[ilrwbwrkhv]

I think it can be done far cheaper than you are imagining. The way LLMs are evolving now, I think they are great for exactly this thing.

Given text X = Create derivates of X.

The way people are using LLMs now is -> given limited number of words -> generate more words, which creates garbage.

But the compression idea is where LLMs really shine and should be explored to disrupt the industry.

And other pricing models can also be explored such as a netflix like subscription etc.

edit:

Just to further play with that idea, I think a big reason why https://standardebooks.org/ are quite popular is because they are simply making these books readable on devices. Just that much value is enough for the market.