Jeffrey Epstein, SEO Expert?

This post isn't for everyone. We now live in a post-Epstein world where we are left to our own devices to unearth the truth, crowdsourcing research and using non-DOJ tools to understand who knew what when.

The Epstein story and the ongoing saga of the release of the “files” are far from over.

Victims are justifiably unsatisfied with the US Department of Justice’s efforts.

The absence of empathy and indictments makes it clear that Justice is entirely lacking in the department in question.

Using the poorly constructed DOJ website (aka The Epstein Library, as they’ve labelled it) is onerous and clumsy. Their site is a UX embarrassment, given the importance of the information contained and what is possible with even moderately better off-the-shelf software.

Third-party search tools and Artificial Intelligence are simplifying but not perfecting the task for both journalists and individuals, arming them with the ability to sift through the 3.5 million documents, 180,000 images and 2,000 videos that currently constitute the bulk of what we call the Epstein Files.

There are obvious risks to trusting large language models with the task of fact-checking millions of sensitive documents.

ChatGPT struggles with Epstein-related questions because of the guardrails OpenAI has put in place for other good reasons.

Meanwhile, Elon’s xAI tool Grok is muddying the waters of earnest people’s research efforts by enabling the creation of fake redactions and AI-generated Epstein images that only make fact-checking harder.

But weren’t LLMs built for this exact sort of task, using expansive data sets to provide some form of answer?

AI and Journalism

The New York Times has been explicit that it has built its own internal tool to allow a team of journalists to research stories, enabling them to dig deeper into the Epstein files than the DOJ site can.

“The first thing we always try to do is make things searchable. But here we also needed ways for reporters to get at the things that weren’t easy targets for search. One way we did that was by leveraging something called “semantic search,” which lets reporters search for concepts and find matching text even if the exact language isn’t in the document. We also built an A.I.-powered tagging and categorization tool to bucket the documents by type and add labels for things that we thought may be useful indicators of newsworthiness.” - Andrew Chavez, NYT

The Times is not alone.

The Miami Herald is using Google’s PinPoint tool for journalists to simplify research.

At the BBC, Ravin Sampat explained that they extracted the released documents into their own database and built a custom search system because the DOJ site was too difficult to navigate

“This really comes down to searching terms and being clever. You can’t rely on Boolean search here – it doesn’t exist – so you have to be strategic about keywords. We’ve found things by being a bit unconventional,” said Sampat. For example, instead of searching for generic words like ‘investment’ between business leaders, we look for terms they would use less frequently to help narrow the search. That’s essentially how we’ve been approaching it.” - Ravin Sampat, BBC

Epstein Ninja

A software developer from Atlanta built the Espstein Ninja, a “Smart search engine for the Epstein files from the DOJ built on a RAG pipeline that indexes and makes them searchable via natural language.”

The Epstein Ninja doesn’t appear to include all of the files, but it does make searching them a lot more like other search experiences we’re accustomed to.

The file’s data is aggregated, summarized, and rendered in a way similar to how Google or Bing might in AI summaries.

Jmail.World

Riley Walz, who Wired describes as a “prankster,” and developer Luke Igel built a web tool called Jmail, a recreation of Epstein’s email.

Jmail is a very Gmail-like application that looks and behaves like you’re viewing Epstein’s emails in his jeevacation@gmail account, in his actual inbox.

Beyond the ease of use, Jmail also lets you query the documents and emails for things you might search for in your own Gmail account.

Here are some examples.

It kinda looks like Jeffrey Epstein ordered a 23andMe kit. It is unclear if he sent it in.

Curious about Epstein’s Spotify recommendations. Sure, weird, but okay, you can do that too.

Want to know who reached out to Epstein to connect on LinkedIn? No problem.

Jmail now includes a “suite” of Google Workspace-like products, and presents them in the way that Epstein might have used and experienced them.

JPhotos is a Google Photos clone containing all the photos released with the documents.

JDrive includes PDFs and other Epstein-related files released in Google Drive format.

Jeffrey Epstein, SEO Expert?

A weird footnote on the absolutely abhorrent story of Jeffrey Epstein is how he hired software engineers to game Google search to bury negative press after his 2008 conviction for sex trafficking minors.

From Forbes:

“Among the roughly three million Epstein emails released by the Department of Justice last month are exchanges with a little-known search specialist called Tyler Shears. Shears was introduced to Epstein in 2013 by a hacker Pablos Holman, who’d struck up a relationship with Epstein over a shared interest in hacking culture and videogames as educational tools. Neither Shears nor Holman responded to requests for comment.

Epstein tasked Shears with basic reputation laundering. That work included placing favorable stories in major publications (including Forbes and Huffington Post) that showcased his philanthropic work. Once published, Shears exploited the search engine’s ranking signals to push them higher in its results, demoting negative stories tied to Epstein’s name.

This wasn’t the first time Epstein paid to have his online reputation laundered. In 2010, he discussed a similar, but more aggressive project with Al Seckel, an optical illusion enthusiast and the husband of Ghislaine Maxwell’s sister Isabel. For $20,000, Seckel claimed he and Holman could assemble a team to hack websites hosting negative stories and “cause all that crap to disappear.” He also promised to use bots to manipulate Google results and to edit Epstein’s Wikipedia page to remove his mugshot and add content about his philanthropic work.

But Epstein was unimpressed. “Why don’t you stop and look, the results are marginal,” he wrote to Seckel in December 2010. “You have made every excuse you can make, it’s tedious and tiresome… Al, you are talented. You should be careful.”

Seckel pushed back, reminding Epstein just how awful his reputation had become., “Your case was the ‘worst’ that anyone has ever come across,” he wrote. “To accomplish the amount that we did was a miracle of a semi-religious nature.”

Epstein moved on to work with Shears in 2013. In 2015, Seckel was found dead at the bottom of a cliff near his home in southern France, following allegations that he had defrauded buyers and sellers of rare books, either by never paying for them or never giving them the book.”

I’m not telling you that Jeffrey Epstein was any sort of search engine optimization expert. But he did understand how important his own search results were to his public reputation, probably more than many businesses do today.

If you really want to, you can read Epstein’s emails with and about Tyler Shears and Al Seckel in Jmail.