I didn't go down the route of LLMs for the clean up, as you're getting into scale and context issues with larger datasets.
I got into semantic similarity networks for this use case. You can do efficient pairwise matching with Annoy, set a cutoff threshold, and your isolated subgraphs are merger candidates.
I wrapped up my code in a little library if you're into this sort of thing.
github.com/specialprocedures/semnet
https://github.com/maxandrews/Epstein-doc-explorer/blob/83ee...
Americans..?
It is very funny that the "unaccountability shield" stops at the US border, though, so it's taken out Prince Andrew.
there's Obama in there only because related to Trump via "invited to White House lunch", now he is part of septic tank?
The Bill Clinton entity is interesting.
> 2009: Bill Clinton discontinued association with Jeffrey Epstein
> 2010: Jeffrey Epstein provided flights on jets to Bill Clinton
> 2010-2011: Jeffrey Epstein traveled via private aircraft with Bill Clinton
> 2011: Ghislaine Maxwell piloted helicopter for Bill Clinton
> 2014: Bill Clinton alleged presence at sex parties
> 2015: Bill Clinton distanced relationship from Jeffrey Epstein
Wasn't very good at discontinuing the relationship it seems.
Guess there is precedent for him lying about sexual activities though.
I think a sentiment analysis between the friendliness and social meetups between Epstein and other individuals would be useful.
Who were his friends after 2008 when he was first convicted?
Those who were still friends with him after 2008 were in on it or guilty by association, if not legally, socially.
Friends like Reid Hoffman and Larry Summers...
> From: Reid Hoffman
> Sent: 7/6/2015 5:04:31 PM
> To: jeffrey E. [jeeyacation@gmail.com]
> Subject: RE: ICYMI
> slow progress.
> planning to see you in August.
> Hope you're well.
Larry Summers has too many to list. Doesn't look good though digging through them.
This is of course ontop of the 2006 Florida prostitution charge though.
But point being those people that were friends with him had to know. Whether it was socially acceptable by the elite because the public wasn't aware isn't very relevant.
Keep in mind that those summaries are AI-generated. There's gonna be a lot of confabulating in there.
Care to dispute the summaries using the sources?
I mean here's a weird example. Searching Donald Trump there's the headline
(1994-06 Wexner Mansion NYC)
Donald Trump forced to perform oral sex and physically abused 13-year-old female plaintiff and 12-year-old female.
A declaration from Tiffany Doe (pseudonym) testifying that she witnessed Jeffrey Epstein and Donald Trump sexually abuse a 13-year-old girl and other minors during parties from 1990-2000 in New York City.
7. It was at these series of parties that I personally witnessed the Plaintiff being forced to perform various sexual acts with Donald J. Trump and Mr. Epstein. Both Mr. Trump and Mr. Epstein were advised that she was 13 years old.
So far the paragraph summaries seem to be accurate in my poking around but the headlines are mixing ordering and have other weird errors like this. Anyways, always good to check when things are as serious as this...
Here's the link: https://drive.google.com/file/d/11KzAOYCjxwEhnyrsiBpKM8OGBJp...
13. I personally witnessed Mr. Epstein physically threaten the life and well-being of the Plaintiff if she ever revealed the details of the physical and sexual abuse she suffered at the hands of Mr. Epstein or any of his guests.
14. I personally witnessed Defendant Trump telling the Plaintiff that she shouldn't ever say anything if she didn't want to disappear like the 12-year-old female Maria, and that he was capable of having her whole family killed.
15. After leaving the employment of Mr. Epstein in the year 2000, I was personally threatened by Mr. Epstein that I would be killed and my family killed as well if I ever disclosed any of the physical and sexual abuse of minor females that I had personally witnessed by Mr. Epstein or any of his guests.
To that, I think the answer is an unambiguous "yes". If someone who is rich, well connected, and successfully covering up heinous crimes at a large scale, then yes, I believe a person threatened has a reasonable belief that such a threat is credible.
Seriously, we are talking about a world famous pedo who was pimping out girls to presidents, royals, billionaires, and when he was finally convicted he was only charged with prostitution and got a extremely light sentence that everyone now calls a "sweetheart deal." So years after does a witness have a credible belief that such a man can post a significant threat to her and her family?
Do you seriously believe that no person has any reason to fear Epstein? I find that laughable considering how much conspiracy there is about him being murdered and how the accusations are towards varying high profile people. You're trying to say that Epstein is a puppy dog that's all bark and no bite?
I agree, nothing is proven but it's absolutely laughable to claim that such a threat is not credible.
Why are you defending a pedo?
