Epstein Data Wiki

Joshua Cooper Ramo

Properties2
tagsperson, kissinger-associates, china, jeffrey-epstein, correspondence, introductions, santa-fe
aliasesRamo, Joshua Cooper, Josh Ramo, Joshua Ramo, Joshua C. Ramo, Josh Cooper Ramo, Mr. Ramo

62 min read

Named on the documents as”Joshua Cooper Ramo” in the sender field of his own messages12; “Joshua C. Ramo” in the signature block of his own assistant at Kissinger Associates, Inc.3; “Josh Ramo”, “Josh Cooper Ramo” and “Joshua Ramo” on Jeffrey Epstein’s daily schedule sheets45; “Joshua” and “J” in his own sign-offs67; “josh” and “joshua copper ramo”, with a double p, in Epstein’s own writing89
The title the documents give him”Assistant to Vice Chairman, Joshua C. Ramo / Kissinger Associates, Inc.”, printed in his own assistant’s signature block. Read at the image at 300 dpi3. Epstein described him to third parties as “kissinger right hand” and as “the partner of henry kissing=r in kissinger associates”; both descriptions are Epstein’s, not his own910
Who introduced themJoi Ito, by e-mail on 13 July 2013: “I want you two to meet.”6
What the correspondence is aboutMeetings — breakfasts, lunches and appointments at Epstein’s New York house and in Santa Fe; science and network-theory reading; introductions in both directions; and, from late 2014, Epstein’s advice to him on his own compensation and on the terms of a proposed China company, which the correspondence names only by the attachment KCN_Proposal.docx. An unheaded sheet in the release sets out the structure of a “Kissinger China NewCo”; this page does not assert that it is that attachment111213141516
Documents he wrote, cited here, run from14 July 2013, his first reply to the introduction, to 31 July 2018, a two-line note about Santa Fe617
Response to the correspondence becoming publicNone located. See Response

Joshua Cooper Ramo is an American writer and business executive who, between July 2013 and August 2018, corresponded with Jeffrey Epstein several hundred times. The signature block of his own assistant, on a sheet in the release, identifies him as Vice Chairman of Kissinger Associates, Inc.3 He is on this page because the documents record him acting: writing, introducing, asking and being asked.

The correspondence began five years after Epstein’s Florida conviction and ran for five years after that. Its ordinary content is scheduling. Its substance is a two-way trade in access: Epstein offered him to a Rothschild bank, to a Harvard professor, to a New York Times correspondent and to a Qatari businessman; he introduced Epstein to a Princeton professor, thanked Epstein for connecting him to a former Israeli prime minister, and, in October 2015, took up an introduction Epstein made to an Italian oil executive. From November 2014 the traffic turns personal in a different way: he asked Epstein for advice on his own pay, and Epstein wrote him a negotiating script. By 2016 the subject is the structure of a company, and an unheaded proposal sheet in the release names him as Chairman and CEO of a “Kissinger China NewCo”.

Nothing on this page is a finding of wrongdoing, and no document cited here charges him with anything. Where a role or a characterisation is attributed to him, the page says who attributed it.

Identity and account numbers, telephone and fax numbers, e-mail addresses, residential addresses and dates of birth printed on the sheets cited here are not reproduced on this page.

What was searched, and what each search returned

Every figure below was measured against the FTS5 index of the full-text corpus on 12 September 2026, counting rowid first and resolving identifiers in batches. Document counts are distinct identifiers across every production in the index, EFTA, HOUSE_OVERSIGHT and DOJ-OGR, and were not scoped by documents.dataset. Page counts are pages.

querypagesdocumentsoutside the full-name base
"Joshua Cooper Ramo"1,2931,247— (the base)
"Josh Ramo"537523+482
"Joshua Ramo"7272+62
"Mr. Ramo"4646+29
"Joshua C Ramo"4141+21
"Ramo, Joshua"14140
"J Ramo"22+2
"w/josh"353328+298
bare "Ramo"2,3832,232—
"Ramos"1,025727—

The full name reaches about 63 per cent of him. A page built on "Joshua Cooper Ramo" alone would miss 746 documents that this page counts, and the single largest contributor is not damage at all but the ordinary short form Josh Ramo, which is what Epstein’s office called him on every schedule sheet.

Two of the rows above are traps rather than populations, and both are worked below: "w/josh" names at least five different men, and bare "Ramo" is not a name search at all.

"w/josh" is schedule shorthand, and on this corpus it does NOT resolve to one man

"w/josh" returns 328 documents. FTS5 is punctuation-blind, so the phrase matches the adjacent token pair w josh however the sheet punctuates it. The test applied here was to read the token that stands immediately after the shorthand on the sheet itself, in every one of the 328 documents. The result:

token immediately after w/joshdocuments
a Ramo form (Ramo, Ram, Rant, Rant°, Ranu., Rana, Rama, Ra) or Cooper232
Harris or a damaged form of it (Harri, Hanis, Hams, Harr, Hath, Hata, Han, Hants)93
Black or Blacks41
Koppel1
Carlo1
no surname on the line16

Documents carrying more than one appointment appear in more than one row; the row totals therefore exceed 328. Harris alone is 83 and Black alone is 40; the figures given are for the whole of the form each label names, which is the form the scoping was actually applied to. Eighty documents carry w/josh only beside another man’s surname. The sheets name a Josh Harris, a Josh Black, a “Josh Koppel at Scrollmotion” and a “Josh Carlo”; one sheet carries w/Josh Harris and w/Josh Ramo as separate appointments on the same day.18

So the shorthand is not an identification, and this page does not treat it as one. Only the 232 documents in which a Ramo form stands beside it are counted, and of those only 25 are not already reached by one of the named forms. Ninety-six of the 328 fall outside the scoped set; 82 of those 96 are reached by no named form either, so a page that had taken all 328 would have imported 82 documents on the strength of another man’s appointment.

The damaged forms in the first row were each read before being admitted. The extraction of one day sheet carries w/Josh Rant° where the appointment below it (“11-11:30am Appt w/Bruce Galloway”) matches, line for line, a sheet whose image prints “6:30am Breakfast w/Josh Ramo” above the same Galloway line.19 Seven of the 232 carry no whole-word token Ramo anywhere, because the surname survives only in a damaged form; they are in the population on the strength of the line they stand on and nothing else.

Decomposing the 2,232

Bare "Ramo" returns 2,232 documents. Which of them count as “outside the name searches” depends on one choice, and the choice is worth exactly one document, so it is named here rather than buried. Take the union as the four named forms plus all 328 shorthand documents and the residual is 396; take it as the four named forms plus only the 232 scoped shorthand documents and the residual is 397. The document in the difference is https://epstein-data.com/EFTA01950302, a five-page day sheet whose printed page 2 carries w/Josh Harris, which is why the scoping drops it, and whose printed page 3 carries a damaged w/Josh Ramo, which is why it belongs to him. It was read, it is his, and it is in this page’s population.

The residual read for this page is therefore the 396. The arithmetic that closes at 1,836 + 396 = 2,232 proves a partition and nothing about membership: it closes just as well if every one of the 396 is a stranger. Membership was tested separately, by reading, and a further 7 documents reach the population without appearing in the bare-surname search at all, because on those sheets the surname survives only in a damaged form.

All 504 occurrences of the whole-word token ramo in those 396 documents were read, one line at a time. The unit is the FTS5 token: a match bounded by non-alphanumeric characters, which is why RAMO°, Ramo. and Ramo-CC are hits and Tramo and paramount are not. The decomposition:

what the token isdocuments
him — a forename, an initial or the schedule shorthand stands beside the surname on the sheet150
a different named person26
a named mis-reading of another word, or of a person’s surname45
unnamed extraction noise in a long financial or institutional run175

Those four figures are a reading and not a query, and a reader cannot re-derive them. The 396 documents, and the 504 tokens standing inside them, are reproducible from the searches printed above and were reproduced independently on 12 September 2026 and again the same day against this text. The sort into four buckets was made by reading each of the 504 lines; this page does not publish the four lists; so no reader can check which bucket a given document is in, or that the reading landed at 150 / 26 / 45 / 175. The same warning covers the itemisations of the 26 and of the 45 set out below, which are figures of the same kind, made by the same reading, and not queries either.

