변호사 이재욱의 미국법 이야기 (Attorney Lee's commentary about U.S. law)

Freedom of Speech, Permissible Tailoring and Transcending Strict Scrutiny

EUGENE VOLOKH *
volokh@law.ucla.edu
Cite as Eugene Volokh, Freedom of Speech, Permissible Tailoring and Transcending Strict Scrutiny, 144 U. Pennsylvania L. Rev. 2417 (1997).

Introduction

       The Supreme Court has often held that content-based restrictions on fully protected speech are valid if they are "narrowly tailored to serve a compelling state interest." 1  I believe this is wrong.
       It is wrong descriptively:  There are restrictions the Court would strike down -- of which I'll give examples -- even though they are narrowly tailored to serve a compelling state interest.  It is wrong normatively:  In striking these restrictions down, the Court would, in my view, be correct.  And the official test is not just wrong but pernicious.  It risks leading courts and legislators to the wrong conclusions, it causes courts to apply the test disingenuously, and it distracts us from looking for a better approach. 2
       After briefly restating strict scrutiny doctrine (Part I), I'll give three examples of speech restrictions that in my view would pass muster if the strict scrutiny framework were taken seriously, but that nonetheless would and should be struck down (Part II).  I'll then point to some of the costs of the Court's reliance on an unsound doctrinal structure (Part III), and finally (Parts IV and V) suggest the rough foundations -- and, I concede, only the rough foundations -- of two alternative approaches.
       The first alternative is for the Court to acknowledge that there is a third prong to strict scrutiny, which I call "permissible tailoring."  Rather than just asking about the strength of the government's interest, or about whether the means are narrowly drawn to accomplish the interest, it asks whether the means are nonetheless impermissible:  Whether, no matter how narrow they are, and no matter how compelling an interest they serve, the means are still contrary to some basic prohibitions that the Free Speech Clause imposes.  This, I'll argue, is an inquiry quite distinct from what the Court requires under the "narrow tailoring" prong.
       The second alternative, which I prefer, is for the Court to shift away from means-ends scrutiny, and toward an approach that operates through categorical rules -- such as a per se ban on content-based speech restrictions imposed by the government as sovereign -- coupled with categorical exceptions, such as the exceptions for fighting words, obscenity and copyright.  I think this framework would better direct the Court's analysis, and would avoid the erroneous results that strict scrutiny seems to command.