Not really. After Epstein got convicted in 2008, he set about trying to rehabilitate his image, to be seen as a philanthropist, a patron of science, and (perversely) a supporter of women and girls. He hired reputation management consultants to help carry out the project, with one of the models they used being Mike Milken (of Drexel infamy), who ultimately secured a pardon from Trump. A lot of prominent people, knowingly or not, served as "useful idiots" in this project, often due to financial incentives that were not wholly selfish. For example, the MIT and Harvard scientists whose labs and research he funded, and who visited his island for science-themed retreats. Clinton was probably another of Epstein's useful idiots, being lured in through his Clinton Global Initiative and the promise that Epstein, with his ample wealth, could help greatly expand it.
I should add that at least one of them, Marvin Minsky, was accused by name by the late Virginia Giuffre.
Just like here you could get a timeline of key events, a graph of connected entities, links to original documents.
Newsrooms might already do this internally idk.
This code might work as a foundation. I love that it's RDF.
Tech Company: At long last, we have created the Torment Nexus from classic sci-fi novel Don't Create The Torment Nexus
These general data models start to become useful and interesting at around a trillion edges, give or take an order of magnitude. A mature graph model would be at least a few orders of magnitude larger, even if you aggressively curated what went into it. This is a simple consequence of the cardinality of the different kinds of entities that are included in most useful models.
No system described in open source can get anywhere close to even the base case of a trillion edges. They will suffer serious scaling and performance issues long before they get to that point. It is a famously non-trivial computer science problem and much of the serious R&D was not done in public historically.
This is why you only see toy or narrowly focused graph data models instead of a giant graph of All The Things. It would be cool to have something like this but that entails some hardcore deep tech R&D.
Using this press release as an example, if you pay attention to the details you'll notice that this graph has an anomalously low degree. That is, the graph is very weakly connected, lots of nodes and barely any edges. Typical graph data models have much higher connectivity than this. For example, the classic Graph500 benchmark uses an average degree of 16 to measure scale-out performance.
So why did they nerf the graph connectivity? One of the most fundamental challenges in scaling graphs is optimally cutting them into shards. Unlike most data models, no matter how you cut up the graph some edges will always span multiple shards, which becomes a nasty consistency problem in scale-out systems. Scaling this becomes exponentially harder the more highly connected the graph. So basically, they defined away the problem that makes graphs difficult to scale. They used a graph so weakly connected that they could kinda sorta make it work on a thousand(!) machines even though it is not representative of most real-world graph data models.
To plug my project, I've wrapped the SuiteSparse GraphBLAS library in a postgres extension [1] that fluidly blends algebraic graph theory with the relational model, the main flow is to use sql to structure complex queries for starting points, and then use the graphblas to flow through the graph to the endpoints, then joining back to tables to get the relevant metadata. On cheap hetzner hardware (amd epyc 64 core) we've achieved 7 billion edges per second BFS over the largest graphs in the suitesparse collection (~10B edges). With our cuda support we hope to push that kind of performance into graphs with trillions of edges.
That is a wild claim. Perhaps for some very specific definition of "useful and interesting"? This dataset is already interesting (hard to say whether it's useful) at a much tinier scale.
Almost every non-trivial graph data model about the world is a graph of human relationships in the population. If not directly then by proxy. Population scale human relationship graphs commonly pencil out at roughly 1T edges, a function of the population size. It is also typically the highest cardinality entity. Even the purpose isn’t a human relationship graph, they all tend to have one tacitly embedded with the scale implied.
If you restrict the set of human entities, you either end up with big holes in the graph or it is a graph that is not generally interesting (like one limited to company employees).
The OP was talking about generalizing this to a graph of people, places, events, and organizations, which always has this property.
It is similar to the phenomenon that a vast number of seemingly unrelated statistics are almost perfectly correlated with GDP.
Aren't LLMs something like this?
Could you point us to any public research on this issue? Or the history of the proprietary research? Just the names might help - maybe there are news articles, it's a section in someone's book, etc.
https://developer.nytimes.com/docs/semantic-api-product/1/ov...
The Guardian has similar:
https://open-platform.theguardian.com/documentation/tag
Either or both could be an interesting starting point for something like that. I tried to find something for the BBC and was surprised they didn’t have anything. I would have figured public media would have been a great resource for this.
That said, some networks of shorter paths than 6 are interesting. Right now, there's a 1:1 direct path from these documents to a bunch of people with an interest in confounding what evidentiary value they have in justice processes. That's more interesting to me, than what the documents say right now.
I have a 4090 and 32 GB of RAM and this thing is chugging at like 2 FPS, with the UI being completely unresponsive.
