Every Lab List Now Shows the Date It Sorts By

Every list of lab records on this site is ordered by the date we published each record. Most of those lists printed a different date on the card: the date the record was last revised.

Both dates are real and neither is wrong by itself. Put together they made a list that looked broken. A page reading newest first would show August, then August, then January, then August, because the order came from one date and the label came from another.

We counted it across the whole lab. Of 724 records, 530 carry a revision date that is not their publication date, and 494 of those differ by more than a week.

The worst of it was on the page that says so out loud. Our full lab list is headed “Everything, newest first” and tells the reader it is in the order we published it, and every card on it printed the revision date.

We had already fixed this three times: once for Deep Dives, once for Lab Results, once for System Updates. Each fix was written for the shelf in front of whoever found it. So three shelves were right, the others were not, and all three of those sets appear again on the full list, where not one of the fixes reached.

The default is now the other way round. A lab list shows the date it is sorted by, and a view wanting the revision date instead has to ask for it by name. No view asks for it today. The revision date keeps its home on each record’s own page, where it sits beside the record it describes and implies no order at all.

The reason we are writing this down rather than filing it as a tidy-up: the fix that was correct three times running was also the reason the fourth case survived. A rule applied to the page in front of you leaves its own class standing, and the class only becomes visible when somebody counts the pages nobody was looking at.