I.  A Brief Restatement of Strict Scrutiny Doctrine

       Content-based speech restrictions, the Court says, are constitutional if they are "narrowly tailored to serve a compelling state interest"; 3 many have aptly called this an "ends and means" inquiry. 4  The Court makes a normative judgment about the ends:  Is the interest important enough to justify a speech restriction? 5  And the Court makes a primarily empirical judgment about the means:  If the means do not actually further the interest, are too broad, are too narrow, or are unnecessarily burdensome, then the government can and should serve the end through a better-drafted law. 6
       The Court has set forth four general principles related to compelling interests:
       1.  The government can have no compelling interest in privileging particular subclasses of core protected speech -- discussion about economic, social and political matters -- over other subclasses.  All such core protected speech "rest[s] on the highest rung of the hierarchy of First Amendment values." 7  The mere interest in furthering a subset of this speech (for instance, labor picketing) "without more, cannot justify [a content-based] exemption" for such speech. 8
       2.  Avoidance of offense and restriction of bad ideas are not compelling interests by themselves:  "`[T]he government may not prohibit the expression of an idea simply because society finds the idea itself offensive or disagreeable.´" 9  Note though that the word "simply" here, like the phrase "without more" in the quote at the end of the last paragraph, leaves a good deal uncertain.  What if society finds an idea offensive and the resulting offense leads to a particular bad result, such as employees of a particular religion, race or sex becoming so offended by workplace speech that they reasonably conclude that their workplaces have become hostile environments? 10  What if society wants to prevent disclosure of embarrassing facts about people, because people might find it "offensive or disagreeable" for their neighbors to know these facts about them? 11  The answers to both these questions, and to others like them, are far from settled.
       3.  A law's underinclusiveness -- its failure to reach all speech that implicates the interest -- may be evidence that an interest is not compelling, because it suggests that the government itself doesn't see the interest as compelling enough to justify a broader statute. 12
       4.  An interest might itself be impermissibly underinclusive, even if the law is quite narrowly fitted to the interest:  The government (at least under some circumstances) may not assert a compelling interest in fighting one particular ill, and then refuse to deal with other ills that seem almost indistinguishable.  There's only one case that squarely raises this point, so the boundaries of this principle are still quite vague. 13
       Outside these general areas, the Court has recognized a number of specific interests as compelling:  "maintaining a stable political system"; 14 ensuring that "criminals do not profit from their crimes" and that crime victims are compensated by the criminals; 15 protecting the right of "members of groups that have historically been subjected to discrimination . . . to live in peace where they wish"; 16 protecting voters from confusion, undue influence and intimidation; 17 preventing vote-buying; 18 "eliminating from the political process the corrosive effect of political `war chests' amassed with the aid of the legal advantages given to corporations"; 19  and protecting "the unique role [of] the press," which may justify otherwise impermissible speaker discrimination. 20  In the associational rights and religious freedom contexts, the Court has held that preventing race and sex discrimination, 21 preserving the integrity of the tax system, 22 and "procuring the manpower necessary for military purposes" 23 are also compelling.  Query whether these interests apply equally in the free speech context.
       On the other hand, the Court has held that the interests in "equalizing the relative ability of individuals and groups to influence the outcome of elections," 24 in "reducing the allegedly skyrocketing costs of political campaigns," 25 in "preserving party unity during a primary," 26 and in protecting speakers who "are incapable of deciding for themselves the most effective way to exercise their First Amendment rights" 27 are not compelling.  If the substantive due process and equal protection cases are any guide, neither is the interest in administrative efficiency. 28 
       Most cases striking down speech restrictions, however, rely primarily on the narrow tailoring prong, which, according to the Court, contains four components: 29
       1.  Advancement of the Interest:  For a law to be narrowly tailored, the government must prove to the Court's satisfaction that the law actually advances the interest. 30  The government need not, however, prove this scientifically; a sufficiently persuasive common-sense foundation is enough. 31
       2. No Overinclusiveness:  A law is not narrowly tailored if it restricts a significant amount of speech that doesn't implicate the government interest. 32  The theory here is that if the government can serve the interest while burdening less speech, it should.
       3.  Least Restrictive Alternative:  A law is not narrowly tailored if there are less speech-restrictive means available that would serve the interest essentially as well as would the speech restriction. 33  The justification for this requirement is similar to that for the overinclusiveness inquiry, though one can imagine a law that isn't overinclusive -- that restricts only the speech that implicates the interest -- but is still not the least restrictive alternative.  This might happen when the interest can be served equally well with a restriction on unprotected conduct rather than on speech, 34 or with a restriction that merely limits the speech in some ways rather than barring it altogether. 35  The government need not, however, choose an alternative that "fall[s] short of serving [the] compelling interests." 36
       These three components are closely related, and all of them could be subsumed within the "least restrictive alternative" inquiry.  If the law doesn't actually advance the interest, then not having the law at all would be a less restrictive but equally effective alternative.  Likewise, if the law is overinclusive, then a narrower law that exempted speech which doesn't implicate the interest would be less restrictive and equally effective.  When the Court says, as it sometimes does, that a law must be "`necessary to serve a compelling state interest,´" 37 it seems to be referring to these three components. 38
       4.  No Underinclusiveness:  Finally, a law is not narrowly tailored if it fails to restrict a significant amount of speech that harms the government interest to about the same degree as does the restricted speech. 39  Underinclusiveness might suggest, as mentioned above, that the interest isn't very important, or that the government's real interest wasn't the stated one but was rather just a desire to favor one form of speech over another, or to suppress offensive or otherwise disfavored speech. 40  Underinclusiveness may also show the presence of content discrimination beyond that justified by the compelling interest.  Because content discrimination is disfavored, and allowed only when justified by a compelling interest, the presence of this extra, unjustified distinction makes the law unconstitutional. 41
       Some cases have condemned content distinctions because speech on both sides of the line implicated the interest equally.  For instance, the Court has held that a bar on nonlabor residential picketing was not narrowly tailored to the interest in preserving residential privacy because labor picketing and nonlabor picketing were equally intrusive. 42  Such cases turn on underinclusiveness principles -- they strike down laws for being strikingly underinclusive with respect to the interest they purport to serve.
       On their faces, all four of these prohibitions call for primarily empirical judgments -- judgments about the closeness of the fit between the law and the interest -- rather than normative ones.  Once the Court concludes that the interest is compelling, the strict scrutiny framework requires that the free speech interest yield to a narrowly drawn law.  The narrow-tailoring prong, then, involves essentially factual questions about whether the law is indeed narrowly drawn:  Does the law further the interest; is the law limited to speech that implicates the interest; does the law cover all such speech; are there less restrictive alternatives that will serve the interest equally well?

