From time to time, one sees an anthology containing stories by authors all belonging to some similar group -- they are women, black, Jewish, Russian, whatever. And some of the stories, and indeed whole anthologies, may indeed be quite good.
Nevertheless, when I see such an anthology, I start from the assumption that it is probably not as good, on average, as an anthology based on some other criterion (such as subgenre or year of publication). Why do I have such a prejudice -- and am I being irrational?
My reason is simple. The more criteria one applies to a selecting a story, other than its quality, the greater the limitation on the set of stories from which one is selecting a subset. This in turn means that the average quality of the stories in the subset is likely to go down.
Suppose, for instance, that one is creating an anthology, The 100 Best SF Stories Of All Time. This anthology could include very good stories written by anyone at any time. If we assume that 1 million science fiction stories have been written in the whole history of the genre, we could select out 100 stories, even the worst of which would fall into the top 0.01%, and these stories would be very good indeed.
Now let's suppose that we create an anthology, The 100 Best SF Stories By Women Authors Of All Time. If we assume that exactly 50% of all science fiction stories have been written by women, and that women write just as well as men, then this means that the worst story in the anthology only falls into the top 0.02% of all science fiction stories.
If the anthology is retitled The 100 Best Science Fiction Stories Written By Women in 1992, and we assume that 1% of all the science fiction ever written was written in 1992, now the stories range over only the top 2% of all science fiction stories ever written.
If we now change it to The 100 Best Science Fiction Stories Written By Black Women in 1992, and we assume that 10% of all science fiction writers are black, now the stories range over only the top 20% of all science fiction stories ever written.
This develops inexorably from the search criteria. The more narrow the criteria, save that of quality, the lower the average quality of any selection assuming the number of stories selected is equal and that the statistical universe of stories is large enough (if my "anthology" consisted of one story, and if we assume that the best SF story of all time was written by a black female gay Mormon, then The Best Science Fiction Story Ever Written By a Black Gay Mormon Woman would perforce contain that tale, even if that was the only science fiction story ever written by a black female gay Mormon).
Those of you who have been paying attention may note a possible flaw in my argument. Namely, it applies to any non-quality based search criteria, so why am I focusing on gender-ethnic-religious based ones?
Indeed, the most common limitation on the selection of science fiction stories is year of publication. Most of the long-running anthology series limit themselves in this fashion. The next most common limitation is subgenre, often cut extremely fine: so we see Elf Fantastic, Horse Fantastic, Plumber Fantastic, etc. etc. etc.
Year-of-publication fails to bother me for two reasons. The first is that, given the lack of access to time machines, new material becomes available to editors on a time-limited basis. It would be impossible to publish an anthology The Best Science Ficton of All Times Including the Future, because future stories haven't yet been written or submitted to the editors of such a tome. Editors want new science fiction stories, so year is the most obvious means of organization.
The second is that I unashamedly confess to a strong interest in the history of science fiction. Even though the stories in The Best SF of 1971 aren't exactly "new," it fascinates me to see what themes were being explored, and how, forty years ago as compared to today. The existence of series of anthologies of this sort is a useful tool in studying the thematic and technical development of the field.
As for subgenre-based anthologies, these are useful because not everyone is interested in the exact same sort of science fiction. Indeed, the assumption that "science fiction," rather than "speculative fiction" or "space-based science fiction" or "space-based science fiction about colonizing Vesta," is the useful unit of examination is arbitrary: it was hardly decreed by the Muses. Sometimes, I want to see an overview of what's been written specifically about spaceships, or robots, or whatever.
Having said that, I will point out that my prediction holds true. The average quality of the work in the year-based or subgenre-based anthologies is generally below that of the all-time or all-genre based anthologies, and when one combines the two limitations, one often sees the publication of truly execrable work.
Ok, but I admitted that I'm not too much bothered by the year-based and subgenre-based anthologies. So why do author group-identity based anthologies bother me?
Basically, it's because I'm a strong believer in the equality of races, sexes and (to some extent) religions (*). Believing in this equality, I don't think that any very important piece of information is being conveyed to me when I learn that a writer is male, female, black, white, Christian or Jewish. Hence, an anthology which limits its selection by such criteria is reducing the average story quality without forming a group of stories which are in any important way similar.
Now, some will say that it is important to produce such group-limited anthologies in order to give (Group X) a chance, because members of (Group X) have been historically underrepresented in the set of science fiction authors, published science fiction stories, or whatever. Or, alternately, that this group has a special "voice," a unique perspective on things, which should be heard in order to promote a diversity of viewpoints in the genre. Surely, we should give the writers from this group what they deserve? Surely, we should listen to their voice?
The problem with the first argument is that what science fiction writers "deserve" is to have their stories fairly judged on content, which is to say quality and length. Being black, or Jewish, or Armenian, or whatever, does not logically give one a special claim on publication resources. If one is good, then one should and will be published regardless of one's ethnic or gender identity. If one is not good, the one should not and will not be so published.
The problem with the second argument is that it is both polylogistic as regards the genre and monolithic as regards the ethnic groups concerned. Polylogistic, because it asserts that there is some special black (or female, or Armenian) conciousness which cannot be understood by whites (or men, or non-Armenians). Monolithic, because it assumes that all members of such a group will agree on certain matters, and in practice authors who do not show the "correct" attitudes will be excluded from such anthologies.
This gets to yet another problem with the "group identity" anthologies. In many case the diversity of ideological content is less than it would be in a non-group identity anthology, even from members of the specific group. What's more, the group identity anthology will tend to focus on a limited set of themes: I've seen this in, to take the most obvious case, "Jewish SF" anthologies, which tend to focus on the religion of Judaism in science fiction settings. This fact glares so obviously because so many of the "big name" SF writers of the 1940's through 1970's actually were Jewish (**), and hence are abundantly represented in more general science fiction anthologies.
Real and meaningful diversity -- especially in an intellectual field such as a genre of literature -- lies in diversity of ideas, not diversity of group membership. We find excellence by seeking the excellent, not by asking people to show us their identity cards.
(*) Since religions are philosophies, and hence are to some degree chosen by their adherents, I consider prejudice against persons based on religious affiliation far more rational than prejudice against persons based on race or sex. Having said that, I don't consider most religions to be any wiser or sillier than their competitors, so my religious-based prejudice is minimal.
(**) In some cases they used nom de plumes, but in most they didn't. Isaac Asimov was their most prominent representative, and he generally wanted to display his own name in as big letters as would fit on a book cover ;-)