There are two ways for the pragmatist to respond. The first is to bite the bullet. If we accept all the premises of the story, then ex hypothesi if we "rescued" the child, other people and other children would, on the whole, suffer even more. Since we know (again by the premises of the story) they would suffer, we would be just as responsible for their suffering as we are for the scapegoat's. The ones who walk away (i.e. those who reject Omelas' solution to the pragmatic dilemma) are responsible for more suffering, not less, and know they're responsible. If knowledge entails responsibility, then the only escape from Omelas is into ignorance. If knowledge doesn't entail responsibility, then no one is responsible for the scapegoat's suffering.
The second response is for the pragmatist to question the premises of the story. Pragmatists have beliefs that are pseudo-moralistic: They are judgments about the intrinsic value of things, but they are shortcuts to deal with uncertainty and ignorance; if we knew the outcome better, we would change the belief. We might be radically uncertain as to why torturing children leads to an overall bad outcome, but we are convinced that it does. All of our beliefs, including our pseudo-moral beliefs, are shaped by social and biological evolution, i.e. by reality exerting selection pressure. If Omelas were physically possible, it would seem to require some physics very different from our own. Indeed, LeGuin gives us no mechanism at all by which the suffering of the scapegoat child actually causes the benefits to Omelas. LeGuin only assumes but does not prove that our modern pseudo-moralistic beliefs fail to conform to reality.
Pragmatism cannot be a purely "rationalistic" project, i.e. that we must predict the outcomes of all possible actions using rigorous, scientific logic, and choose the action that will produce the best outcome. Instead, pragmatism entails adding a conscious, intentional selection pressure to our moral beliefs. We can at times determine that a pseudo-moralistic belief leads to a bad outcome or prevents a good outcome. When we can make that determination, we change the pseudo-moralistic belief. The burden of proof is still on the pragmatist ("if it works, don't fix it" conservatism is obviously correct).
But pragmatists can meet that burden. We know, for example, that homosexuality causes no real harm: to oppress and marginalize homosexuals causes them considerable suffering to no real benefit. Our pseudo-moralistic beliefs about the evils of homosexuality must fall to our knowledge. We know that childhood sex education (real education, not abstinence-only ignorance fetishism) prevents pregnancy and sexually transmitted diseases; our pseudo-moralistic beliefs about protecting children's sexual