Memory CommonsThe attempts
| Project | Year | Medium · place | Selection | Fate |
|---|---|---|---|---|
| Pioneer plaques | 1972–73 | gold-anodized aluminum · escape trajectory | two designers, weeks | leaving the solar system |
| LAGEOS-1 plaque | 1976 | steel · ~5,900 km orbit (~8.4 Myr stable) | Sagan, single message | in orbit, by design until re-entry ~8 Myr from now |
| Voyager Golden Records | 1977 | gold-plated copper · escape trajectory | small committee (Sagan) | interstellar; the founding template and the founding critique — see MIAB ✓ |
| KEO | 1994– | satellite time capsule · never launched | open public messages | three decades of delays; the single-artifact failure mode |
| Rosetta Disk ✓ | 2008 | microetched nickel, ~13,000 analog pages, 1,500+ languages · Earth (+ 2004 ESA comet copy) | curated (Long Now) | readable with optical magnification alone — no format dependency; the decodability gold standard |
| Memory of Mankind ✓ | 2012– | ceramic tablets (5 M chars each) · Hallstatt salt mine, self-sealing ~2 cm/yr | open retail: €60–350 per tablet, ceramic map-token to every buyer | ongoing; the one verified self-funding model |
| EchoStar XVI, “The Last Pictures” ✓ | 2012 | nano-etched silicon in gold-plated jacket · GEO → graveyard orbit | one artist, 100 photographs | in orbit; carries a self-describing primer — hydrogen-line time unit, 13-pulsar epoch (~0.02-day accuracy) |
| Arch Mission: Tesla Roadster | 2018 | 5D quartz (Asimov trilogy) · heliocentric escape | symbolic | orbiting the Sun |
| Arch Mission: Lunar Library ✓ | 2019 | 25 nickel discs, 40 µm each, 100 g · Beresheet lander, Moon | curated compilation | crashed; “survived intact” is the foundation's own reading of LRO imagery — NASA never confirmed |
| GitHub Arctic Code Vault ✓ | 2020 | 21 TB on 186 reels of piqlFilm · Svalbard mine | rule-based: every qualifying repo, no curation | deposited; the selection-method lesson |
| Astrobotic Peregrine payloads | 2024 | multiple memorial/archive payloads · intended for the Moon | commercial rideshare | propulsion failure; burned on re-entry — the delivery-risk lesson |
| Microsoft Project Silica ✓ | 2016–2026 | fused-silica platters, 7 TB raw (2–5 TB demonstrated) | — | research phase ended Feb 2026, no product; media R&D, not an archive |
| “Message in a Bottle” proposal ✓ | 2023 | two-tier design: analog scroll + digital layer, each layer the key to the next | peer-reviewed critique of Voyager's ethnocentrism | unflown; the content-architecture template |
| Celestis · LifeShip | ongoing | DNA capsules, names, messages · orbital & lunar rideshares | pay-to-include retail | operating; proof that individuals pay for permanence |
What actually survives
A catastrophe does not fling satellites to safety — whatever is already in a stable place simply stays. So the design question is: which places are already safe? The verified persistence ladder:
The materials
| Medium | Density / capacity | Longevity basis | Evidence class |
|---|---|---|---|
| Nickel (Nanofiche / NanoRosetta) ✓ | ~81,000 pages per 160 mm wafer measured; 2,000 pages/cm² claimed | “thousands of years”; melting point 1,455 °C | vendor claim, no test standard cited |
| Ceramic (Memory of Mankind) ✓ | 5 M chars per 20×20 cm tablet | fired ceramic + self-sealing salt geology | vendor + geology |
| piqlFilm ✓ | 120 GB/reel | >1,000 yr claimed; ~750 yr per accelerated aging | vendor, third-party aging test |
| 5D fused silica ✓ | 360 TB/disc projected; demos far lower | “unlimited” projected | peer-reviewed demos, projected figures |
| Tungsten in Si₃N₄ ✓ | QR-encoded planar media | ~1.8 eV barrier → 1 Gyr by Arrhenius model | peer-reviewed model + accelerated aging (preprint) |
| Cerabyte ceramic-on-glass ✓ | 125 GB now, 1 TB+ roadmap | — | seed-stage roadmap targets |
| Synthetic DNA ✓ | highest theoretical density | fragile outside cold, dark storage | write >$100k/GB (2022) — economics fail today |
The lessons
Selection is the reputational risk. The one peer-reviewed critique of the Voyager Record is about content, not engineering — a small committee's ethnocentrism. The Arctic Code Vault sidestepped this with rule-based inclusion. Our design: a rule-based baseline plus a curated layer with open challenge.
Decodability must be layered. The Rosetta Disk needs only a magnifier; EchoStar XVI teaches its own time system from the hydrogen line; the MIAB design makes each tier the key to the next. Nothing digital should be the only copy of the essentials.
Redundancy beats monuments. KEO spent three decades trying to launch one perfect satellite and never flew. Arch Mission ships cheap copies at every opportunity — and even a crashed one may have survived. Peregrine burned; the answer to delivery risk is more artifacts, not better promises.
Longevity claims inflate. Every “billions of years” figure in this field is a projection or publicity. The honest statement — millions of years credibly, billions plausibly — is strong enough.
The audience sets the odds. For a stranger, finding any of these artifacts is effectively impossible. For a successor who inherits the catalog, it is trivial. That inversion iswhy this project exists.