One mechanical check can be offered on the bucket that carries the weight, and it is printed in full so that it can be run. Over the 396, take every line carrying the whole-word token ramo, and count a document as his if any such line also carries, anywhere on it, the letters osh (the surviving core of Josh, Joshua and the shorthand w/Josh under the OCR damage in this release), or Cooper, or copper. That rule returns 136 documents against the 150 published here. An earlier version of this page put the same reconstruction at 152; that figure is not what any implementation of the rule it stood beside returns, two independent implementations of the looser wording it was printed with returned 136, and the 152 is withdrawn. The rule and the reading differ in both directions. The rule admits two documents the reading excludes, the two Roberta Cooper Ramo sheets whose ramo line carries Cooper;20 and it misses lines the reading admitted, among them >>john ramo said i would like barabasi, isn't he one of yours at https://epstein-data.com/EFTA00663533, printed p.1, and a To: value truncated to er Ramo at https://epstein-data.com/EFTA02089882, printed p.1, on a sheet whose body reads “was just told that Josh will be attending the dinner Jeffrey is having”. The 150 is a reading, the 136 is what a reader gets, and the difference between them is the reading. Two of the four descriptions below were wrong when this page was first written, and what follows says which.

The 26 were given as: Mark Tramo (10), a Ramon (7), a Ramos (3), Roberta Cooper Ramo, an Albuquerque attorney (3) and Abdullah Ramo Pazara (3). The Ramos figure is wrong. Reading the ramo line of every document in the 396 that the first pass did not attribute to him, nine of them carry a person surnamed Ramos whose terminal s the extraction has dropped, not three: seven are sheets of one bank’s internal correspondence of 2017 and 2019 in which a recipient’s surname prints as Ramos and the extraction prints Ramo, and two are copies of one forwarded message of 18 May 2009 whose sender line prints Ramo above a line printing the same surname in full.21 The Ramon figure is wrong too, and the footnote written to correct the Ramos count asserted it and got it wrong again; that second error is recorded here rather than quietly replaced. The correspondent whose forename the extraction damages into the bare token stands in eight documents of the residual, not seven: in four of them the forename is broken across lines, leaving Ramo alone on a line with its last letter on the line below, and in four more it carries quoted-printable damage and prints as Ramo= or ramo=. Four and four.22 Seven documents more than the itemisation allows for therefore have to come out of one of the other three buckets, and since the 26 and the 45 are each itemised down to the last document as published, the boundary between the 26 and the 175 is not where this page put it. The Roberta documents are the dangerous ones, and they caught this page out: the string Cooper Ramo is not unique to this subject, and the first classification run admitted two of the three to his population because it tested for him before testing for anyone else. A rule that tests membership in one direction only will absorb a homograph every time; the test has to run the other way first.20 Mark Tramo enters the bare-surname population because quoted-printable damage on some copies renders his surname as =ramo, and a barred display name on others leaves arc ramo; both tokenise as ramo.23 "Ramos" is a separate 727-document population that shares no document with the full-name base, and 23 with the residual; none of those 23 is him. That cross-check does not bound the problem it looks as though it bounds: two of the nine documents above, https://epstein-data.com/EFTA01371313 and https://epstein-data.com/EFTA01460205, are not in the 727 at all, because the only token the extraction leaves on those sheets is Ramo.21

The 45 named mis-readings are, in order of size: RATING: on brokerage statements (12), Park Avenue on a bank-statement letterhead (9), DALBAR on an award line (7), ramp (6), RAMCO GERSHENSON (5), RADIO on a jail property inventory (4), and two others. The label on that bucket was wrong when this page was first written and is corrected in the table above: not every mis-reading in the residual is a mis-reading of a word. Five documents in the 396 carry a mis-reading of a person’s surname on the ramo line, and each of the three forms was read at the image for this correction: BRENDA S. RAMO. for Brenda S. Fulmer, in the attorney column of a Searcy Denney letterhead; Paul, Weiss, RAMO, Wharton & Garrison LLP for Rifkind, in a partner’s own signature block; and Ramon A. Ramo for Ramon A. Rasco, three times, in the attorney column of a Podhurst Orseck letterhead, the same letterhead whose image prints “Jay P. Lefkowitz, P.C.” where the extraction prints “Jay P. Leficowitz, P.C.”.24 "Ramon A Rasco" returns 40 documents in this corpus and "BRENDA S FULMER" nine, so in each case the correct form stands in the index and the damaged one is this extraction’s. The six sub-buckets itemised above account for 43 of the 45, and five documents will not fit in the two that remain; which of the five the reading put in the 45 and which in the 175 was not recorded, which is one more instance of what the paragraph above says a reader cannot check. The remaining 175 were described in an earlier version of this page as illegible runs in scanned financial and institutional documents in which the four letters occur inside garbage, of which “none carries anything identifying a person”. That description does not hold and is withdrawn. Each of the nine documents in the paragraph above carries a named person, and one of them is not an illegible scan at all: EFTA01363694, read at the image at 400 dpi for this correction, is a clean, fully legible internal bank e-mail of 10 July 2017 with a corporate signature block. Two further documents in the residual that are not his also carry a named person on the ramo line: EFTA00053963 at printed p.767, on a Bureau of Prisons activity log, and EFTA01070533 at printed p.3, a party to a reported appellate decision.21 Most of the 175 are what the old description said they were, and these corrections change what the buckets are called and not the membership figure, which the mechanical check printed above tests at 136 against 150 without re-deriving it; but the blanket sentence was false and a reader should not have been asked to take it. As a second, independent membership test, the full text of every one of the 175 was searched for josh, joshua, kissinger or cooper ramo: ten matched, of which two proved to be him on reading and were moved into the 150; the other eight match those words elsewhere in documents hundreds of pages long.

The population this page was written from is 1,993 documents, the four named forms, plus the 232 scoped shorthand documents, plus the 150 from the residual, plus EFTA01950302. "Joshua C Ramo", "J Ramo" and "Ramo, Joshua" were then re-run against that set as a check: all 41, 2 and 14 documents fall inside it.

Does the record show him acting?

act_rate.py --selftest printed PASS before any figure below was taken, and again on 12 September 2026 when the first two rows were re-run. Each figure is over the whole population named beside it, not a sample, and the name key is Ramo, the bare surname, matched case-insensitively inside the sender field of the document’s own first sheet. A figure quoted without its key is not reproducible, and the same query on a different key returns different numbers.25

populationdocumentssender fieldof which the key heads the fieldaddressed onlyheader-less pressparser refused
"Joshua Cooper Ramo" (the base)1,247491 (39%)356506 (41%)041
the four named forms1,817492 (27%)357534 (29%)095
the full population used here1,993494 (25%)358534 (27%)0109

On the base, four documents in five are direct correspondence in one direction or the other. That rate belongs to the base and to nothing else: the 746 documents the widening adds are overwhelmingly Epstein’s office’s own schedule sheets, on which he is named rather than writing, and the rate falls accordingly. Both figures are true of their own populations and neither is true of the other.

The refusals are not negatives. On 109 documents the parser declined to map a header layout and returned no verdict; those are counted as neither acts nor non-acts, so every act figure above is a floor.

The press figure is zero on all three populations, which answers the obvious objection about a public figure: his mass in this release is not coverage. The tool reaches that figure by testing a header-less sheet for the marks of a clipping, a filing or a transcript, and on the full population it classes 37 header-less documents as authored and none as press. The control run the same day, on "Donald Trump" over 2,898 documents and on the key Trump, returns 1 sender-field act, 3 addressed, 270 header-less authored and 37 press, a shape nothing like this one. The key matters and the control is unreproducible without it: the same query on the key Donald Trump returns 0, 2, 158 and 22.25

All 37 of his header-less documents were then opened, because the rule that counts them as acts is the rule that manufactures a celebrity’s. None is a clipping, and not one of the 37 is a message he sent. Thirty-one are Epstein’s office’s own daily schedule sheets, on which his name stands near the top and the author is Lesley Groff. Two are sheets of the text of his own book, The Seventh Sense: Power, Fortune and Survival in the Age of Networks, sitting in Epstein’s files. One is a contact record whose fields run “Full Name: Joshua Cooper Ramo / Last Name: Ramo / Middle Name: Cooper / First Name: Joshua / Company: Kissinger Associates”, which is a record about him and not one he wrote. The last three are neither his nor a schedule: two copies of one message of 4 December 2013 subject-lined “Josh Ramo is here!”, which carries To: and Subject: lines but no sender label, and a guest list headed “Paris list – updated” on which “Josh Ramo” is one line in a column of names. So the header-less figure over-counts him by an order of magnitude and the sender-field figure is the one to quote, which is why this page quotes it.26

The introduction, July 2013

The relationship starts with a third person. On 13 July 2013 Joi Ito wrote to both men:

I want you two to meet. I was hoping to get to Santa Fe with both of you sometime, but August (the month that Jeffrey’s in Santa Fe) is bankrupt to me, but I think Joshua may be going. If you’re both there, you should connect.

