Digital hoarding or archive? When keeping the ZIP is wise and when it's fear

A folder with four 100MB export ZIPs, none of them opened in a year, is a
small monument to an unsolved decision. You do not throw them away because
they might matter; you do not open them because they might hurt; so they sit.
Psychology has a name for the pattern and a surprisingly clear description of
why it reproduces itself.

The machinery of keeping

• Loss aversion, multiplied. Discarding is coded as a loss, and losses

  loom larger than equivalent gains.¹ Digital items sharpen this:
  deleting costs storage worth cents, but the fear attached to the
  irreversibility is priced like losing the contents themselves.

• Ambiguous value. In studies of digital disorganisation, people struggle

  to discard because the item's worth is unclear and future need is
  unimaginable - so the default becomes indefinite retention.² An old
  export is the perfect vague item: priceless if you ever need it, worthless
  until the day you do.

• The cost of keeping is invisible. Physical clutter imposes daily visual

  tax; a zip file imposes none until the drive is full. With no carrying
  cost, there is no pressure that forces the sorting decision.³

• Sentimental authority. Digital objects inherit the emotional weight of

  their content - your entire 2019-2026 conversation history in one file -
  which makes "it's just a file" untrue in exactly the direction that stops
  deletion.⁴

Hoarding becomes a problem - per the research - when the accumulation
itself causes distress or blocks normal use of the space. Four zips are not
that. The useful moment is the earlier one: when you notice keeping has
become a decision you keep not making.

An export is a special case: know what you'd actually lose

Before any rule, the facts about what the ZIP is:

1. It is re-requestable only partly. You can ask the platform for a new

   export any time, but exports are snapshots with sample caps and
   retention windows - the searches from 2021 in this file will not be in
   next year's. Deleting this copy does not delete the data at the platform;
   it deletes this window's view of it.⁵

2. Nothing else holds a copy. The whole local-first posture means no

   server has it: the file on your disk is the only place the derivation can
   ever run from. There is no account, no cloud copy, no "restore" to click.

3. **The derived answers live in your browser's local storage, not in a

   service.** Clearing site data, changing browsers or moving machines means
   the next session needs the ZIP again to rebuild what it showed.⁶

That is the exchange in full: keeping costs almost nothing and carries a
snapshot nobody else has; deleting costs you the window, not your history at
the platform.

A decision rule that actually ends the decision

The rule exists to make the fear specific instead of ambient:

• Keep one full-fidelity master. Choose the newest complete export (the

  checklist post covers which settings make it complete - all-time range,
  both formats where you need them). Encrypt it if your threat model calls
  for it, put it where backups already exist, and date it in the filename.
  One dated file with a purpose is an archive. This is the wise half.

• Delete duplicates and superseded copies. Older exports that the master

  fully covers are the fear, not the archive. Their removal is the point -
  and it is the safe kind of deletion, because the master was checked against
  them first.

• Decide by question, not by guilt. Each remaining file should answer a

  standing question ("this is the pre-deletion copy", "this is the period I'm
  researching"). Files with no question attached graduate to the duplicate
  pile.

• Re-verify, then release. Open the master once - confirm it loads, the

  counts look sane, the period is right - before removing anything. The
  brief discomfort of verification is the price of an unambiguous decision.

• Accept the residual. One day you will delete a copy and later wish you

  had something from it. That is the cost the rule prices in; without it the
  decision never terminates at all.

Related reading

The export-before-you-delete checklist for what makes an export complete
before you rely on it as the master; the keeping-control post for deletion of
everything the product itself stores; the local-first-vs-cloud post for why
"just store it in my account" is not a trade this tool offers.

¹: Kahneman, D. and Tversky, A. - prospect theory and loss aversion
(1979); cited as the framing for why discarding outweighs keeping.
²: Vyas, L. - research on ambiguity of digital objects' value and
indefinite retention of files with unclear usefulness; cited as framing for
digital disorganisation.
³: Studies of digital clutter note the absence of the constant visual
cue physical clutter provides, so accumulation proceeds without friction;
cited as background.
⁴: Sunderland, M. et al. work on emotional attachment to digital
objects; cited as framing for why "it's just a file" fails under sentiment.
⁵: The sample-caps and retention disclosures across this repo -
packages/shared/src/analytics/compare.ts (100-500 search window), the
eight-things post (media and irrecoverables), the request/how-to posts
(all-time range setting).
⁶: SECURITY.md section 1 (zero network calls) and
content/blogs/2026-10-08-local-first-vs-cloud-instagram-export-analysis.md
(local storage, literal deletion, no server-side remainder).