What if it's a "Lossy Encoder" for 15th-Century Latin?

[Pointless..]

There is no point in this procedure.
With such text, a conventional decipherer won’t be able to understand it properly.
But in reality, there is a possibility that the encryption algorithm may distort the text, and the decipherer will end up with something different from what they would have wanted to see. In the case of a book that has remained undeciphered for about 600 years, this is indeed possible...

Think of it this way. I have a writing system that does not record every detail of the original text. Instead, it preserves the essential information and adds markers or hints where the meaning might otherwise be unclear.

The intended reader has instructions explaining how the system works, how the markers should be interpreted, and how the missing information can be reconstructed from context. The result may not reproduce the original wording exactly, but it should allow the reader to recover the intended meaning.

Anyone who does not know the rules would see only an unfamiliar and ambiguous text. They might recognize patterns, but they would not know what had been omitted, how the remaining elements were organized, or how the markers resolved ambiguities. The decoding procedure would therefore function rather like a key, making the text difficult for an outsider to interpret.

This could also provide a degree of protection against unintended readers. The secrecy would not lie only in the symbols themselves, but in the combination of the writing system, the reconstruction rules, and the specialized knowledge needed to apply them.