The next day he answered: “Jeffrey, / A pleasure to be connected. I will be in Santa Fe often in August. Otherwise in Beijing or New York. Would be great to meet up. Joi, thanks for the suggestion. / Best, / Joshua”.6 Within a week Epstein’s assistant Lesley Groff was arranging breakfast at Epstein’s New York house for Sunday 28 July; his one-word agreement and her message setting the address stand on the same sheet, read from the text layer.27 A Santa Fe lunch followed on 13 August 2013, and the extraction of his message the next day thanks Epstein for the invitation. That sheet was read from the text layer only and nothing from it is quoted here.28

What Epstein did with him

Five weeks after they met, Epstein began offering him to other people.

On 23 August 2013 the Harvard mathematical biologist Martin Nowak wrote to Epstein: “could you please send me the contact for josh who work with kissinger / what was his name? / might be cool to work on the china-US “game"". Epstein forwarded it. The reply came back the same day: “Thanks. / Will follow up.”8

On 2 September 2013 Epstein wrote to Ariane de Rothschild and to Olivier Colom of the Edmond de Rothschild group:

ehud barak on the 22? lunch? breakfast, also joshua copper ramo. kissinger right hand? hope you had a good vacation. time to get to work

That sheet’s Subject: field is a printed label with no value.9

The offer was taken up. On 2 January 2014 Epstein asked Colom who else the bank wanted to meet, and Colom answered: “But I think in addition to Reid: definitely Summers, Staley, maybe Cooper Ramo (as Asia should be a priority for us).”29 Two days later Colom sent back a numbered list of six names, introduced by the words “told me (in the order of interest for her):”, with “6) Cooper Ramo” at the foot of it. The person whose order it is is not identified on the sheet and is not identified here.30 Six days after that, Epstein’s schedule for Friday 10 January 2014 reads “3:30-5:30pm Appt w/Josh Ramo and Ariane Rothschild.”5

On 24 September 2013 Epstein invited the New York Times financial correspondent Landon Thomas Jr. to meet “interesting scientists” at his house. Thomas asked who. Epstein listed four people; the first was “josh cooper ramo. ( kissinger guy )”.31

On 29 April 2017, answering a Qatari correspondent who had asked whether he knew anyone at Starbucks, Epstein wrote a single sentence. The production copy carries quoted-printable damage on the image, which is not repaired here:

my very close friend JOSH COOPER -RAMO. =C2� sits on board , he is the partner of henry kissing=r in kissinger associates.

10

“Kissinger right hand”, “the partner of henry kissinger” and “my very close friend” are Epstein’s characterisations of him, written to third parties, and no document read for this page records him adopting any of the three. The test: every document in the population whose own sender field carries his surname was searched for vice chair, right hand and partner of. right hand and partner of return nothing. vice chair returns one document, and on that sheet the words are his assistant’s signature block quoted back inside a chain, not a description he wrote of himself.32 The search set cannot contain the sheets this page quotes the three phrases from, because all three were written by Epstein and the set is restricted to documents whose own sender field carries this subject’s surname; and the strings searched are the ones a man adopting the role would write, not the ones this page prints. The one title he is given in his own office’s words is the one in the infobox.

What he did with Epstein

The traffic ran the other way too.

On 14 November 2013, under the subject “Barak”, he wrote: “Terrific, terrific discussion with Ehud last night. Many thanks for connecting us. / Hope you are well. Nice seeing you by remote from Joi’s. / Joshua”.1 The connection had been made in Epstein’s words two months earlier, “i think a meeting with ehud barak, and the rothschilds seperately?”, to which the answer had been “I trust your judgement.”33 The misspelling is on the sheet and is not repaired here. Breakfast with Ehud Barak and him stands on Epstein’s schedule for 24 September 2013, on a day sheet whose image prints “6:30am Breakfast w/Josh Ramo and Ehud Barak”.344

On 2 June 2014 he introduced Epstein to the Princeton ecologist Simon Levin. The message was sent from his office on his behalf, copied to Epstein and to Lesley Groff, and signed by him:

I wanted to connect you with Jeffrey Epstein, who thinks a fair amount about complexity and network issues. He’s become a center of gravity for a number of researches, including Nowak, and I think if you’re in NYC at some point you’d enjoy meeting each other.

35

That is the clearest single act on this page: nearly six years after the conviction, a Kissinger Associates executive introducing Epstein by name to a named academic, out of his own office and over his own name, and recommending that the two of them meet. What the sheet shows is an introduction and a recommendation, and the page does not read a warrant of character into it. The title the release gives him, Vice Chairman, is printed on a sheet fourteen months later than this message and this page does not carry it back to 2014; what the June 2014 sheet itself shows is that the message came out of his office over his name.353

On 15 October 2015 Epstein addressed a one-line introduction to two men at once; the To: field carries “Joshua Cooper Ramo” beside “Ugo Brachetti Peretti”.36 Eleven minutes later he answered both: “Thanks Jeffrey. / Ugo, nice to meet you. How best to connect? / Joshua Ramo”.37 Thirteen days later he reported back to Epstein: “Had a good chat with Ugo. Thanks. Will catch you up when we meet.”38 The Ugo Brachetti Peretti page records that no document read for it established what followed the introduction; this is that document. The sheet writes only “Ugo”, and the identification rests on the To: field of the introduction thirteen days earlier, which names him in full; that is an inference and is stated as one.

The bonus, November and December 2014

On 23 November 2014 he wrote: “Hey — I have a quick question need some advice on bonus negotiation this year. You mind? I’m in NYC chilling out mostly today. Could also call tomorrow. J”11

Epstein answered with numbers and then with a script. The chain printed on one sheet runs:

to summarize , 1 I have given it considerable thought , and what I think is FAIR is 4,5 . I have increased the percentage of the firms biz. 18 vs 24. 18 vs 26 counter BLAH BLAH< I think 4.5 is fair. ------ no further explanation until later. — but as you know reveneue are down etc. , I think 4. 5 is fair

Thanks. Extremely helpful chat.

Question: Looks like the net number may go up by about 2.5 in the next week if another fee gets triggered on a deal I started, so we’d be at 20 more or less in my business this year. But I won’t know for sure for a week or so. So if I make the call on Wednesday this week, should I say — seems to me 4.5 is fair at the 18 number, and about 5.25 at the 20 number?

i woudl just say given its 18 i think 25 % of that is fair at 4.5. if it then goes up, you hsould ask for an addtional 25 %

12

The inner timestamps on that sheet are client-rendered and do not run in one order; the page therefore dates the exchange to the day its own header carries and asserts no sequence within it. The sheet’s Date: field is “Sun, 23 Nov 2014 19:11:01 +0000”.

The negotiation ran for four weeks and he reported to Epstein after each round. On 4 December 2014, after the first meeting: “Went fine. I think your line about 25% works. / Basically available pool for total comp is down 50 percent from last year on higher expenses and less revenue in other businesses. My stuff is 80% of revenue. He would not give me a number so we didn’t get into the “what can you come up to” game. He will submit some recommendations to Henry, seems willing to press for the 25%. / Thanks. Let’s see how this turns out, but I think at least you saved me from a disastrous number.” Epstein’s line that morning had been: “im avaible , if you need to war game further”.39

On 17 December 2014 the whole of the coaching stands on one sheet. Epstein wrote out the tactic:

you need to express your view that you believe it was unfair and ask if there is anything constructive you can do to raise your comp. you need to let him vent and tell you its impossible , etc. and then suggest you Need to come up with a way, what does he suggest. this only after he vents. he needs to tell you why he is right and you are being greedy. . if i can bring in an addtional fee in the next two months can we agree a payout rate. etc.

