the right to pull the plug, but if neither of us did, the deal would extend through the 2022 midterms by default.

The signals I was getting from ABC News were quite bearish, but they also weren’t quite ready to make a decision. What I wanted, frankly, was to pare down my responsibilities to Disney — and take an appropriate pay cut — in exchange for non-exclusivity. I’d license the models to them for 2022, produce some designated number of columns and TV appearances. But I wanted to be officially free of managerial responsibilities, serving as more of a mentor/founder, and also free to start a Substack and begin building an audience for it.

ABC’s response to this was basically radio silence. There was never any sort of counteroffer or serious conversation about it. Surely, then, they’d exercise their opt-out right, and we’d each go our separate ways in December?

Well, no. Disney repeatedly asked to push back the deadline for making a go/no-go decision, and my agents and I agreed to several extensions until we got tired of the delays.

The eventual, thrice-delayed deadline came on an exceptionally cold Saturday in February.⁸ I had a late dinner reservation with my sister and some friends, and we went back to her apartment afterward. I checked my inbox in between sips of wine, half expecting to get an email from ABC at 11:59 p.m., but nothing came. I don’t know if they literally forgot about it, but that’s how it felt.

To be fair, I could have opted out also, and if I’d known at the time how much better life after Disney would be, I would have. But I was pretty explicitly in a lame-duck period from that point onward. I’d totally busted my ass for the first seven or eight years of my 10-year tenure with Disney, but in this late phase, I was mostly focused on my book, which I explicitly had the right to work on per my contract.

I received an urgent text message from my boss one morning in April 2023. There had been rumors of job cuts at ABC News, and it could only mean one thing. What was surprising, though, was that while the cuts were both deep and haphazard — while you might have been able to take a go at 2024 with a skeleton crew, you probably wouldn’t have wanted to cut all the managers and editors, as ABC did — they weren’t quite ready to shut down the site entirely. I was not, technically speaking, a part of this layoff, but with only two months to go on my contract, it was safe to assume that they weren’t going to offer me a new deal and I wouldn’t have taken one anyway with the staff gutted and the remaining staffers left in an incredibly challenging place.

To be honest, I’m not sure that ABC News realized that they had no more rights to the models: the license term on the election models expired with my contract.⁹ But they did eventually hire another “model guy”, G. Elliott Morris, to replace me. I am absolutely not looking to extend a beef with Morris, who, like me, seems much happier with his post-Disney life on Substack. But it’s important context to state that this isn’t the person I’d have hired; we’d had a long-running feud, in fact.

I chatted with my ex-boss at about this point in time and asked if they’d consider discontinuing use of the FiveThirtyEight trade name. Even leaving the Morris hire aside, there were a number of other things I did not like, such as beginning to stylize “FiveThirtyEight” as “538” and replacing our carefully refined site design with an ugly ABC News template. (Even though FiveThirtyEight routinely drove more web traffic than the entirety of ABCNews.com.) There was no more sports section, either. The brand was very much being depreciated, and I didn’t want people to associate this stage of the site with its glory days.

What happened in 2024 isn’t something I’d have scripted, though. Basically, their new election model was literally broken, continuing to show Joe Biden virtually tied with Trump even after his disastrous debate. (Evidently because Morris’s design for it had been overcomplicated. These models are hard to design, by the way.) The model was taken offline for more than a month after Biden dropped out, missing basically the entirety of the Kamala Harris “Brat Summer” period. My understanding is that the new, debugged version had been ready earlier, but ABC News PR was exceptionally sensitive about public perceptions around the model, speaking about it only cryptically in contrast to the transparency the site had been known for.

Meanwhile, I had a good year. Silver Bulletin received more support than I ever expected, and I was all over the media during my book tour. I’m not going to lie: after 10 years feeling jerked around by Disney, the whole sequence was pretty satisfying. But it also proved the point about a subscription-based model being the right call.

ABC finally fully shut down FiveThirtyEight in March 2025, 11 years after its debut at Disney. Eleven years is a long time in the media business, and the site covered one of the most tumultuous periods in American political history with its unique blend of analytics, brutal honesty and irreverence.

Disney refuses to negotiate with me

It would be nice if that work could be preserved for the public record. I don’t know what plans Disney has for FiveThirtyEight, if any. But I did approach Disney a year or two ago, through my agent, about acquiring the remaining IP. I’m probably the logical high bidder, though the value is rapidly depreciating as what’s left of the site falls into disrepair. At a minimum, we’d restore the archive, with prominent links to Silver Bulletin.

We were told to basically get lost: ABC was annoyed with my critical public comments about their management of FiveThirtyEight. It apparently wasn’t a long conversation, so I don’t have a lot more color to report than that.

Hanlon’s Razor states: Never attribute to malice that which can be adequately explained by stupidity. But honestly, I don’t know which explanation is better suited to ABC. During the second half of my tenure with Disney, it felt like they were putting almost literally zero effort into any decisions involving FiveThirtyEight (other than my being featured prominently in their election night coverage¹⁰).

The one good thing about the bitter ending is that it prevents the temptation to feel overly awash in nostalgia. Because FiveThirtyEight was always produced with a lot of care, including attention to copy editing and graphics, it tended to impress people as a more smoothly running operation than it actually was. Internally, there was always a lot of conflict: between bright but opinionated staffers, all of whom had slightly different ideas about what “data journalism” was, between staffers and their opinionated boss/founder, between the news cycle and our deadlines, and between pretty much everyone and Disney. The relationship with Disney wasn’t particularly heated, so there weren’t a lot of stories that would make for good movie scenes. But mostly we just felt neglected.

Disney HR did send me a literal Mickey Mouse plaque after I left, celebrating my 10 years as a “castmember”. There was one small problem, though: they misspelled my first name as “Nataniel”.

1. Some late-stage content published under abcnews.com URLs still survives.

2. Although I still own the IP to the original versions of the sports models, most of them were last worked on 5-10 years ago, so I’ve basically decided to do a complete refresh on our suite of sports models instead.

3. That’s probably about right. The average Silver Bulletin story takes something like 15 hours from start to finish, and I write faster than most people.

4. The Times is a different newsroom now, both more data-friendly and more pluralistic, but at the time, the main tensions were with the politics desk, which did not appreciate FiveThirtyEight’s implicit criticism of “horse-race” type coverage during what was ultimately a pretty boring 2012 election

5. In Bloomberg’s case, primarily the Bloomberg Terminal.

6. We’d also been dismissive of Bernie Sanders’s chances against Hillary Clinton, a “call” that was technically correct but maybe not correct in spirit given that Sanders won quite a few states in a way that foreshadowed Clinton’s weaknesses.

7. And we’d probably have wanted to hire a couple of staff devoted to things like pricing strategy and customer retention. Let’s call it $400K-$500K/year for two senior staffers. Still a pretty good return on investment.

8. Or maybe a cold Friday night in January. A good night to eat pizza at Lucali, in any event.

9. Disney retained a non-exclusive license to continue publishing the sports models (but not the politics models) in the condition they were in as of my departure. Instead, they fired the entire sports staff and stopped publishing them anyway.

10. ABC was also constantly begging us to replicate the New York Times “needle”. The needle is an exceptionally challenging engineering and statistical product; we had nowhere near the resources for it, and always found a way to turn them down.