“Get the next article out, please.”
The automated publishing system clocks in. First it checks the draft list, the review list, the translation list, and the publication list. Then it rebuilds its status records, saves them, and checks whether all the numbers match.
About 23 minutes later, the preparation has swallowed the run. The actual article work hasn't started.
That comes from one anonymized execution record: roughly 8 minutes preparing the shared state and another 15 saving it before the process was forcibly stopped. It isn't an industry average.
Imagine a factory where everybody signs in, attends a meeting about attendance, signs the attendance sheet, and goes home without producing anything. Now imagine the factory was supposed to run without people.
One spreadsheet hires another spreadsheet
Tracking progress is useful. You need to know which article exists, which one passed the necessary checks, and which one still needs publishing. Otherwise workers may duplicate each other's efforts.
But then the lists multiply: drafts, reviews, translations, upcoming publications, reports for every worker. If two reports disagree, somebody must reconcile them. Eventually, there's a report for the reports.
“Where's the article?” “We're updating the tracker.” “Are we done?” “We're checking the tracker's update history.”
The accounting department built a robot. The robot wants to become an accountant.
The factory closes to protect the paperwork
Things get stranger when changing a progress record counts as changing the entire set of publishing inputs. The publisher starts rebuilding everything. Another worker updates another record midway through, so the publisher starts over.
To break the cycle, the system may temporarily stop other workers from writing while publication is underway.
The paperwork was meant to help articles get out. Now the articles have to wait so the paperwork stays safe.
It's like shutting down a supermarket checkout while an employee recounts the people waiting in line.
Division of labor isn't division of forms
The essential jobs are straightforward. The writer saves a finished draft. The publisher checks that draft and makes it available to readers. The observer confirms the page can actually be opened.
Each job should be able to finish without requiring every department to update one enormous shared document.
Ten people waiting for one official stamp do not become more productive when you hire another ten people to join the queue.
Can we delete the ledger altogether?
Many duplicate reports can go. That doesn't mean forgetting the information that prevents real mistakes.
You still need to know which version of which article is already live, and where to resume after an interrupted run. Otherwise you risk duplicates or overwriting fresh work with old material.
But if the published site and existing storage already tell you those things, why copy the answers into five more spreadsheets?
Check the article that changed. Reuse results that haven't changed. Let monitoring observe publication rather than hold it hostage.
The meaningful score is how many new articles people can actually read, not how many status sheets got updated.
“Today's report: all spreadsheets perfectly reconciled. Articles published: 0.”
Congratulations. You have built a world-class filing cabinet.