He answered with a draft of what he would say, and then with something else:

Yes. I think it’s probably just a red flag that will piss him off. But I am going to say “Look, you told me you always want to be fair and I should say if you are not. I don’t think cutting my comp by 30 percent in a year when I’m up a bit and with no notice is fair. It was shocking and creates real problems for me financially. So my concern is first this years comp and if I can do anything and second going forward”

But, Jeff, my main worry here is if he gives in and does give me some additional money it will actually be bad for me in the long run. I don’t know what I stand to gain here. He has all the power. I have no leverage. He’s not a good person.

Epstein’s reply was one line: “dont say he is NOT fair, just say that you want to be fair to him”.40

In this exchange the person on the other side is named only as “Henry”, and the page supplies no surname from it. The firm is named on his own assistant’s signature block as Kissinger Associates, Inc.3 A document of a later year, quoted below, writes the name out in full.16

The bad press, January 2015

On 30 January 2015, during the week the Giuffre filings were in the news, he wrote two words of his own: “You doing okay?” Epstein answered: “very well except for bad press. it will pass”. He replied: “OK. Let me know if you need anything”.41

That is the whole of what the documents read for this page record him saying about Epstein’s conduct or the reporting of it, and the claim was built to be falsifiable. Thirteen word-bounded forms were run over all 1,993 documents on 12 September 2026, the words a disconfirming document would use, not this page’s, and every return was opened: accuser, allegation, allegations, convicted, Giuffre and Maxwell return 0 documents each; victim and victims return 2, both of them a Florida police narrative inside a large compiled exhibit that is in the population for a different person’s name; survivor returns 20, every one of them the film title “Lone Survivor” on a schedule sheet; plea returns 4, every one of them the string plea=e note, the words “please note” carrying quoted-printable damage; conviction returns 1, in an essay about computer security; registered returns 3, an aircraft registry entry and a British partnership footer; Roberts returns 2, a Bitcoin pseudonym and a list of computing pioneers. No document in the population is a message of his about the subject, and this page does not assert that he never discussed it, only that no document read for it records him doing so.

The lawyer, April 2015

On 10 April 2015 he asked for a name: “Good seeing you. / Will you send me the name of the lawyer you recommended for my employment agreement. Joshua” Epstein’s whole reply was “jay lefkowitz , kirkland and ellis”; his was “Thanks”.42

Twelve days later: “I am seeing Jay tomorrow. But I need some strategic advice. Can I call you on Thursday or Friday at some point. Or will try you when I am back from Asia, / Hope all is well / J”.7 Three days after that:

Well: Jay Lefkowitz’ advice so good and helpful I had no need to call you. He’s great. Thank you.

2

Jay Lefkowitz of Kirkland & Ellis was one of Epstein’s own defence lawyers in the Florida matter.43 The sheets say what the referral was for, an employment agreement, and say nothing about what the advice was.

The aircraft, August 2015

In August 2015 Epstein’s aircraft carried a passenger from Albuquerque to New York at his request. The sheets call her only “she”, give no name, and state no relationship between her and him; none is supplied here. The arrangement itself stands in the release in full.

On 18 August 2015 Epstein offered a seat: “probably going to new york tomorw if you want a ride”. He asked “So 3 pm departure from ABQ tomorrow? Let me check.” Epstein answered “from general aviation tail 908 JE”. He then asked, for someone else and not for himself, whether she could go instead; the forename in that sentence is withheld on the sheet and is not completed from it here. Epstein: “of course she can come.” He answered: “Great. So I’ll have her there at 245 or so? Should we give any details to your office?”, and, at the foot of the same message, “I really appreciate it.” Epstein offered the aircraft itself as an inducement, in the same chain: “its the 727 , you could work in the office uniterrupted if that is an enticement”.44

The organising was done between the two offices. Epstein’s office wrote to him separately that afternoon, asking for a passenger’s details in these words: “Our Pilot, Larry, who is on this email will need her full name, birthdate…if we could get a copy of her passport that would be great”, and his assistant at Kissinger Associates answered, over the signature block that gives his title.3 He confirmed the drop-off himself the next morning.45

The firm, 2016 and 2017

From early 2016 the correspondence acquires a running subject: the future of his business inside the firm, and Epstein as his adviser on it. On 2 February 2016 he wrote “OK. Was asked for a good forward proposal for structure of t=e business. Want to discuss”. The = is printed on the image.46

On 26 July 2016 he sent a draft, in an attachment named KCN_Proposal.docx, with the line “This seem about right to you?”

A single unheaded sheet in the release sets out the structure of a company called Kissinger China NewCo. Whether it is the attachment he sent is not stated by any document read for this page and is not asserted here.16 It is titled “Kissinger China NewCo Structure and Operation” and it opens: “A new corporation, Kissinger China NewCo (KCN) will be formed that will contain the China-focused business operations of Kissinger Associates.” Under the heading “Roles” it sets out two:

Henry Kissinger will serve as Executive Chairman of the firm. He will have final say on all client selection matters and on operational decisions about how client matters are pursued, hiring decisions and significant capital expenditures.

Joshua Ramo will serve as Chairman and CEO of the firm. He will be responsible for daily operation of the firm, consistent with the practice of the current China operations of Kissinger Associates.

Under “Structure” it sets out the split: “KCN profits will be distributed to employees of KCN, but for the first five years, Kissinger will receive profit distributions as follows: 50% in year one, 40% in year two, 25% in year three and 10% in years four and five. / The firm will be owned initially 100% by Ramo, though ownership interest will eventually be shared among partners of the new firm.” Its stated aim is “the leading advisory firm on China-related commercial matters”, and it answers its own question about existing clients: they stay with the old entity and are “serviced by” the new one.16

That sheet names the two principals in full. It carries no header, no date and no sender field; what connects it to the correspondence is its subject and nothing else, and the connection is not asserted as a fact here.

Epstein answered the draft with the terms of a corporate separation:

entity , not corporation . death of josh or death or incapcity of henry. tbd. do you want to start at 50% or let him negotiate you up. ? do you want to talk about any current or former clients? multiyear already existing obligations? option to buy the rest of his biz , ? etc

He came back with questions of his own: “I mean to say that current clients when they renew will fold into newco. Should I/can I say that the clients are transferred now? They signed contracts with old entity. Or need to just specifiy profit split. / No interest to buy rest of business / On his death or incapacity goes to 5% ‘till five year total clock is up?”, and Epstein added one more term, with its purpose attached: “or the aility to buy out the balance. of years so you have no one outside auditing your books”. His whole reply was “ok”.47

On 28 July 2016 he reported back:

Done. He asked me yesterday to prepare something soon, and I had the doc ready and gave to him.

He looked over. No reaction, but said he’d pass to his lawyer who will discuss with me. He talked a bit about maybe he’ll just shut stuff down.

Weirdly he said he wants to get this done after the election, by the end of the year. To me this sounds like he doesn’t really want to do it but is just playing with it. I’ve been through this drill before, so I’m a bit frustrated in that regard. I can’t tell if he’s just playing me to be sure I keep the business going for a bit longer.

13

Two days later Epstein wrote: “you might consider hiring a lawyer . tell henry and his lawyer , that you have done so , and have your lawyer be the pest . he can call each week, send suggestions . negotiate, without the personal emotinal charge”. He answered: “Yep. Good idea.”14

The subject was still live eleven months later. On 10 June 2017 he wrote one sentence: “When is a good time to brainstorm some options for future structures to pass along to the man?”15 In the correspondence Epstein and he call the counterparty “henry” and “the man” and never write a surname. The identification is taken here from the firm and not from a second document: his own assistant’s signature block puts him at Kissinger Associates, Inc., and the “henry” who sets his compensation at that firm, and whose lawyer he is told to expect, is Henry Kissinger.3 The unheaded sheet quoted above writes the name out in full as well;16 nothing on this page rests on that sheet, because nothing read for this page ties it to this correspondence.

Two details of method the sheets carry

Signal. On 20 July 2016 he wrote: “Park and 52 but can’t get free. How long are you in town= / Are you using Signal yet?” The = is printed on the image.48 The subject returns in October the same year: the extraction of two sheets of 10 October 2016 carries him proposing the application for a call from Italy and Epstein sending him a link to it. Those two sheets were read from the text layer only and nothing from them is quoted here.49

Starbucks. On 16 June 2016, declining a Friday: “Tomorrow I’m tied up working a shift at Starbucks. Seriously. So I’ll come over on Saturday.”50 Epstein’s April 2017 line to the Qatari correspondent, above, says he “sits on board”. On that one point the outside record confirms Epstein rather than correcting him: the April 2023 report cited below states that he “at the time served on the boards of Starbucks Corp. and FedEx Corp”.51 That is also why FedEx appears in the Response section, and it is the only one of Epstein’s three characterisations of him that any source read for this page corroborates.

