Pruning and Retention

A long record accumulates. At some point the question arises of whether to keep all of it, and the answer differs from the one that applies to most files.

To try the approach with real entries rather than memory, start with this time-management reference. The result should remain easy to correct, export and interpret after the immediate details have faded.

Why the default is keep

It is small. Ten years of entries is a few megabytes. Storage is not the constraint and has not been for a long time.

A related source worth keeping beside this note is Jira, particularly for decisions involving issue histories.

Value increases with age. A 2016 entry is worth more now than it was in 2016, because it is now part of a comparison that did not exist then.

And deletion is irreversible in a way that no other operation here is.

For the data itself, the default should be permanent retention.

What to prune anyway

Duplicates from sync failures, which corrupt totals.

Entries you know to be wrong, marked rather than deleted where possible.

Test data from when you set up a tool.

And attachments or screenshots, if your tool allows them — these are what actually make a record large, and they age badly.

What the retention question really concerns

Not the numbers. The notes.

Notes accumulate detail about other people, about difficult periods, about health and relationships and things said in confidence.

Ten years of that is a document of some sensitivity, and the boundary section covers what follows.

This is the part where a retention decision is genuinely warranted.

A workable policy

Numbers: keep indefinitely.

Notes: keep, with a review. Once every few years, read the oldest year's notes and remove anything you would not want read — about others particularly.

Attachments: keep two years, then remove.

This takes an hour every few years and it is the only pruning most records need.

The archive split

At some point the working file becomes unwieldy.

Split by year rather than pruning: the working file holds the last two or three years, archives hold the rest, all in the same format.

Analysis across them is a concatenation, which any tool handles.

This solves the weight problem without losing anything.

What people delete and regret

The years that look bad.

A period of illness or difficulty, deleted because reading it is unpleasant.

These are frequently the entries people most want later — for a claim, for a comparison, or simply to see how long something took to pass.

If reading it is the problem, move it to an archive rather than deleting it.

Inheritance

Uncomfortable and worth a sentence: a multi-year personal record is a document somebody may read after you.

Decide whether that is acceptable. If not, that is a reason to prune the notes rather than to stop recording.

Most people have never considered it and it takes one decision.

What to check

How large is your record actually — is weight a real problem or an impression?

Have you ever removed duplicates?

When did you last read the oldest year's notes?

And is there anything in there about another person that you would not want read?

The point

The retention question is about the notes rather than the numbers.

Ten years of notes is a document of some sensitivity and that is where a decision is genuinely warranted.

Worth adding

Split by year rather than pruning when the file becomes unwieldy.

The working file holds recent years, archives hold the rest, all in the same format.

Additionally

People delete the years that look bad and regret it.

If reading them is the problem, move them to an archive rather than deleting.

Finally

Numbers keep indefinitely; notes deserve a review every few years, removing what concerns other people and no longer serves a purpose..

In summary

The default for a long record is permanent retention: it is small, its value rises with age, and deletion is the one irreversible operation available..

Last word

The practice this points at is unglamorous: a small daily act, a short weekly correction, an annual read, and a written page of conventions beside the data. None of it is difficult and all of it is easy to skip, which is why most records do not reach the point where they start paying.