A row can tell the truth
Suppose a system stores this accurate line: “A visitor said she dislikes being hurried.” We should be glad it kept the line. Without a record, the next conversation may begin in ignorance. But a table can preserve those words for years while every reply still rushes her.
Follow the record forward
Imagine two uses of that row. One service retrieves it only when asked to quote the archive. Another brings it into the right encounter, knows who said it and when, and leaves space before pressing for an answer. If the visitor later says she has changed her mind, it records the correction rather than treating the first sentence as permanent law.
The difference is not the storage format. It is whether the past can be attributed, interpreted, revised, and allowed to matter in a later choice. The second service might still get that choice wrong; its behavior gives us something to examine.
Pause here: If the row vanished from the next decision, in what useful sense did this system remember the visitor?
Memory can also point toward tomorrow
PHS’s Engram Spine design includes intended events as well as completed ones. “Visit the shop at three” is a plan, not an account of having gone. If the visit occurs at 3:33, the later event should be linked to the plan without silently changing the plan’s time. That difference may explain a missed appointment—or may simply be unimportant. Either way, the record keeps its tense.
My answer: Rows are valuable evidence. PHS aims for memory that also preserves ownership, time, and meaning well enough to influence what happens next—and to show where that influence failed.