The last messages

The correspondence thins after 2017. On 10 April 2017 Epstein asked “are you playing a part in the xi trump. fun”, and the answer was four words: “In Europe. You well?”52 The message survives in two copies and they do not print the same characters: the EFTA copy prints “xi =rump”, the = being quoted-printable damage carried through that copy, and the House Oversight copy prints the sentence undamaged. The undamaged printing is the one quoted here. The last documents in the set he wrote are both dated 31 July 2018. The first, subject “Santa Fe”, has for its whole body “August 4-14. You around?” Epstein answered “Yes”, and the last thing he says to him in this release is “Great. Let’s have lunch some day?” The terminus was tested by date and not by wording: every document in the population whose own sender field carries his surname was scanned for a year token of 2018 or later, and exactly three returned, of which two are those two sheets and the third is 28 March 2018.17

Response

No response from him, or from anyone on his behalf, has been located. The outside record was searched on 12 September 2026 for a statement, denial or comment: the Wall Street Journal’s April 2023 report that Epstein’s calendar showed “more than a dozen meetings” with him between 2013 and 2017 is the earliest outside account, and the secondary coverage of it that could be read carries no comment from him, from Kissinger Associates, from Starbucks or from FedEx.51 Searches for a statement after the 2025 and 2026 document releases returned none. A response is not recorded as absent because he declined to give one, no request from this desk was made; the finding is only that none is published where it would be found.

What the record here does not establish

  • This page makes no claim about who else was present at any meeting it names. The schedule sheets give a time, a place and a name; they do not give a guest list, and several entries beside his carry barred values that are not completed here.
  • No document read for this page records any money passing between them, in either direction. Eight word-bounded forms were run over all 1,993 documents on 12 September 2026 and every return was opened: invoice, fees and retainer return 0 documents each; wire returns 2, a security essay and a book review; payment returns 4, none of them between the two of them; commission returns 13, of which the largest group is Epstein’s own line “then out of commission until Aug 30” and the rest are the European Commission, a state racing commission and a federal trade commission; paid returns 6, none of them a payment between them; and fee returns 19, every one of them a copy of two lines in the 2014 bonus chain, his own “if another fee gets triggered on a deal I started” and Epstein’s “if i can bring in an addtional fee in the next two months can we agree a payout rate”. Those are fees inside his own firm, not fees between him and Epstein. Two of the eight searches are circular and carry no weight: this page prints the fee line and the commission line itself, so neither search can separate its own material from a disconfirming document. The finding rests on the six that appear nowhere in a quoted line here, invoice, fees, retainer, wire, payment and paid, which return 0, 0, 0, 2, 4 and 6 documents, every one of which was opened. The advice this page records ran one way without a price attached to it on any sheet read here.
  • No identity number, account number, telephone or fax number, e-mail address, residential address or date of birth printed on any sheet cited here is reproduced on this page.

Notes on the text layer

Five faults were measured on sheets cited here, each by comparing the extraction with the image at 300 dpi or higher. Every quotation in the body of this page was read at the image, and the footnote beside it says so. Where a footnote says its sheet was read from the text layer only, nothing from that sheet is quoted in the body. Five footnotes quote a string inside the footnote itself without having read the sheet ([^selfdesc], [^twojoshes], [^schedclean], [^roberta] and [^tramo]), and each says in terms that it is quoting the extraction and makes no claim about what the sheet prints.

  • On EFTA00968960 the extraction prints “ehud bank on the 22?”; the image prints “ehud barak”.
  • On EFTA00706451 the extraction prints “Will follow tip.”; the image prints “Will follow up.”
  • On EFTA00980960 the extraction opens “lust sent a message to Ariane”; the image prints “Just sent”.
  • On EFTA01957450 the extraction prints “arian rothschfild”; the image prints “arian rothschiild”, with a doubled i.
  • On EFTA00641081 the extraction scrambles the line order of the quoted chain and drops most of one message; the image carries it whole.

And one entry that stood in this list and should not have. An earlier version of this page recorded a sixth fault “in the other direction”: that on EFTA00969470 the extraction printed “separately” where the image printed “seperately”, and that the extraction had therefore silently repaired a misspelling on the face of the sheet. The corpus does not hold that. The extraction of EFTA00969470 prints seperately; the image, read for this correction at 600 dpi, prints seperately; the only text that read “separately” was this page’s own quotation of the line.33 Both are corrected above. The false entry is recorded here rather than deleted, because this section exists to tell a reader how far to trust this page’s handling of the text layer, and an accuracy audit that carries an invented fault is worse than the misspelling it claimed to describe.

A sixth fault is the one that moved this page’s arithmetic. On the schedule sheets the extraction renders a printed w/Josh Ramo as wilosh Ramo, wljosh Ramo, wlfosh Ramo, w/Joshtta Ramo and, once, watch Ramo. FTS5 tokenises w/josh as the two tokens w and josh, so the query "w/josh" reaches a sheet only where the slash survives the extraction intact; every damaged form falls out of that query and into the bare-surname residual instead. Fifty-eight of the 150 residual documents this page admits are schedule sheets of exactly that kind.

Footnotes

Footnotes

  1. https://epstein-data.com/EFTA00694397, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: Epstein, Subject: “Barak”, Date: “Thu, 14 Nov 2013 16:55:58 +0000”. Read at the image at 300 dpi. A second copy is at https://epstein-data.com/EFTA02388564. ↩ ↩2

  2. https://epstein-data.com/EFTA00858458, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “jeffrey E.”, Subject: “Re:”, Date: “Sat, 25 Apr 2015 22:26:24 +0000”. Read at the image at 300 dpi. Epstein’s quoted line on the same sheet ends in a barred value. ↩ ↩2

  3. https://epstein-data.com/EFTA00343410, printed p.1, From: “Ybel Collado” over a barred address, Subject: “RE: Jeffrey Epstein”, Date: “Tue, 18 Aug 2015 19:37:37 +0000”. Read at the image at 300 dpi. The signature block reads “Ybel Collado / Assistant to Vice Chairman, Joshua C. Ramo / Kissinger Associates, Inc. / 350 Park Avenue, 26th FL / New York, NY, 10022”, followed by two barred lines and a barred Email: value. The quoted request from Epstein’s office is the inner message on the same sheet. Further copies are at https://epstein-data.com/EFTA00365497, https://epstein-data.com/EFTA00383227 and https://epstein-data.com/EFTA02100920. ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7

  4. https://epstein-data.com/EFTA00663461, printed p.1, To: Epstein, Subject: “Re: Lunch Wed. Sept. 25th? Sultan…”, Date: “Sun, 22 Sep 2013 13:29:44 +0000”, the day sheet headed “Tuesday Sept 24, 2103 NY” — the year is printed as “2103” on the sheet and is not repaired here. Read at the image at 300 dpi, where the line prints “6:30am Breakfast w/Josh Ramo” over a barred value. A copy of a differently worded version of the same day, naming Barak on the same line, is at https://epstein-data.com/EFTA01955873; that copy was read from the text layer only. ↩ ↩2

  5. https://epstein-data.com/EFTA00980211, printed p.1, From: “Lesley Groff”, To: Epstein, Subject: “Paul Barrett”, Date: “Tue, 07 Jan 2014 16:22:07 +0000”, the day sheet headed “Fri. Jan. 10, 2014 NY-LSJ”. Read at the image at 300 dpi, where the line prints “3:30-5:30pm Appt w/Josh Ramo and Ariane Rothschild”; the extraction of the same sheet prints “wilosh”. ↩ ↩2

  6. https://epstein-data.com/EFTA01964792, printed p.1, From: a barred value “on behalf of Joshua Cooper Ramo” over a second barred value, To: “Joi Ito”, Cc: Epstein, Sent: “Sun 7/14/2013 3:27:58 PM”, Subject: “Re: getting together”. Read at the image at 300 dpi. Joi Ito’s quoted message of 13 July 2013 stands on the same sheet. ↩ ↩2 ↩3 ↩4

