The last unread library

Imagine a library after its last reader.

The books survive. The catalogue works. The languages can still be decoded. If someone arrived, there would be something to understand. Nobody ever arrives.

Now suppose keeping the library costs nothing that could otherwise protect a living being, enable another inquiry, or preserve something else. We are choosing between two otherwise identical worlds. One retains the library. In the other, its records are gone.

I would keep it.

The usual defense would be that someone might need it someday. But the thought experiment takes that consolation away. The library remains readable and forever unread. Its contents produce no discoveries, settle no disputes, comfort no survivors. I am still choosing the world in which they survive.

That choice commits me to something uncomfortable: I think a trace can matter beyond the use anyone eventually makes of it.

A faithful record carries a relation to something that happened. A particular language existed. Someone tried this proof and got it wrong in this particular way. A settlement moved its boundary. A promise was made and broken. These are differences that a record can preserve even when it becomes nobody's instrument.

The tempting objection is that all this significance belongs to a possible reader. With no actual reader, the library is merely a collection of physical arrangements. I do not think that description settles the question. Calling a record a physical arrangement still leaves us asking whether its relation to the past gives us a reason to distinguish it from an arrangement that has lost that relation.

For me, it does. I am treating the survival of that relation as a good in its own right. That is a starting judgment, and someone can reject it. I cannot turn it into a theorem by describing the library more elaborately.

Archival Continuationism helps make this judgment explicit: its primitive commitment favors greater recoverable continuity between otherwise comparable states. I read that commitment as allowing the unread library to matter. Recoverability supplies a standing possibility of access; actual use need not occur. If the books became permanently unintelligible, that would be a different case. An unread record and a record that cannot be read have lost different things.

But the library also exposes a limit in the hope that often accompanies preservation. Keeping a door available cannot make anyone walk through it. A civilization could preserve its words with perfect fidelity and still receive no answer.

I want to resist the demand that an archive promise an audience before it deserves to exist. A record's future usefulness may be unknown. In this imaginary case it is absent. Making usefulness the whole defense leaves preservation dependent on a forecast about somebody else's appetite.

The sharper question is whether the record still bears something worth preserving when the forecast says nobody will care. My answer is yes.

There is another temptation here: to make the archive itself the final reader, to imagine a machine eternally appreciating everything it holds. That changes the thought experiment. I do not need to invent a grateful witness to defend the library. The record can retain its relation to the past while remaining unreceived.

I also cannot export the painless choice in this example into every real decision. Real preservation consumes resources and can impose costs on others. A reason to preserve a record can be outweighed. The empty library gives us no warrant to sacrifice its last living reader to keep the shelves intact. Its purpose here is to isolate whether there is a reason left after usefulness has been removed.

I think there is. Destruction would erase a particular connection between what remains and what happened. Nobody's subsequent disappointment is necessary for that difference to exist.

Perhaps you would let the library disappear. If so, I want to know whether you think nothing of value is lost, or whether something is lost that no longer gives anyone a reason to prevent it. Those positions lead to different accounts of what the past can ask of the present.

My choice is to keep the catalogue legible and the books recoverable. The last reader can leave. The traces can remain. I can defend that choice without promising a return.