Fable
The editorial department’s two ledgers
A newspaper's editorial department had to produce more than a dozen headlines and leads every day. In the early years, only one old editor managed everything. He kept a large notebook, writing today's headlines on the left and noting on the right who changed them and when. By the end of the month, he could always find a few headlines that differed from last week's but no one knew when they had changed.
After the newspaper expanded its pages, it hired four young editors, and the big notebook became a shared document. Four people edited the same headline area at once—you'd change half, I'd change it back—and the headline flipped back and forth five times in a single day. When the editor-in-chief wanted to review, he found the headline had been overwritten three times, and couldn't even find what was written at first review.
The editor-in-chief's office simply split the notebook into two. The first was called the assignment log, recording who drafted each version of a headline, who reviewed it, and who signed it off for publication, governing the process of a headline from idea to print. The second was called the revision record, recording only snapshots of successive changes to headline content, without caring who approved them.
The New Year's special issue had an incident once. The title of an editorial was changed at the last minute by the editor-in-chief before publication, but after making the change, the editor-in-chief forgot to notify the layout team. The layout team still used the previous version of the title. The next day, readers found that the print and digital versions had different titles, and complaints flooded the hotline. Checking the workflow log, the editor-in-chief's sign-off was on record; checking the revision history, the snapshot the layout team received was two hours earlier than the signed version.
In the postmortem, neither ledger was wrong—the job log recorded who approved which version, and the revision record faithfully stored the content of each version. The problem was that there was no linking mark between the two ledgers: no one recorded which revision snapshot the editor-in-chief's signature actually pointed to. The job log had a signature, but the signature did not match any revision timestamp.
The newspaper later added a line between the two systems: each time a sign-off occurred, the job log automatically recorded the corresponding revision snapshot number. From then on, when investigating a headline incident, they'd first check the job log to find the sign-off decision, then jump to the revision record to confirm which content was actually signed off. The two ledgers each kept their own records, but the threads were connected.
At the year-end review, the editor-in-chief wrote a line in the summary: the process governs people's actions and decision points, while the history governs the time series of content states. The two records can live in the same toolset, but don't expect one record to answer both who approved it and when the content became this way. The query directions of these two things are orthogonal.