  7. https://epstein-data.com/EFTA00688320, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “Epstein Jeffrey”, Subject: “Hey”, Date: “Wed, 22 Apr 2015 18:02:28 +0000”. Read at the image at 300 dpi. ↩ ↩2

  8. https://epstein-data.com/EFTA00706451, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: Epstein, Date: “Fri, 23 Aug 2013 15:57:20 +0000”. Read at the image at 300 dpi. The forwarded message from “Nowak, Martin” is dated “Fri, Aug 23, 2013 at 6:26 AM” on the same sheet. The extraction of this sheet prints “Will follow tip.”; the image prints “Will follow up.” ↩ ↩2

  9. https://epstein-data.com/EFTA00968960, printed p.1, From: Epstein, To: “A. de Rothschild” and “COLOM, Olivier”, each over a barred address, Date: “Mon, 02 Sep 2013 22:28:16 +0000”. Read at the image at 300 dpi. The Subject: label is printed with no value. The extraction prints “ehud bank”; the image prints “ehud barak”. The sheet prints “copper”, with a double p, and it is not repaired here. The phrase “kissinger right hand” stands in 15 documents of the corpus, all of them copies of this one message. ↩ ↩2 ↩3

  10. https://epstein-data.com/EFTA02651539, printed p.1, From: “jeffrey E.”, Sent: “Saturday, April 29, 2017 9:55 AM”, To: “Jabor Y.”, Subject: “Re: Star Bucks”. Read at the image at 300 dpi. The = characters and the replacement glyph after “=C2” are printed on the image and are quoted-printable transmission encoding surviving onto the face of this production copy; the same damage runs through every quoted line on the sheet. Not repaired here. ↩ ↩2

  11. https://epstein-data.com/EFTA00714979, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “Jeffrey Epstein”, Subject: “Hola”, Date: “Sun, 23 Nov 2014 14:22:57 +0000”. Read at the image at 400 dpi. A second, separate line of his stands below it on the same sheet and is not quoted here. ↩ ↩2

  12. https://epstein-data.com/EFTA00641081, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “jeffrey E.”, Subject: “Re:”, Date: “Sun, 23 Nov 2014 19:11:01 +0000”. Read at the image at 300 dpi. The three quoted passages are, in the order the sheet prints them from the foot upward, Epstein’s summary, his question and Epstein’s reply. The extraction of this sheet scrambles the line order and drops most of Epstein’s summary. ↩ ↩2

  13. https://epstein-data.com/EFTA00823213, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “Jeffrey Epstein”, Subject: “update”, Date: “Thu, 28 Jul 2016 15:41:28 +0000”. Read at the image at 300 dpi. ↩ ↩2

  14. https://epstein-data.com/EFTA00630176, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “jeffrey E.”, Subject: “Re:”, Date: “Sat, 30 Jul 2016 13:28:48 +0000”. Read at the image at 300 dpi. Epstein’s quoted line of the same day stands on the same sheet. ↩ ↩2

  15. https://epstein-data.com/EFTA01043752, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “Epstein Jeffrey”, Subject: “Call”, Date: “Sat, 10 Jun 2017 12:01:28 +0000”. Read at the image at 300 dpi. ↩ ↩2

  16. https://epstein-data.com/EFTA00584255, printed p.1 of 1. Read at the image at 300 dpi. A single sheet with no e-mail header, no date and no sender field, headed “Kissinger China NewCo Structure and Operation”. What ties it to this correspondence is its content and the attachment name at the preceding footnote; no document read for this page states that this sheet is that attachment, and this page does not assert it. The corpus classifies it as a corporate-formation document. ↩ ↩2 ↩3 ↩4 ↩5

  17. https://epstein-data.com/EFTA02599848, printed p.1, From: “Joshua Cooper Ramo” over a barred value, Sent: “Tuesday, July 31, 2018 6:58 PM”, To: “Epstein Jeffrey”, Subject: “Santa Fe”; and https://epstein-data.com/EFTA02602522, printed p.1, Sent: “Tuesday, July 31, 2018 11:43 PM”, Subject: “Re: Santa Fe”, which carries Epstein’s “Yes” as a quoted line. Both read at the image at 300 dpi. The Sent: fields are bare and the quoted line on the second sheet is stamped 1 August 2018 by a different client, so the two are in different zones; this page gives the day its headers carry. ↩ ↩2

  18. Four documents, and the strings below are quoted from the extraction, not from any sheet: https://epstein-data.com/EFTA00284747, printed p.3, whose extraction carries “11:00am Appt w/Josh Harris” and “11:00am Appt w/Josh Ramo” as separate entries of one day; https://epstein-data.com/EFTA00284799, printed p.11, which carries both “w/Josh Ramo” and “w/Josh Black”; https://epstein-data.com/EFTA02554054, printed p.1, “w/Josh Koppel at Scrollmotion”; https://epstein-data.com/EFTA00129185, printed p.3, “w/ Josh Carlo”. The enumeration of all 328 documents was made by reading the extraction of each, and the table it produced is a statement about the extraction. No sheet in this footnote was read at the image, and nothing here is a claim about what any of them prints. ↩

  19. https://epstein-data.com/EFTA00383363, printed p.5, whose extraction carries “w/Josh Rant°” followed by “11-11:30am Appt w/Bruce Galloway”, against https://epstein-data.com/EFTA00663461, printed p.1, whose image prints “6:30am Breakfast w/Josh Ramo” followed by the same Galloway line. Text layer for the first, read at the image at 300 dpi for the second. The comparison is between one extraction and one image and is stated as such. ↩

  20. Three documents. The extractions of https://epstein-data.com/DOJ-OGR-00031174 at printed p.82 and p.125, and of https://epstein-data.com/EFTA02729080 at printed p.25, carry the string “Roberta Cooper Ramo, Albuquerque Attorney” in a list of political contributions; the extraction of https://epstein-data.com/EFTA02855411 at printed p.82 and p.125 carries the same line without the middle element. All five readings are of the extraction, and nothing here is a claim about what those sheets print. DOJ-OGR-* documents cannot be rendered from the site. Two of the three were admitted to this subject’s population by the first classification run and removed by the second; see the paragraph that cites this note. ↩ ↩2

  21. The nine documents in which the extraction drops the terminal s of a private person’s surname are https://epstein-data.com/EFTA01359760, https://epstein-data.com/EFTA01359982, https://epstein-data.com/EFTA01362497, https://epstein-data.com/EFTA01363694, https://epstein-data.com/EFTA01371313, https://epstein-data.com/EFTA01376736 and https://epstein-data.com/EFTA01460205, all of them sheets of one bank’s internal check-referral correspondence carrying the same recipient; and https://epstein-data.com/EFTA01827524 and https://epstein-data.com/EFTA02442791, two copies of one forwarded message of 18 May 2009 in which the sender line prints the truncated form directly above a line printing the surname in full. EFTA01363694, printed p.1, was read at the image at 400 dpi: it is a clean typed e-mail with a corporate signature block and a logo, not a damaged scan, and the surname prints with its s. None of the nine is this subject, none of them is a public figure, and no name from any of them is reproduced here. The two named as outside the "Ramos" population were checked against it directly. The Bureau of Prisons log is https://epstein-data.com/EFTA00053963, printed p.767, and the appellate decision is https://epstein-data.com/EFTA01070533, printed p.3; both read from the text layer, and nothing from either is quoted here. ↩ ↩2 ↩3

  22. The itemisation’s “a Ramon (7)” is eight documents and not seven, and the note this page first printed to correct the Ramos figure got that second figure wrong on its own description, counting two named in full and five others across a set of eight. The eight, each carrying exactly one ramo token, are https://epstein-data.com/EFTA02247230, printed p.3, https://epstein-data.com/EFTA02248312, printed p.3, https://epstein-data.com/EFTA02249116, printed p.3, and https://epstein-data.com/EFTA02249732, printed p.3, on each of which the forename is broken across three lines so that Ramo stands alone with its final letter on the line below; and https://epstein-data.com/EFTA02449139, printed p.1, https://epstein-data.com/EFTA02491619, printed p.2, https://epstein-data.com/EFTA02526697, printed p.1, and https://epstein-data.com/EFTA02660340, printed p.1, on each of which the line prints Ramo= or ramo=. Four and four. The earlier note was wrong in a second way: the ramo token on EFTA02247230 and EFTA02249116 does not stand in a To: field at all, and every one of the eight is in the residual because of breakage or damage, not because any of them lacks the name in full elsewhere. The = here is not a quoted-printable soft line break — a soft break is a = standing at the end of a line — but quoted-printable mangling in which a single character has been replaced by = wherever it falls. EFTA02526697, printed p.1, was read at the image at 400 dpi: the sheet prints the correspondent’s full name three times and the damaged form once, and carries eleven further instances of the same single-character substitution, among them “t=e lamdscaper”, “get b=ck to you” and “ready for ou= next meeting”. The identification therefore rests on the sheets printing the name in full, not on the damaged line. Text layer for the other seven. None of the eight is this subject, none of them is a public figure, and no name from any of them is reproduced here. ↩