That is not an exhaustive list. But people who want things to improve should also shut down their ability to confect scandals or distractions, like the "Obama tan suit" controversy. Once Americans have a reasonable selection of non-insane non-compromised candidates, things may get better. The election of Mamdani is a good start in that direction, because all the other (D) candidates were horribly compromised.
[1] - https://www.youtube.com/watch?v=XcebHfZ2LbU [video][4 mins]
We need to expand on network mapping with data and areas as well.
### Purposes Behind a "Manufactured" Narrative
- *Distraction From Systemic Issues* The media spectacle around Epstein's crimes and network shifts public attention toward lurid details, celebrity involvement, and political intrigue, while potentially obscuring broader questions of elite accountability, institutional corruption, or failures in law enforcement and intelligence oversight. This phenomenon is typical in high-profile scandals: rather than fostering reform, they may act as pressure-release valves, venting public outrage in "safe" directions away from actionable reform or scrutiny of underlying systems.[1][2]
- *Polarization and Conspiratorial Thinking* The Epstein case has fueled intense binary narratives — "elites vs. the people," "deep state cover-up," and similar populist themes. QAnon, MAGA circles, and conspiracy-oriented media have recast Epstein as evidence of a secretive cabal undermining America, intentionally stoking distrust of government, media, and political adversaries. This fragmentation of public trust can benefit actors seeking to create division, distract from policy failures, or delegitimize political opponents.[3][1]
- *Reinforcement of Powerlessness and Fatalism* The widespread belief that Epstein's death was the result of elite conspiracy (i.e., "he knew too much," and "they'll never let the truth out") can breed social fatalism and apathy — the sense that ordinary citizens are powerless against entrenched interests, which can reduce civic engagement or demands for accountability.[2][1]
- *Information Warfare and Blackmail Speculation* Persistent rumors about espionage, blackmail, and covert intelligence operations surrounding Epstein (Israeli ties, spy theories, etc.) serve to keep the public focused on speculation, preventing consensus and muddying facts. This cacophony can be exploited by political or intelligence actors seeking to obscure real mechanisms of elite control or leverage.[4][5]
### Target Audience: American Public
- *General Population (Non-Government Workers)* The intended psychological effect is a mix of outrage, curiosity, and demoralization, wrapped in a sensation of being on the outside of elite secrets. The public is kept vigilant about "pedo networks" and government corruption but largely passive in meaningful action, with most energies channeled into online speculation and politically polarized media.[1][2][3]
### Patterns & Implications
- The Epstein narrative quickly became a "choose-your-own-adventure" for conspiracists, activists, and mainstream skeptics, reinforcing pre-existing suspicions and worldviews.[3][1] - The framing allows elites implicated by proximity to claim victimhood in a supposed media witch-hunt, while others use it to fuel anti-establishment politics.[2][3]
### Conclusion
Considering the story as a manufactured psyop, its chief functions seem to be distraction, polarization, and demoralization of the public, as well as muddying the waters of elite accountability, with the target audience being ordinary Americans outside the machinery of state and intelligence.[5][4][1][2][3]
[1](https://www.lemonde.fr/en/opinion/article/2025/07/20/the-eps...) [2](https://www.nytimes.com/2025/07/16/opinion/epstein-trump-sca...) [3](https://www.newyorker.com/news/fault-lines/donald-trump-jeff...) [4](https://www.trtworld.com/article/16616743) [5](https://www.businessinsider.com/jeffrey-epstein-spy-epstein-...) [6](https://www.youtube.com/watch?v=1Et4ujSsluA) [7](https://www.vanityfair.com/news/2019/08/curious-sociopathy-o...) [8](https://www.nytimes.com/interactive/2021/11/23/us/document-e...) [9](https://therapynearme.com.au/mental-health-blog/f/psychoanal...) [10](https://www.vox.com/2018/12/3/18116351/jeffrey-epstein-case-...) [11](https://www.pbs.org/newshour/politics/how-trump-spent-years-...) [12](https://www.nbcnews.com/news/us-news/trove-newly-released-je...) [13](https://www.politico.com/news/magazine/2021/05/14/jeffrey-ep...) [14](https://www.npr.org/2025/07/25/nx-s1-5478620/jeffrey-epstein...) [15](https://www.npr.org/2025/05/30/nx-s1-5407856/conspiracy-theo...) [16](https://www.pbs.org/newshour/nation/new-records-detail-how-e...) [17](https://www.bbc.com/news/articles/cp8j3e5g74no) [18](https://www.vpm.org/npr-news/npr-news/2025-05-30/how-conspir...) [19](https://www.bbc.com/news/articles/c20r07dg6kro) [20](https://www.youtube.com/watch?v=ZsykRb16QRU)
wnevets•2mo ago
analog31•2mo ago