This post is going to be loooonnnngggg. Mainly because it contains some Butler quotes.
I don't think that Butler's position on pornography is that simple, OutsSelf. And even it it was, what does she propose that we actually do? I mean if she thinks that MacKinnon's analysis is wrong? What does Butler think we should be doing instead of challenging the First Amendment? Butler always seems keen to take down other women's (especially MacKinnon's) efforts to use the available tools to actually do something - so what is her great idea then?? Seems like Butler finds just about everything that women want to actually do about oppression to be a bad idea because doing anything about oppression inevitably makes that oppression worse. Gosh darn it. That rather puts a spanner in the works doesn't it? Thanks for the positive attitude Butler. Now what do you suggest that we actually do? Oh, yes, use that incredible tool against social injustice and violent systems of oppression; parody. 
I don't think that Butler has anything like the problem with pornography that you suggest. Bizarrely, for one who gives so much consideration to 'performativity', Butler doesn't think that pornography is performative. She thinks it is phantasmatlc.
Some quotes from Excitable Speech by Butler.
Page 18. Significantly, MacKinnon's argument against pornography has moved from a conceptual reliance on a perlocutionary model to an illocutionary one. In the work of Mari. Matsuda, hate speech is understood not only to act upon its listener (a perlocutionary scene), but to contribute to the social constitution of the one addressed (and, hence, to become part of a process of social interpellation). The listener is understood to occupy a social position or to have become synonymous with that position, and social positions themselves are understood to be situated in a static and hierarchical relation to one another. By virtue of the social position he or she occupies, then, the listener is injured as a consequence of that utterance. The utterance also enjoins the subject to reoccupy a subordinate social position. According to this view, such speech reinvokes and reinscribes a structural relation of domination, and constitutes the linguistic occasion for the reconstitution of that structural domination. Although sometimes this view on hate speech enumerates a set of consequences that such speech produces (a perlocutionary view of the matter), there are other formulations of this position where the force of the performative is secured through conventional means (an illocutionary model). In Mari Matsuda's formulation, for instance, speech does not merely reflect a relation of social domination; speech enacts domination, becoming the vehicle through which that social structure is reinstated. According to this illocutionary model, hate speech constitutes its addressee at the moment of its utterance; it does not describe an injury or produce one as a consequence; it is, in the very speaking of such speech, the performance of the injury itself, where the injury is understood as social subordination.
What hate speech does, then, is to constitute the subject in a subordinate position. But what gives hate speech the power to constitute the subject with such efficacy? Is hate speech as felicitous as it appears in this account, or are there faultlines that make its constituting power less felicitous than the above description would imply?
I wish to question for the moment the presumption that hate speech always works, not to minimize the pain that is suffered as a consequence of hate speech, but to leave open the possibility that its failure is the condition of a critical response. If the account of the injury of hate speech forecloses the possibility of a critical response to that injury, the account confirms the totalizing effects of such an injury. Such arguments are often useful in legal contexts, but are counter-productive for the thinking of nonstate-centered forms of agency and resistance.
Even if hate speech works to constitute a subject through discursive means, is that constitution necessarily final and effective? Is there a possibility of disrupting and subverting the effects produced by such speech, a faultline exposed that leads to the undoing of this process of discursive constitution?What kind of power is attributed to speech such that speech is figured as having the power to constitute the subject with such success?
Matsuda's argument presumes that a social structure is enunciated at the moment of the hateful utterance; hate speech reinvokes the position of dominance, and reconsolidates it at the moment of utterance. As the linguistic rearticulation of social domination, hate speech becomes, for Matsuda, the site for the mechanical and predictable reproduction of power. In some ways, the question of mechanical breakdown or "misfire" and of the unpredictability of speech is precisely what Austin repeatedly emphasizes when he focuses on the various ways in which a speech act can go wrong. More generally, however, there are reasons to question whether a static notion of "social structure" is reduplicated in hate speech, or whether such structures suffer destructuration through being reiterated, repeated, and rearticulated. Might the speech act of hate speech be understood as less efficacious, more prone to innovation and subversion, if we were to take into account the temporal life of the "structure" it is said to enunciate? If such a structure is dependent upon its enunciation for its continuation, then it is at the site of enunciation that the question of its continuity is to be posed. Can there be an enunciation that discontinues that structure, or one ,that subverts that structure through its repetition in speech? As an invocation, hate speech is an act that recalls prior acts, requiring a future repetition to endure. Is there a repetition that might disjoin the speech act from its supporting conventions such that its repetition confounds rather than consolidates its injurious efficacy?
Page 20 In MacKinnon's recent work, Only Words, pornography is construed as both speech and conduct, indeed, as "performative utterance;' and is understood not only to "act on" women in injurious ways (a perlocutionary claim), but to constitute, through representation, the class of women as an inferior class (an illocutionary claim). The burning cross is understood to be analogous to the pornographic utterance to the extent that both of them represent and enact an injury. But can the illocutionary claim be made about pornography as easily as it can about the burning cross? The theory of representation and, indeed, the theory of performativity at work differs in each of these cases. I will argue that, taken generically, the visual text of pornography cannot "threaten" or "demean" or "debase" in the same way that the burning cross can. To suggest that both examples instantiate the same kind of verbal conduct is not only a mistake in judgment, but the exploitation of the sign of racial violence for the purposes of enhancing, through a metonymical slippage, the putatively injurious power of pornography.
Page 22 Similarly, MacKinnon's appeal to the state to construe pornography as performative speech and, hence, as the injurious conduct of representation, does not settle the theoretical question of the relation between representation and conduct, but collapses the distinction in order to enhance the power of state intervention over graphic sexual representation. In many ways, this very extension of state power, however, comes to represent one of the greatest threats to the discursive operation of lesbian and gay politics. Central to such politics are a number of "speech acts" that can be, and have been, construed as offensive and, indeed, injurious conduct: graphic self-representation, as in Mapplethorpe's photography; explicit self-declaration, such as that which takes place in the practice of coming out; and explicit sexual education, as in AIDS education.