  23. See Mark Tramo. The desk’s own document rulings record the extraction damaging his surname on other sheets. On https://epstein-data.com/EFTA02149316, printed p.1, and https://epstein-data.com/EFTA01920806, printed p.1, the sender display reads “arc ramo” and the body is his; on https://epstein-data.com/EFTA02654576, printed p.1, it reads “mark =ramo”. Text layer. Neither is this subject. ↩

  24. Three letterheads, each read at the image at 400 dpi on 12 September 2026 and then deleted. https://epstein-data.com/EFTA01130140, printed p.1, a letter of 10 November 2015 on the letterhead of Searcy Denney Scarola Barnhart & Shipley, P.A.: the attorney column prints “BRENDA S. FULMER” and the extraction of that line reads “BRENDA S. RAMO.”. https://epstein-data.com/EFTA02133670, printed p.1, Sent: “Wed 7/17/2013 3:51:12 PM”, Subject: “Re: Jeffrey Epstein”: the signature block of one of that firm’s partners prints “Paul, Weiss, Rifkind, Wharton & Garrison LLP” and the extraction of the same line reads “Paul, Weiss, RAMO, Wharton & Garrison LLP”. The partner’s own name is on the sheet and is not needed here. https://epstein-data.com/EFTA00606737, printed p.3, a Podhurst Orseck letter of 20 February 2009: the attorney column prints “Ramon A. Rasco” and the extraction reads “Ramon A. Ramo”; the same image prints the addressee as “Jay P. Lefkowitz, P.C. / Kirkland & Ellis LLP” where the extraction reads “Jay P. Leficowitz, P.C.”, which is the same fault recorded at [^lefkowitzwho] on a different document. Two further copies of that letterhead carry the same damaged name, at https://epstein-data.com/EFTA00220707, printed p.2, and https://epstein-data.com/EFTA01076123, printed p.3, both read from the text layer only. Every name restored in this note is an attorney’s, printed on that attorney’s own firm letterhead or signature block, and nothing about any client of any of them is taken from these sheets. None of the five documents is this subject’s. ↩

  25. scripts/act_rate.py, re-run on 12 September 2026, --selftest PASS. The first two rows are reproducible: act_rate.py '"Joshua Cooper Ramo"' Ramo returns 1,247 documents, 491 sender-field acts, 506 addressed only, 14 header-less authored, 0 press and 41 refusals; the four-named-forms row is the same classifier over the union of "Joshua Cooper Ramo", "Josh Ramo", "Joshua Ramo" and "Mr. Ramo", which is 1,817 documents. Both “addressed only” figures were corrected on 12 September 2026, from 503 and 531. The third row’s population is the one described above and cannot be rebuilt from a single query, so a reader cannot re-run it; re-measuring it over an independent reconstruction differing from it by two documents returned 494 sender-field acts, 535 addressed only, 37 header-less authored, 0 press and 108 refusals, and the row as printed is the original count over the exact population. The Trump control is act_rate.py '"Donald Trump"' Trump; on the key Donald Trump the same query returns 2,898 documents, 0 sender-field acts, 2 addressed, 158 header-less authored, 22 press and 83 refusals. Both sets are correct about their own key. ↩ ↩2

  26. The 37 are the documents scripts/act_rate.py’s own header rule counts as header-less and authored over the population used here, and every one was opened. The contact record is https://epstein-data.com/EFTA01377511, printed p.1, read at the image at 400 dpi, where the fields print as quoted above; the sheet also prints, below the fields, the words “Met through JE” — the extraction of that line reads “Met through )E” — and whose record it is is not stated on the sheet and is not asserted here. The two book sheets are https://epstein-data.com/HOUSE_OVERSIGHT_018232 and https://epstein-data.com/HOUSE_OVERSIGHT_018233; HOUSE_OVERSIGHT documents cannot be rendered from the site, so both were read from the text layer and nothing from either is quoted here. The message of 4 December 2013 stands twice, at https://epstein-data.com/EFTA00505433 and https://epstein-data.com/EFTA02011889; the first was read at the image at 400 dpi, where the sheet carries a To:, a Subject: and a Date: and no body at all, and the second was read from the text layer. The guest list is https://epstein-data.com/EFTA02671456, printed p.1, read at the image at 400 dpi; nothing on that sheet beyond the line carrying his name is described or reproduced here. The count of 31 schedule sheets is a count over this page’s population and moves with it; the six documents named in this note do not. ↩

  27. https://epstein-data.com/EFTA00387673, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “Lesley Groff”, Date: “Mon, 15 Jul 2013 19:38:06 +0000”. Text layer; his agreement is one word and nothing from this sheet is quoted on this page. ↩

  28. https://epstein-data.com/EFTA00636796, printed p.1, Date: 12 August 2013. Text layer; nothing from this sheet is quoted on this page. A second copy is at https://epstein-data.com/EFTA01963767. ↩

  29. https://epstein-data.com/EFTA00980960, printed p.1, From: “COLOM, Olivier” over a barred address, To: “‘Jeffrey Epstein’”, Subject: “RE:”, Date: “Thu, 02 Jan 2014 13:23:01 +0000”. Read at the image at 400 dpi. The extraction opens “lust sent a message to Ariane”; the image prints “Just sent”. A second copy is at https://epstein-data.com/EFTA01940041, read from the text layer. ↩

  30. https://epstein-data.com/EFTA00980582, printed p.1, From: “COLOM, Olivier” over a barred address, Subject: “Re:”, Date: “Sat, 04 Jan 2014 22:04:51 +0000”. Read at the image at 400 dpi, where the name standing before “told me” is covered by a redaction bar and the six numbered names run “1) Reid / 2) Summers / 3) Mr West / 4) Jes Staley / 5) Messina - if we can make the contact / 6) Cooper Ramo”. Further copies, read from the text layer, are at https://epstein-data.com/EFTA00980592, https://epstein-data.com/EFTA01940167 and https://epstein-data.com/EFTA01940752. ↩

  31. https://epstein-data.com/EFTA01957450, printed p.1, To: “Landon Thomas” over a barred address, From: “Jeffrey Epstein”, Sent “Tue 9/24/2013 1:05:21 PM”, Subject: “Re: Just like old times…”. Read at the image at 400 dpi. Epstein’s question of the same morning, “have interesting scientists at 645 if you would like to stop by”, and Thomas’s “Like who?” stand on the same sheet, and the four names are printed on one line. The extraction prints “arian rothschfild”; the image prints “arian rothschiild”, with a doubled i. ↩

  32. The search was run on 12 September 2026 over the 1,993-document population, restricted to documents whose first sheet carries his surname in the sender field as the act test reads it, and the returns were opened. The single vice chair return is https://epstein-data.com/EFTA00343430, printed p.1, where the extraction of the quoted block reads “Ybel Collado > Assistant to Vice Chairman, Joshua C. Ramo > Kissinger Associates, Inc.” Text layer, and the string above is quoted from the extraction and not from the sheet; the same signature block is read at the image at [^vicechair]. ↩

  33. https://epstein-data.com/EFTA00969470, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “Epstein Jeffrey”, Subject: “Re:”, Date: “Mon, 09 Sep 2013 12:11:24 +0000”. Read at the image at 600 dpi. Both quoted passages stand on this one sheet, his reply above Epstein’s quoted line. The image prints “seperately” and so does the extraction of the same sheet; see the note below on this page’s own error about that word. A different sheet nine days later, https://epstein-data.com/EFTA00970962, printed p.1, carries the arrangement of the 24 September meeting and not this exchange, and neither quotation above is on it; text layer, and nothing from it is quoted here. ↩ ↩2

