The Curation Question: What to Keep and What to Release

The average smartphone user takes roughly 4,000 photos per year. A parent with two children, documenting birthdays, milestones, vacations, and ordinary Tuesday afternoons, might accumulate 40,000 images in a decade. They are all stored. Almost none are lost.

Almost none are found.

The problem with infinite digital storage is not storage. It is selection. For nearly all of human history, families preserved what they could afford to preserve — a few photographs, a handful of letters, whatever fit in the box under the bed. Scarcity imposed curation by default. The question of what to keep was answered partly by capacity, partly by cost, and only partly by conscious judgment about what mattered.

That constraint is gone. And with it, the forced act of meaningful selection.

What Archivists Have Always Known#

Professional archivists have practiced the art of selection for well over a century. They use a precise term for the judgment-intensive process of deciding what to keep: appraisal. The Society of American Archivists defines it as the process of determining the value of records — and thus their ultimate disposition: retention, transfer, or destruction.

The archival concept of appraisal carries a counterintuitive implication that most families never encounter: a collection is defined as much by what it excludes as by what it holds. The act of selection is not a loss. It is an interpretation about what is significant. A well-appraised archive is not the one that kept the most — it is the one that kept the right things, understood to be the things with enduring value to people who were not present for them.

Archivists make a further distinction worth borrowing. Between primary value — the original reason a record was created — and secondary value — the record's usefulness to later researchers and family members who weren't there when it was made. A photo of a toddler eating birthday cake has genuine primary value: pure joy, captured in the moment. Its secondary value, to that child at thirty-five, may also be substantial: this is who I was, this is what we celebrated, this is whose arms were around me. A photo of a parking lot, taken accidentally while fumbling with a phone, has neither primary nor secondary value worth preserving.

The archival question is always: what is the future value of this object to someone who wasn't present for it? Most families have never been asked to think in those terms. They have simply kept everything, because everything is now keepable.

The Argument for Forgetting#

In 2009, digital rights scholar Viktor Mayer-Schönberger published a provocation titled Delete: The Virtue of Forgetting in the Digital Age. His central argument ran against the instinct of most technologists: the ability to forget is not a flaw in human memory. It is a feature.

Biological memory forgets by default. Digital memory remembers by default. The human system evolved to keep what matters, fade what doesn't, and allow the past to recede to an appropriate distance — creating room, cognitively and emotionally, for present experience and future planning. Digital systems do the opposite. They preserve everything with equal fidelity, imposing a kind of total recall that serves storage capacity rather than human flourishing.

Mayer-Schönberger's argument has particular force when applied to families. A family that has never deleted a photo, never discarded a redundant capture, never made any conscious choice about what its memory archive should actually represent, does not have a richer history than a family that has. It has a larger hard drive. The richness comes from selection — from the act of deciding what matters enough to carry forward and what can be allowed to go.

This is not a comfortable argument. Loss feels like loss, even when the thing lost was never particularly meaningful. But an unmanageable archive carries its own weight — the weight of inaccessibility, of opacity, of the inability to return to the moments that actually shaped who you are, because those moments are buried in forty thousand images that look, from a thumbnail view, almost identical.

Abundance Creates Its Own Problem#

Roy Rosenzweig, a pioneering digital historian who helped establish the Roy Rosenzweig Center for History and New Media at George Mason University, wrote searchingly about what abundance does to the historical record. In work that anticipated many of the challenges families now face with personal archives, he argued that the shift from scarcity to abundance in documentation was not an unambiguous gain. Abundance without curation produces noise, not richness. The signal — the genuinely significant, the truly irreplaceable — becomes harder to find as the surrounding volume grows.

The Library of Congress has encountered a version of this problem at enormous scale, grappling with what to preserve from the vast and rapidly expanding digital record of human activity. Their digital preservation programs reflect a recurring insight: the challenge of digital preservation is not primarily technical. Keeping bits alive is largely a solved problem. Deciding which bits are worth keeping is the hard work — and it requires human judgment, institutional commitment, and explicit criteria for what counts as significant.

For families, the practical consequence of this is visible in any family group chat. Someone shares a hundred photos from a vacation. Some are beautiful; many are redundant; a handful are blurry or accidental. Because the beautiful ones aren't labeled, organized, or distinguished from the rest, they are experienced collectively — and the experience of scrolling through a hundred indifferent images diminishes the power of the ten that genuinely matter.

The archive that holds everything holds nothing particularly well. Meaning requires selection.

The Paradox of Curation Paralysis#

Knowing that curation matters does not make it easy. Most families find the prospect of curating their existing archive so overwhelming — so open-ended, so emotionally loaded, so vast — that they never begin. The task of sorting ten years of photos feels impossible as a single project. It probably is. And the suggestion that families simply "curate better" — sort through their decades of images, label the significant ones, delete the rest — asks for a level of sustained attention and emotional labor that most people cannot realistically commit on a recurring basis.

This is where the perfectly curable meets the actually impossible. Curation matters. Curation is also, as described above, practically undoable at the scale most families have accumulated.

The way out of this paradox is not to ask less of curation but to distribute the work differently.

The Middle Path: AI-Assisted, Human-Overseen#

The most defensible approach to family curation combines two things: AI triage for the obvious cases, human judgment for everything else.

AI can reliably handle the mechanical work of triage that most families never do:

• Duplicate identification and grouping. When three nearly identical photos of the same candle-blowing moment exist, a human shouldn't need to compare all three side by side. AI can surface the best of a burst and group the rest for confirmation or deletion.
• Low-quality flagging. Severe blur, accidental captures, exposures that reveal nothing useful, photos of the floor or the ceiling — these are unambiguous candidates for deletion that don't require human agonizing.
• Organizational clustering. Grouping by face, date, event, and place transforms a flat scroll of 40,000 images into a navigable structure, dramatically reducing the cognitive cost of human review.

What AI cannot do, and should not be trusted to do alone, is judge significance. The blurry photo of a grandparent who is no longer living may be the only image that captures her particular way of laughing — the slight tilt of her head, the way her eyes nearly closed. An algorithm has no access to that knowledge. The photo of an ordinary kitchen, unremarkable by any objective measure, may be the one a grandchild wants most — because that kitchen was where everything important happened for twenty years.

The appropriate human role in curation is not to review every photo but to make judgment calls about the ones that AI identified as borderline, and to annotate the ones that have secondary value not visible in the image itself. A caption that says "the last Thanksgiving before Grandma moved to assisted living" transforms an ordinary dining-room photo from storage into memory. Without that annotation, the image survives but its meaning doesn't.

What Families Are Really Deciding#

When a family chooses what to keep, they are not making a decision about storage. They are making a decision about identity. The archive a family builds and tends is a statement about who they are, which experiences shaped them, and which parts of their shared history they want to carry into the future.

A thoughtful selection — smaller than everything that was captured, richer than what would survive by accident — is a more honest representation of a family's life than a complete archive would be. It reflects judgment, not just technology. It says: this mattered to us. This is what we want to remember. This is the story we are telling about ourselves.

The village that held family memory for most of human history was selective by necessity. Stories that didn't get retold disappeared. What survived was what survived the test of retelling — what remained useful, meaningful, and worth passing on across generations. The digital archive that now does the village's work can aspire to the same standard, not by accident but by design.

Curation is not the enemy of memory. It is memory's most honest act — the acknowledgment that not everything deserves to last, and that knowing the difference is what makes a family's story a story rather than simply a record.