Editors’ note: In light of reports that, in his testimony yesterday, Lt. Col. Vindman said he tried to correct the record of the July 25 phone call, we are reposting this piece by James Goldgeier, written before the call record’s release, on how such records are produced and why they cannot be treated as verbatim transcripts.

What did President Trump say to Ukrainian President Volodymyr Zelensky in the now-infamous phone call flagged by a whistleblower? Did he imply that Ukraine wouldn’t get its promised military aid unless the government would investigate whether former vice president Joe Biden tried to protect his son Hunter from a Ukrainian corruption probe? House Speaker Nancy Pelosi announced a formal impeachment inquiry Tuesday based on reports that Trump did — and, in response, the president has said he will release a “complete, fully declassified and unredacted” record of the call, something everyone has been calling a “transcript.”

Of course, the phone call is only one part of this story. The administration has also been refusing to give Congress — as required by law — a whistleblower complaint alleging that Trump made an improper “promise” to a foreign government. By unanimous consent in a voice vote on a nonbinding resolution, the Senate called on the Trump administration to release this complaint to the Senate Intelligence Committee.

Nevertheless, the release of the record of the call will be major news. It is important to understand how documents like it come to be produced — and to remember that this is just one document of one moment in a longer series of events regarding the release of U.S. assistance to Ukraine.

When a ‘transcript’ isn’t a transcript

Here at TMC, Henry Farrell has noted that Trump may well have signaled what he wanted in the phone call without directly saying it, using “a form of communication that has much in common with the language of Mafia bosses.” But even when a president does not speak in “Mafia code,” there are still plenty of reasons to be wary about treating a memorandum of conversation as “truth.”

The word “transcript” implies a verbatim record, the kind someone might get from transcribing a recording. But as far as we know, no White House has recorded phone calls since Richard Nixon did so, since the Nixon tapes led to his downfall during the House impeachment hearings.

How are these records produced? When I served as a note-taker in 1996 on phone calls between President Bill Clinton and Russian President Boris Yeltsin, typically several of us from the National Security Council staff sat in the White House basement with headphones on, taking notes that we would then use to produce a formal memorandum of telephone conversation. (Last year, the Bill Clinton Presidential Library released the records of nearly all the Bill and Boris conversations.)

Note-taking is an art, not a science

The quality of the notes varies, depending on the number and quality of the note-takers. It might sound easy. It isn’t — particularly if there is only one note-taker trying desperately to make a record of the conversation. Normally that record would list who the note-takers were; look for that fact in the record when it’s released. If there is more than one note-taker, they confer and go over their notes to produce the most accurate account possible; the greater the number of those taking notes, the more likely the final memorandum can be relied upon as accurate. Often there are one or more officials in the room with the president, listening in on the phone call. If so, they might be listed as well on the formal memorandum.

When the notes are typed up to produce the formal memorandum, that draft memorandum goes up the NSC chain for formal approval for distribution to a select group of officials. In that process, one of the top NSC officials could direct the staff to take certain material out of the memorandum — because once it goes out for distribution, a number of officials, including those at other agencies such as the State Department, will see the record. Given the chaotic nature of the NSC under Trump, we do not even know what kind of formal process has been followed for producing these memorandums.

But here’s the key point: The record we will see should not be called a transcript. Beyond any coded way of speaking that the president might have, the record could be misleading in any number of ways. The notes may not have captured the exact wording — not for nefarious reasons but simply because of the note-takers’ limited capacity to write as they listen, especially if only one person was taking notes. Certain things may have been removed from the record before it was approved for distribution within the government. And certain things may be removed before the president releases the memorandum publicly, regardless of the president’s assurance that it will be “complete and unredacted.”

Anyone who has produced documents or who has done archival research knows that documents are not absolute truths. They are windows into events captured by individuals who wrote and approved those documents for the records. They are usually written not for the public, but for the government’s internal use in conducting its business.

If this memorandum of telephone conversation is in fact released to the public, it will be just a single document freighted with all the challenges involved in understanding such materials. Its release is unlikely to end the debate. Instead, expect it to provide fuel to the ongoing partisan fire.

James Goldgeier (@JimGoldgeier) is Robert Bosch Visiting Senior Fellow at the Brookings Institution and a professor of international relations at American University.