The idea that marriage exists to bind a woman to a man for the purposes of property and inheritance is to say the least ethnocentric. My own ancestors (Cherokee) were matrilineal for property and inheritance rights, but they found marriage a beneficial institution as well. In that culture the man was bound to the woman for the purposes of mutual defense and protection of the children.
The current literature suggests strongly a two parent heterosexual marriage provides the best outcomes for children over the generic two parent household. The literature also suggests that a two parent gay household is superior to any single parent household.
Bottom line — before we start making political changes to an institution that has existed over several thousand years we must develop a thorough knowledge of how it works and why it exists. I can say this without fear of contradiction: existing across all cultures, languages, and civilizations it is safe to conclude that religion has no significant effect on the existence of marriage.
That said, the religious argument over a state definition of marriage is irrelevant. Whichever way you want it, you must make a sociological or biological argument for or against the institution in general. I strongly suspect that one must make the sociological argument within a given culture. Religion can strongly reinforce the cultural definition of marriage, but does not actually define the institution.
Valid scientific studies of gay couples with children exist in the literature. As far as they go most of these studies contribute to the necessary knowledge base. However, in the public debate a serious problem arises when we try to compare homosexual couples when one partner is previously divorced to heterosexual couples where divorce does not occur.
Children of couples who stay together until they reach majority are generally more successful as adults. When divorce occurs while the children are still in the home the literature indicates that whatever family arrangements come afterwards, the two worst situations are single parent and orphan status. In this case the proper conclusion has nothing to do with sexual orientation of the parents because the major event for the child is the divorce.
As I understand the literature (and I know it better than most), the best child rearing environment is a permanently mated heterosexual couple with strong community and ancillary support networks. Second best is a second marriage involving another mated heterosexual couple. After these two family configurations come gay couples permanently mated (rare among gay males, more common among gay females), single parent homes regardless of sexual orientation, foster home arrangements, and outright orphanage. All of these family arrangements improve in direct proportion to the positive influence of the community and ancillary support networks available to the family.
Finally, note that while a good support network is a major factor for the child, it is not generally sufficient to overcome the effects of a divorce. As an additional data point most researchers find that when all else is equal and the missing parent is deceased instead of divorced, positive outcomes for the child are significantly better (provided of course that death of the missing parent did not involve domestic abuse or violence).
Footnote to the above discussion: It is rare to find two studies in this area that define “successful”, “good”, “positive”, “better”, or “best” the same way. It is even more rare find two studies using the same criteria to measure. Therefore attempts to fix conclusions in a mathematical way are most often futile.