16. The Seal That Could Not Be Forged
Domain: Cryptography — Hashing, Digital Signatures POV: Sable Reading time: ~15 minutes
The order arrived on official magistrate’s stationery, bearing what appeared to be Magistrate Penric’s personal seal and signature, and it instructed the gate-guards of the Middle Ring to open all barriers immediately and unconditionally.
Sable was the one who caught it, because Sable reviewed every official order that crossed her desk now — a habit she had developed after the discovery of the erased entries in the Spire’s access log. The order looked authentic. The paper was the correct weight and watermark. The seal was pressed in the correct position. The signature matched the sample of Penric’s handwriting that Sable kept in her logbook for reference. Everything about the document proclaimed its legitimacy.
And everything about it was wrong.
“Penric would never issue this order,” Sable said to the emergency council, laying the document on the table before them. “He has spent weeks helping us reinforce the barriers. He knows what will happen if they are opened prematurely. This order is either a forgery — someone has created a document that looks exactly like an official order — or a corruption. Someone has taken a legitimate order and altered its contents without changing the seal.”
Penric, who was present at the council, examined the document with the careful attention of someone who had spent decades reading and writing official correspondence. “This is not my signature,” he said after a long moment. “It looks like my signature — the shape is right, the pressure is right, the angle of the letters is right. But it is too perfect. I have a slight tremor in my right hand, the result of an injury I sustained many years ago. It is barely visible, but it is there, in every document I have signed for the past twenty years. This signature does not have the tremor. Someone has copied my handwriting from an old sample and used it to forge this order.”
“So we have a problem,” Sable said. “How do we verify that an order is genuine? If a seal can be copied and a signature can be forged, then how do we know whether any given document came from the person it claims to come from?”
The answer, which Quill unearthed from the same cryptographic treatises that had helped them decrypt the anonymous message, was something called a hash.
“A hash,” Quill explained, addressing the council with the slightly pedantic enthusiasm that he always brought to technical explanations, “is a mathematical function that takes any input — a document, a message, an entire book — and produces a fixed-length output called a digest. The digest is unique to that specific input. If you change even a single character of the input — a comma, a space, a letter — the digest changes completely. And the function is one-way: you cannot reverse the hash to recover the original input. You can only verify that a given input matches a given digest.”
“So if we had a hash of the original order,” Kip said, “we could compare it to the order we received. If the hashes match, the order is genuine. If they do not match, the order has been altered.”
“Exactly,” Quill said. “Hashing provides integrity verification. It does not tell you who created the document or whether they were authorized to do so. But it tells you, with mathematical certainty, whether the document you received is exactly the same as the document that was originally created. Any alteration — no matter how small — will produce a different hash.”
“And if we combine hashing with something else,” Sable said slowly, her mind already working through the implications, “something that proves who created the document… we could have both integrity and authentication. We could know not only that the document is unaltered, but that it genuinely came from the person who claims to have sent it.”
“A digital signature,” Quill said, and the phrase seemed to light up something in his expression. “The sender creates a hash of the document, then encrypts that hash with their private key. The recipient decrypts the hash with the sender’s public key, then computes their own hash of the document. If the two hashes match, the document is both unaltered and genuinely from the claimed sender. The signature proves authenticity; the hash proves integrity.”
The implementation of digital signatures across the Citadel’s administrative systems took a week of intensive work, and it required the fellowship to solve problems that no one had ever solved before.
The first problem was key distribution. Digital signatures required public-key cryptography — each official needed a key-pair, and the public keys needed to be available to anyone who might need to verify a signature. Quill designed a public registry — a single, authoritative source where every official’s public key was recorded and could be looked up by anyone. The registry itself was protected by its own set of signatures, so that no one could alter the registry without detection.
The second problem was the hash function itself. The old hashing mechanisms in the Library’s treatises were functional but slow, designed for an era when documents were few and verification was rare. Sable worked with Tessa and the gear-works engineers to design a new mechanism — a specialized gear-train that could compute hashes quickly and reliably, producing a fixed-length digest for any input document in a matter of seconds. The mechanism was installed in the Archive of Record and at every major administrative office in the city.
The third problem — and this was the one that occupied Kip’s attention — was education. The magistrates and the archivists and the clerks who actually produced and processed official documents needed to understand not just how to use the new system but why it mattered. Kip wrote the instructions — simple, clear, designed for people who had no background in cryptography — and spent two days walking from office to office, explaining the concept of hashing and digital signatures to anyone who would listen.
“Think of it this way,” Kip said to a roomful of skeptical archivists who had been processing documents the same way for thirty years. “Every document now carries two things: a hash, which is like a fingerprint — unique to that specific document, impossible to forge. And a signature, which is like a personal seal — proof that the document came from the person who claims to have written it. If anyone alters the document, even by a single letter, the hash will change and the alteration will be detected. If anyone tries to forge the signature, the public key in the registry will not match, and the forgery will be exposed.”
“We have been sealing documents with wax for centuries,” one of the older archivists said. “Wax has served us well. Why should we trust this new system over the one we know?”
“Because wax can be copied,” Kip said. “A seal can be lifted from one document and pressed onto another. A signature can be forged — Penric’s signature was forged, and it was good enough to fool everyone who looked at it. But a hash cannot be forged, because it is not a mark on paper — it is a mathematical relationship between the document and its digest. And a digital signature cannot be forged, because it depends on a private key that only the sender possesses. The wax seal has served us well for centuries. But the centuries have changed. The threats have changed. And our protections must change with them.”
The new system caught its first forgery three days after implementation.
A order purporting to come from the Chief Archivist’s office — Vale’s office — arrived at the Spire, instructing the Keeper of the Clock to disable the new access controls and restore the old system. The order bore what appeared to be Vale’s signature and seal. But when the Spire’s verification mechanism computed the hash of the order and compared it to a registered hash supposedly provided by the Chief Archivist’s office, the two did not match. The order had been altered — subtly, almost imperceptibly — but the alteration was enough to change the hash, and the change was enough to flag the document as fraudulent.
“Someone forged this,” Sable said, holding the document up to the light. “Either Vale himself, trying to circumvent our new controls, or one of his recruits, operating on his instructions. But the forgery failed. The hash detected the alteration. The signature could not be verified because the hash did not match.”
“So the system works,” Kip said, and there was a note of quiet satisfaction in his voice. “A document can be corrupted. A seal can be copied. A signature can be forged. But a hash cannot be faked, and a signature backed by a public-key registry cannot be impersonated. Integrity and authenticity — two protections that the old system could not provide.”
Sable set the forged order aside and made a note in her logbook. The Shroud was still pressing against the walls. Vale was still plotting. The city was still fighting for its survival. But now, at least, when an order arrived, they could verify — with mathematical certainty — whether it was genuine or whether it was another weapon in the long, slow campaign of corruption that Vale had been waging against the city.
And verification, Sable understood, was the whole point. A wall could be breached. A door could be forced. But a hash — a small, fixed-length digest that represented the entire contents of a document — could not be altered without detection. It was a different kind of protection than the brass barriers and the layered defenses they had built against the Shroud. It was quieter. More abstract. More dependent on mathematics than on materials.
But it worked. And in the end, that was all that mattered.