“Heavily redacted” records in the ARC

[Note that some material linked in this post has moved from my Google Drive to Dropbox, I have changed the links and refs for these.]

This post discuses my recent attempt to count redactions in the John F. Kennedy Assassination Records Collection (JFKARC) at the National Archives and Records Administration (NARA).

This attempt is summarized in an excel spreadsheet available here. For more on the spreadsheet and the records it represents, see here.

All of this deals only with CIA records; redactions in records from other agencies, such as the FBI, probably differ from CIA records. I may post something on these other redactions, but it will have to wait until after the next round of releases, now scheduled for December 2022.

Redactions and withholding distinguished–again

According to the latest published count from NARA, there are still 14,236 records at NARA that have text redacted.1The count is at https://www.archives.gov/research/jfk/processing-project-2021. This takes into account the 1491 records which shed their last redaction in December 2021. “Text redacted” means these documents are available to the public, but some of the text is held back.

According to NARA, there are also 515 records that are withheld in full. This means the records are not available to the public, except for some basic metadata, such as how many pages, document dates, etc. Almost all of these records are tax returns. The JFK Act, which established the JFK ARC, specifically exempted this sort of information from release.

These two types of documents, redacted and withheld, are frequently confused. As a result, there is often confusion in on-line and journalistic discussions over exactly how much material is not yet released in the ARC.

In an effort to clarify this confusion, I recently uploaded several dozen pdf compilations of redacted records to my google drive, and in my last note posted a brief explanation with links to the compilations and documentation.

Examples of redactions

Those who are actually interested in looking at this material should check my dropbox folder. If you are at a loss what to download, I suggest getting the introduction, the spreadsheet with the basic metadata for the documents, and the first pdfs of one page, two page, and three page documents.

Using the spreadsheet, one can sort on pages and number of redactions/document. This gives some interesting results. For example, there are over 2500 one page documents with just one redaction apiece. This is one quarter of all the redacted CIA documents. A quick look at the redactions shows that most of them are short indeed, consisting of single words or short phrases.

Redactions are now indicated in ARC documents using blank rectangular boxes in documents. Following is an example. Careful, there are actually two redactions in this document.

From the context, both of these redactions are clearly people’s names or titles, and this is true for most of the redacted material.

If you want further confirmation of what is being redacted many redactions also have “substitute language codes” written next to them, which tell us the type of word or phrase redacted. Following is an example from a cable which gives the substitute code “12-23”.

“12-23” is a code for a “cable prefix” and gives the location of the CIA station which originated the cable. CIA generally redacts this information due to the political sensitivities of host countries, and also no doubt for operational reasons, since giving the location where events take place can reveal much of the context of sensitive cables.

In this case, “Dawn” and “Pak Times” are newspapers published in Pakistan, so that one is certainly tempted to identify the origination of the cable as probably Pakistan and perhaps Karachi. There is always a little uncertainty over such deductions, however, (maybe Lahore? maybe Islamabad?). This, of course, is the point of the redaction: CIA wants to leave wiggle room.

Redactions, light and heavy

Documents like these I call “lightly redacted”. The amount of text held back is small, minute even. On the other hand, there are, of course, documents that have more than a single redaction per page, sometimes a lot more.

There are, for instance, something around a hundred CIA records which have whole pages redacted. There is no way of telling exactly what is being held back there. There are also documents that have big redaction boxes, which people, sometimes correctly, sometimes incorrectly, believe cover a lot of interesting stuff. There are, however, surprisingly few of these. Instead, we mostly have documents with lots and lots of single item boxes. These are what I call “heavily” redacted documents. How many is “lots”? There is always a subjective element to such counts, but certainly a one to three page document with 30 or more redactions must count as “heavily redacted”.

Here are some examples of what I think anyone would call heavily redacted documents. A lot of these are just lists:




Several of the more interesting “heavily redacted” docs are memos from the CIA to the ARRB, the federal board which assembled the JFK collection. These memos appeal for the Board to hold back various types of sensitive information, and are of course themselves heavily redacted:



Other documents are chapters in old stories with names still left out, as in this excerpt from the tale of Herbert Itkin:


With the exception of the ARRB-CIA memos, explaining why these documents were put into the JFK collection would take multiple, lengthy posts. Most of them have only the most distant relationship to the JFK assassination. The redacted material here adds nothing to our knowledge concerning even the most peripheral aspects of the assassination and the government investigations thereof.

A quantitative approach

Perhaps you find it hard to believe that all these records are so peripheral to the JFK story. Well, let’s count. How many of these “heavily redacted” CIA docs are there left in the ARC? Limiting ourselves to the one, two, and three page documents in the compilations I put on-line (and remember that these make up 80 percent of all redacted CIA documents), there are only 20 documents with more than 30 redactions. How many 1-3 page documents with more than 20 redactions? 52. How many with more than 10 redactions? 250. And how many with 10 or less redactions? About 7800.

This quantitative approach to redaction in ARC documents has convinced me there just can’t be that much information left concealed, especially when one considers the large number of redactions that cover simply the name of a CIA employee, or the location of a CIA facility.

To repeat, much of this redacted information has only the most marginal relevance to the assassination of President Kennedy. The attention that these redactions have received in the press and online is disproportionate, to say the least. Just in terms of the amount of text here, these marginally relevant redactions are a tiny fraction of the total text which is available for public inspection.

Print all these redacted words/phrases/prefixes/numbers together in a single document and how many pages would it take? Less than a hundred, perhaps. Certainly no more than a couple of hundred. Print all the text that has been released up until now in a single document and it would take easily over a hundred thousand pages.

Others, however, take the mere fact of a redaction as having special significance. My question is still this: why is so much meaning being attached to the redacted words, and so little to the released words?