It is indeed a lot of assumptions and also ignores, for example, the fact that those shared and least-interesting features are still, in many cases, what people like most about things, which would make it a good approach for a recommendation engine even though it's a bad approach for critical work. Musical and narrative genres are rife with this kind of 'subscription mentality'[1]: people who simply enjoy the genre and will like almost any entry into it.
> Secondly I can imagine taxonomies which are not based on shared textual features, but on chronological, geographical, and genre features.
Note that these aren't meant to be excluded by the quote, which is from a book that deals largely with the three things you mention. Il n'y a pas de hors-texte and all that.
> Thirdly, I don't see why a shared feature would automatically be the "least interesting". What's interesting is rather subjective and context-dependent. If for example I had a particular interest in literature coming out of the first world war, this shared feature could be at the top of my list.
No doubt, but the quote is specifically concerned with literary/artistic value. Historians or critics reading through a different lens will of course have different interests and see value in different things.
Would you not agree, though, that a particular interest in literature coming out of the first world war is simply a filter for your data set, and that once you get down to specific works, it's how they differ or assert themselves against one another that is most interesting about them?
1. I just made that phrase up; someone may have come up with a better one already. But you know the type: the inveterate sci-fi fan, the fantasy aficionado, the lover of all things zombie. These people are less concerned with the quality of a particular work than with the genre trappings it exhibits.
I think that if your theory is that the quality is in the sum total of all differences with other works, then it's not saying much. On the surface, texts are unique as snow flakes, so that's not a useful perspective. Some differences or features will surely play a larger role than others in determining quality. For the features that matter, commonalities might be more interesting than differences.
I think there might not be an effective difference between a particular interest acting like a filter, and the rest of the differences identifying a work. They appear to be different because the former is explicit and particular, but in the end our taste is conceivably described as a filter.
And I don't see why this couldn't pertain to literary or artistic value. You seem to suggest that shared features automatically imply genre works which are typically deemed of lower value. If artistic value is said to only reside in an ineffable difference of the whole work, then to me this is just a cop out to avoid having to define what that value actually is.
> I think that if your theory is that the quality is in the sum total of all differences with other works, then it's not saying much.
I wouldn't say that quality and value are the same thing. Oftentimes works of questionable quality are incredibly valuable because they open up new domains for other artists to develop.
> On the surface, texts are unique as snow flakes, so that's not a useful perspective.
This isn't true at all. In fact, most people, even trained critics, have trouble telling writers apart, let alone texts. There are exceptions, of course -- mostly the avant-garde, because you know when you're reading Joyce or Pound -- but even then the vast majority of people can't distinguish the real thing from a parody. When I was in grad school a quiz (http://reverent.org/poetry_or_parody.html) made its way through our poetry seminar, and it was hilarious. Two of us missed only one; the rest missed at least six! I just retook it now, three years later, and missed three, so I guess I'm losing my touch.
> Some differences or features will surely play a larger role than others in determining quality.
Of course, but again, quality <> value. Also I'm not sure what you're responding to with this thought. There's nothing about 'sum totals' or anything like that in the Perkins quote, and I don't think I suggested it elsewhere.
> For the features that matter, commonalities might be more interesting than differences.
Do you have an example? The 'might' is important here: I agree that it's possible, but I can't think of an example. Most likely I would think the commonality would be some shared trope that the text then proceeds to spin off in an entirely different direction, so that the commonality is only interesting as staging for the difference. But I can be persuaded otherwise.
> I think there might not be an effective difference between a particular interest acting like a filter, and the rest of the differences identifying a work. They appear to be different because the former is explicit and particular, but in the end our taste is conceivably described as a filter.
I'm not sure if this is what you meant, but I think it's an interesting point: namely, that every difference is a difference from, so by shifting the context in which a work is understood, you've effectively changed what its commonalities with and differences from other works are. And that leads us down the oh-so-tricky rabbit hole of establishing contexts, e.g. via period (Romanticism), tradition (form, genre), and so on.
As far as taste being a filter, that just depends on the person. It's certainly possible to imagine tastes that can't be used as filters because they don't involve surface properties like genre trappings.
> You seem to suggest that shared features automatically imply genre works which are typically deemed of lower value.
I'm not suggesting either of those things. Shared features don't imply genre works (all works have shared features), and genre works aren't necessarily of lower value (they just attract certain audiences regardless of their value).
> If artistic value is said to only reside in an ineffable difference of the whole work
It's neither ineffable nor anything to do with the 'whole' work. I'm sure it seems abstract, with us talking about all of art in one go here, but once you get down to specific texts and contexts any competent critic is going to talk in excruciating detail about what specifically is valuable in a given work. At least, that's the idea -- in practice most professional critics are crap.
That texts are unique is simply an objective fact. As soon as you change just one letter, you have a new text with 'difference'. I'm saying this is not a useful perspective because it only gets interesting when you can point out what kind of difference there is and why it is important, and then you end up with something that higher quality texts may have in common.
While critics may have some difficulty telling authors or texts apart because of human limitations, computers attain very high accuracies with various authorship attribution methods. This underscores my point: if you focus on differences, you will end up identifying very particular things such as a particular author or even the style of a particular work, that's just the idea taken to its extreme.
On the other hand, I believe that the quality of a work, which is one kind of value indeed, can lie in commonalities instead of differences. I don't have a concrete example, but from reading literature I do get the feeling that there is something definite that they share. I don't mean something on the level of a trope, that would be too superficial. I think it's OK to admit that we don't know, but not a good idea to preclude it from being described as part of a taxonomy because it's supposedly only about differences. There's simply too much difference in all the unique things around us for that to be informative.
> Secondly I can imagine taxonomies which are not based on shared textual features, but on chronological, geographical, and genre features.
Note that these aren't meant to be excluded by the quote, which is from a book that deals largely with the three things you mention. Il n'y a pas de hors-texte and all that.
> Thirdly, I don't see why a shared feature would automatically be the "least interesting". What's interesting is rather subjective and context-dependent. If for example I had a particular interest in literature coming out of the first world war, this shared feature could be at the top of my list.
No doubt, but the quote is specifically concerned with literary/artistic value. Historians or critics reading through a different lens will of course have different interests and see value in different things.
Would you not agree, though, that a particular interest in literature coming out of the first world war is simply a filter for your data set, and that once you get down to specific works, it's how they differ or assert themselves against one another that is most interesting about them?
1. I just made that phrase up; someone may have come up with a better one already. But you know the type: the inveterate sci-fi fan, the fantasy aficionado, the lover of all things zombie. These people are less concerned with the quality of a particular work than with the genre trappings it exhibits.