Let's start at the top (or, rather, given the nature of blogger, the bottom), with Tushnet's basic position, in which she argues:
Marriage developed over centuries to meet several specific, fundamental needs: children's need for a father. A couple's need for a promise of fidelity (and consequences for breaking that promise). Young people's need for a transition to manhood or womanhood. And men's (and women's, but mostly men's) need for a fruitful rather than destructive channel for sexual desire--a way of uniting eros and responsibility....
At best, marriage only addresses one need of homosexual couples: sexual fidelity. Even there, it should be obvious that same-sex couples will be less likely to insist on physical fidelity than heterosexual couples. If your man might make babies with someone else, you're more likely to see the point of restrictions on male sexuality.
I think the above is perhaps too easily dismissive of the needs addressed by homosexuals entering into marital unions, on several levels. Frankly, the only need Tushnet lists that I don't find served in the homosexual variant, as it were, is "children's need for a father." If, as she points out later "The wedding ring is the sign that the careless kid has finally grown up, become a man with a man's responsibilities," if marriage is a sign of adulthood, and people crave this recognition, how is it any less a sign of adulthood and responsibility when the marriage is between same-sex partners? While I'll agree that "Marrying a woman is significantly less proof of one's manhood when a woman can do it," what it is not is any less proof of one's adulthood. And unless Tushnet's classifying homosexual intercourse under "destructive channel[s] for sexual desire," then marriage seems just as capable of joining eros to responsibility regardless of gender-pairing.
On fidelity, I have to say that the fact that a partner's casual infidelity is capable of killing you makes for rather strong encouragement toward finding a means of strengthening fidelity. It may or may not be a stronger practical consequence of infidelity than that of illegitimate children, but I at least contend that it's of equal severity. While a heterosexual union has both concerns to deal with, I think the homosexual community still focuses far more strongly on the danger posed by STD's than its heterosexual counterpart, arguably to an extent which compensates for the lack of "someone might get pregnant" as it concerns marital desires.
Gender (vs.) Identity
The sole, unserved supposed need of marriage by homosexual union, the need of children for a father, is probably where our arguments become hopelessly irreconcilable. If I'm understanding her correctly, Tushnet agrees that gender is a social construct, but whereas I have no use for it, she feels gender (and gender roles) still serve some essential purpose in society:
Children want gender because they want a role in the world, a place in the story unfolding around them, a role to live up to and by which they can judge their actions. Gender fits us into the cycle of family and fruitfulness; it connects us to our parents, who have taken on the most obviously gendered roles of all when they became mother and father; and it provides children with a connection to their future maturity, and thus to sex. Children want a sexual identity even when they do not plan on kids or a spouse, because they want an adult identity.
My take: children want identity because they want a role in the world, and gender is one of a handful of simplistic means (like race, or size / body type) of generating an identity ("Who are you, Johnny?" "I'm a tall, white boy."). My biggest problem with this, however, is the fact that humans have proven all but incapable of accepting the existence of non-qualitative differences. Once you point out the difference between two objects (apples and oranges be damned), one of the two becomes "better." Tushnet's own examples from the article quoted above distinctly point to this: "That's boy stuff, ewww!" tells us boy stuff is something to be despised; "Boys don't cry" because it's beneath them (and thus girls, who do cry, are similarly subordinate). Given that there are any number of actual, qualitative differences among people, what useful purpose does it serve to give precedence to the arbitrary, non-qualitative ones (especially in a post-industrial environment, which no longer requires the division of labor along sex boundaries)?
What it comes down to, then, is that I don't believe children do need a father. In discussing lesbian couples raising children, Tushnet argues that the family structure "makes it harder for both daughters and sons to form a sense of what it means to be a man and how men fit into the family." I rather think the persistence of gender roles themselves makes it harder to figure out "what it means to be a man," in that we continue to insist that being a man is innately different from being a woman on a level other than a rudimentary, biological one.
Tushnet goes on to compare the lesbian child-rearing couple to an unmarried woman raising children with the help of her mother, a comparison that soundly ignores a number of important differences: 1) There is of necessity a sizable age difference in the members of the mother-daughter pairing which is generally less likely to exist (or exist in as noteworthy a gap) in a lesbian couple. 2) While a lesbian couple has likely made the joint decision to rear a child (both members are doing this for the child), the mother-daughter "union" is forged out of a level of general familial obligation (the mother is doing this as much if not more for her child as for her grandchild). 3) Mother-daughter pairings are built upon a relationship with a history of (necessary) inequity, and are being shoehorned into a dynamic which requires peer cooperation. One relationship now has to serve two, not-remotely-complementary purposes. I think children in these situations do suffer, and they suffer from poorly-defined familial roles, but it isn't "from not having a male role model and a sense of men's place in the family." Rather, it's because their grandmother is trying to be three different people: grandparent, parent to her daughter, and parent to her granddaughter. Children who are raised by their adult older siblings (regardless of the gender of said siblings) suffer for the same reason: because their (adult) brother is trying to be their father. Plenty of role-confusion, but it's not uniquely gender-based by any means.