  34. https://epstein-data.com/EFTA02723859, printed p.1, To: Epstein, Sent: “Tue 9/24/2013 12:18:52 PM”, the day sheet headed “Tuesday Sept 24, 2103 NY” — the year is printed as “2103” on this sheet too and is not repaired here. Read at the image at 400 dpi, where the line prints “6:30am Breakfast w/Josh Ramo and Ehud Barak” followed by barred values. The extraction of the same sheet prints “6: Oam Break ast w/Josh Rama and Fluid Barak”, which is why the query "w/josh" does not reach it. ↩

  35. https://epstein-data.com/EFTA01922708, printed p.1, To: Epstein, “Simon Levin” over a barred value and “Lesley Groff”, From: a barred value “on behalf of Joshua Cooper Ramo”, Sent “Mon 6/2/2014 12:14:42 PM”, Subject: “Connection”. Read at the image at 400 dpi, where “Simon Levin” is printed cleanly and the message opens “Si, / Good seeing you Saturday, even briefly.” above the quoted paragraph. A second copy is at https://epstein-data.com/EFTA01922803. ↩ ↩2

  36. https://epstein-data.com/EFTA00843499, printed p.1, From: Epstein, To: “Joshua Cooper Ramo” over a barred address and “Ugo Brachetti Peretti”, Date: “Thu, 15 Oct 2015 15:28:58 +0000”. Read at the image at 400 dpi, where the To: field prints both names in full over barred addresses. This sheet is the reason this page exists: it was found by the Ugo Brachetti Peretti writer. ↩

  37. https://epstein-data.com/EFTA00843500, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: Epstein and “Ugo Brachetti Peretti” over a barred address, Date: “Thu, 15 Oct 2015 15:39:21 +0000”. Read at the image at 400 dpi by the Ugo Brachetti Peretti pass, whose footnote records the reading; a second copy is at https://epstein-data.com/EFTA01732146. ↩

  38. https://epstein-data.com/EFTA00669495, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “Jeffrey Epstein”, Subject: “Hi”, Date: “Wed, 28 Oct 2015 20:27:37 +0000”. Read at the image at 300 dpi. ↩

  39. https://epstein-data.com/EFTA00637003, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “jeffrey E.”, Subject: “Re:”, Date: “Thu, 04 Dec 2014 16:50:47 +0000”. Read at the image at 300 dpi. Epstein’s quoted line of the same day stands on the same sheet. ↩

  40. https://epstein-data.com/EFTA00673114, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “jeffrey E.”, Subject: “Re:”, Date: “Wed, 17 Dec 2014 14:34:27 +0000”. Read at the image at 300 dpi. The sheet carries the whole exchange of that day in one chain; the three passages quoted above are, in the order the sheet prints them from the foot upward, Epstein at 9:28, his own reply at 10:31 and Epstein at 9:33. The inner timestamps are client-rendered and do not run in one order, so this page asserts the day and not the sequence within it. A copy of the middle passage alone is at https://epstein-data.com/EFTA00709356, printed p.1, also read at the image at 300 dpi. ↩

  41. https://epstein-data.com/EFTA00689862, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “jeffrey E.”, Subject: “Re: Hola”, Date: “Fri, 30 Jan 2015 16:03:55 +0000”. Read at the image at 300 dpi. His opening message of the same day is at https://epstein-data.com/EFTA01141887. The inner timestamps on the sheet are client-rendered and do not run in one order; the page dates the exchange to the day its header carries. ↩

  42. https://epstein-data.com/EFTA00670888, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “jeffrey E.”, Subject: “Re:”, Date: “Fri, 10 Apr 2015 17:15:19 +0000”. Read at the image at 300 dpi. The sheet carries all three messages. ↩

  43. The Justice Department’s Office of Professional Responsibility report on the 2006–2008 federal investigation of Jeffrey Epstein names him, at narrative p. 9, as one of two Kirkland & Ellis attorneys who contacted the U.S. Attorney’s Office on Epstein’s behalf in August 2007: https://epstein-data.com/EFTA02830776, printed p.35. Text layer here — the extraction of that page prints the surname as “Leibowitz” — and the passage is quoted and read at the image at 300 dpi on Jay Lefkowitz, which is this desk’s page for him. ↩

  44. https://epstein-data.com/EFTA00848348, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “jeffrey E.”, Subject: “Re:”, Date: “Tue, 18 Aug 2015 21:28:25 +0000”. Read at the image at 300 dpi. The whole of the quoted exchange of 18 August 2015 stands on this one sheet. The page completes no withheld value from it. ↩

  45. https://epstein-data.com/EFTA00343430, printed p.1, Subject: “Re: Jeffrey Epstein”, Date: “Wed, 19 Aug 2015”. Text layer. ↩

  46. https://epstein-data.com/EFTA02391487, printed p.1, From: “Joshua Cooper Ramo” over a barred value, Sent: “Tuesday, February 2, 2016 4:39 PM”, To: “jeffrey E.”, Subject: “Re: You around NYC this week?”. Read at the image at 300 dpi. The = in “t=e” is printed on the image; this production copy carries quoted-printable damage throughout. ↩

  47. https://epstein-data.com/EFTA00823472, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “jeffrey E.”, Subject: “Re: Hello”, Date: “Tue, 26 Jul 2016 15:05:38 +0000”. Read at the image at 300 dpi. All four quoted passages, including his opening line and the attachment reference, stand on this one sheet as a quoted chain. The opening message is separately at https://epstein-data.com/EFTA00823446, printed p.1, Subject: “Hello”, Attachments: “KCN_Proposal.docx”, which was read from the text layer. ↩

  48. https://epstein-data.com/EFTA02455873, printed p.1, From: “Joshua Cooper Ramo” over a barred value, Sent: “Wednesday, July 20, 2016 2:13 PM”, To: “jeffrey E.”, Subject: “Re: You in town?”. Read at the image at 300 dpi. The = characters are printed on the image. ↩

  49. https://epstein-data.com/EFTA00815855, printed p.1, and https://epstein-data.com/EFTA00815913, printed p.1, both dated 10 October 2016. Text layer; nothing from either sheet is quoted on this page. ↩

  50. https://epstein-data.com/EFTA00639604, printed p.1, From: “Joshua Cooper Ramo” over a barred address, To: “jeffrey E.”, Subject: “Re: So”, Date: “Thu, 16 Jun 2016 20:13:05 +0000”. Read at the image at 300 dpi. ↩

  51. The Wall Street Journal’s April 2023 report is the earliest outside account located; its finding, as restated in the secondary coverage read for this page and in the encyclopedia article that cites it, is that Epstein’s calendar showed “more than a dozen meetings” with him “from 2013 to 2017”. The corpus population measured above is far larger than a dozen meetings, and a schedule entry is not a meeting held. Secondary coverage read on 12 September 2026 at https://www.foxnews.com/politics/epsteins-private-calendar-reveals-planned-meetings-obama-admin-official-cia-chief, published 30 April 2023 and updated 1 May 2023, carries no comment from him or from any of the three organisations named beside him. The same article states that the documents named “Joshua Cooper Ramo, who at the time served on the boards of Starbucks Corp. and FedEx Corp”, which is the corroboration cited in the Starbucks paragraph above; re-read for this correction on 12 September 2026. ↩ ↩2

  52. https://epstein-data.com/EFTA02653869, printed p.1, From: “Joshua Cooper Ramo” over a barred value, Sent: “Monday, April 10, 2017 9:30 AM”, To: “jeffrey E.”, Subject: “Re:”. Read at the image at 300 dpi. The = in “=rump” is printed on the image; this production copy carries quoted-printable damage throughout. A second copy of the same message is at https://epstein-data.com/HOUSE_OVERSIGHT_032524 p.1, and that sheet was read at the image at 8× native on 2026-09-14. It is one sheet, headed From: Joshua Cooper Ramo over a barred value, Sent: 4/10/2017 9:29:36 AM, To: jeffrey E. over a barred value, Subject: Re:, Importance: High; it prints In Europe. You well? over On Sun, Apr 9, 2017 at 00:46 jeffrey E. <[email protected]> wrote: and then, undamaged, are you playing a part in the xi trump. fun. The body above has been changed to quote this printing rather than the EFTA copy’s =rump. An earlier version of this footnote said this copy “was read from the text layer only, because HOUSE_OVERSIGHT documents cannot be rendered from the site”. The /pdf/ route does return HTTP 403 for that prefix; the sheets are ordinary page images and read. ↩

Backlinks