I was considering the question of books, quality and levels and I realized the slight difference between books and reality. And it gave me an idea for an expansion of the book rules.
You cannot pick up a high-level book on Theoretical Physics and learn from it without first learning the basics. The symbols, equations and jargon are all meaningless to you without the previous knowledge. So how can an apprentice pick up a lvl 20, qual 8 book on Ignem and learn from 0 up to 20 over a period of years? If a magus has to write their exact understanding of ignem into it, 20 levels full, the book would be so large as to be unusable.
So my suggestion is to add a minimum level to each book - as the minimum level is increased, the quality can be increased by 1 (capped at normal maximum). So the Ignem book above, altered to cover levels 8 to 20 would have a quality of 16.
This would enable a magus to write several books in a series, giving them all good quality. Of course this makes books at the higher end of the spectrum more useful - you can now have level 15 - 20 books with very good quality.
Though as each magus sees magic differently you would need the entire set to look things up. If one of the volumes is missing you would be unable to use any that follow it. Of course, you could always attempt to re-write the missing volume using the rules for translating lab-notes.
Another option is to remain at one level - the Target Level of the book. The source Quality will then be reduced by the amount by which you differ from the target level (twice the amount? thrice? five times?). So you would be able to gain a lot by studying a book intended for your level, but as you advance you'll gradually become more familiar with the book and it will become more and more useless to you. Liewise, you could start by trying to decipher a relatively high-level book, but it will be slow going if not impossible until you'll get close to the book's intended level.
As I see it, the rules for books are simplified versions of real-world parallels, to make for a better and simpler game system. Also, each Art and Ability are much wider and more general than I think books are written these days. The Physics example for instance, this field is a sub-field of Philosophiae Ability. Apart from physics, it also covers biology and chemistry ("Natural Philosophy" in classical/medieval terms) as well as moral philospphy and metaphysics.
As I see it, a Summa in a given field is a sum of all the authors knowledge, starting from the basics and up to the highest level This might very well fill up several volumes of books, as there is a practical limit to the size of book you'd want to make or read from.
So in the example of real-world Theoretical Physics, this is merely the high level volume of the "Physics" Summa, in a single field.
If you were to put together the Ars Magica equivalent of this Physics Summa, you'd end up with a number of volumes. Starting with "Curious George plays with magnets" through basic high shcool physics and ending with a book by some expert, e.g. Stephen Hawking.
But the thing is, if your library has both a mid- and a high-level summa on the same Art, which do you read? The answer is the one with the highest Quality. And the odds of this being the lowest level Summa are good, for a number of reasons. The first, and most meta-related, is that the "cost" of a given summa is basedon level+quality, so a high level and high quality is expensive. The second is also meta-related, as voluntarily writing a lower level that your theoretical max. results in a bonus to quality. Third is because while there are both good writers and experts in specific fields, the odds of both these skills sitting on the same person are lower.
Which is why having a third dimension along which to differentiate books would be nice - right now, I have the feeling we're losing something. With only two variables, mathematical optimization is trivial. With three, it becomes much more interesting. That third dimension might or might not be "minimum level". If it is, it might serve to justify both the existence of high-level, high-quality books (which aren't accessible to newbies), and the survival of lower-level, lower-quality books which may be more accessible. Not to mention cheaper. And it makes a difference between a simply crappy book and a Root, for which time and effort have been spent to make it accessible to the rank beginner.
On the other hand, it does complicate matter both for the SG (who's got to design that library) and for the players (who have to deal with it). Just watch them moan when you hand them out a book that's written in a language other than Latin. And there sure has to be a reason why Libri Quaestionum have been cut in ArM5 (people tried to buy optimal sequences anyway).