II.  Applying the Doctrine to Some Cases

A.  The Wartime Election

       The problem with the strict scrutiny framework is that if taken seriously it would reach results startlingly at odds with some fundamental free speech principles.
       Consider the following example:  The United States is fighting a slow, limited, Vietnamesque war against, say, Serbia.  (For the sake of simplicity, assume that we are the key partner in the coalition, and that our departure would mean a Serbian victory.)  Things seem to be going slowly in our favor, but the Serbs appear able to hold out until at least the next election.  The current U.S. administration is offering the Serbs peace on relatively harsh terms.
       The U.S. presidential campaign is about to start, and Congress is concerned:  What if some candidates announce that they'd give the Serbs a more favorable offer?  If this happens, the Serbs might refuse to come to terms until after the election, hoping a dovish candidate will win.  And, of course, every month that peace is delayed could mean thousands of lost lives, both American and foreign.  So Congress passes a law that prohibits all prospective presidential candidates from saying that they'd offer peace terms lighter than those the current administration is proposing. 43
       Such a law should be unconstitutional, even if one believes that more speech restrictions should be allowed in wartime than in peacetime.  This is not a ban on revealing the sailing dates of troopships, or even on encouraging people to resist the draft.  This bars a candidate from bringing before the voters alternatives that they, as the sovereign power in a democracy, are entitled to consider.  They've considered such alternatives in wartime elections throughout American history, especially in 1864, 1968 and 1972.  Surely their right to consider these alternatives is a key aspect of democracy, precisely because so much rides on their decision.
       But, the Court tells us, content-based speech restrictions are constitutional if they pass strict scrutiny -- if they "are narrowly tailored to serve a compelling state interest." 44  (Of course, the fact that the speech here relates to elections doesn't give it any protection greater than does strict scrutiny; the three cases that have upheld speech restrictions under the strict scrutiny framework -- Buckley v. Valeo, 45 Austin v. Michigan Chamber of Commerce 46 and Burson v. Freeman 47 -- all involved election-related speech.) What answer, then, does the official strict scrutiny framework yield?

1.  Compelling Interest

       The government interest in winning the war more quickly -- and in preserving lives -- is as compelling as they come. 48  The interest is not in "prohibit[ing] the expression of an idea simply because society finds the idea itself offensive or disagreeable." 49  The interest is in winning the war, not simply in suppressing ideas.
       Some might suggest that the interest is impermissible because it's an interest in winning the war through the suppression of speech.  This argument, though, confounds the two prongs of the test.  The government doesn't want suppression of speech for its own sake.  Rather, the government's interest is in the enemy quickly accepting the peace terms, just as the interest the Court accepted as compelling in Buckley was in avoiding corruption.  In both cases, the suppression of speech is only the means by which the interest is served.  If there is any sense in separating the inquiry into the end from the inquiry into the means -- and the strict scrutiny framework assumes there is sense in this -- then the interest must be the goal the government is serving, without reference to the means it's using.
       But even if one reads the interest here as including a "through the suppression of speech" clause, there can't be anything per se unacceptable about this interest.  The premise of strict scrutiny is that sometimes the government may suppress speech and suppress ideas, so long as the means of suppression is narrowly tailored to serve a compelling end.
       In Burson, for instance, the interest in preventing fraud and intimidation was jeopardized by people communicating their ideas, and was served by suppressing this communication. 50  In Austin the interest in preventing corruption and the "corrosive and distorting effects [on the political process] of immense aggregations of wealth" was jeopardized by corporations communicating their ideas, and was served by suppressing this communication. 51  Despite this resulting suppression, a plurality in Burson and a majority in Austin concluded that the law passed strict scrutiny; even the Burson dissent didn't doubt that the interest was compelling. 52

2.  Narrow Tailoring -- Advancement of the Interest

       The factual predictions underlying the law seem eminently plausible.  If the Serbs think they might get a better peace if a certain candidate is elected, they have a powerful incentive to keep fighting.
       Granted, candidates' statements might not make a difference in some situations, for instance if the Serbs can't hold out until the election regardless of who is likely to win, or if they will fight to the end in any case.  But in many other circumstances, an enemy might indeed behave differently depending on the various presidential candidates' views.  "Dissension within a country is a high source of comfort and assistance to its enemies; the least intimation of it they seize upon with jubilation.  There cannot be the slightest question of the mischievous effects of such agitation upon the success of the national project . . . ." 53  Surely this is especially true when one of the dissenters is a candidate for president.
       Of course, the candidates will have views, whether enunciated or not.  If the enemy knows that, for example, Governor Jones has dovish opinions, even if he's not expressing them in his campaign -- or even if the law forces him to express contrary views in the campaign -- the enemy might still decide to wait to see if Jones is elected.
       But even if Jones's general views are partly known -- which will not always be so, especially if he is currently only a governor and not a federal official -- there's a big difference between what he said before the campaign and what he says on the campaign trail.  Jones is probably much more likely to feel bound by his announced platform than by any views he might have expressed before he began his campaign.  Moreover, if he is elected, he'll have a much easier time implementing an announced policy, for which he can credibly claim he has a popular mandate, than an unannounced one.  Campaign statements should thus matter a great deal to the enemy's resolve.
       The law would not by itself go far in advancing the government interest; but the Court has never demanded that the law make a particularly huge dent in the problem it's trying to solve.  Certainly the contribution restrictions upheld in Buckley didn't go far by themselves in preventing corruption -- the Federal Election Campaign Act never purported to address all aspects of corruption. 54  Likewise, the law upheld in Burson was one of the less potent tools that the state had for "protecting voters from confusion and undue influence" or protecting the election process from fraud. 55  The Court, though, was untroubled by this; it demanded only that the law in some measure advance the interest, not that it advance it remarkably far. 56
       One can undoubtedly construct scenarios in which the campaign speech restriction wouldn't actually serve the asserted interest.  It's possible that the Serbs would be emboldened by the law's passage because they saw it as evidence that the United States thought itself in trouble, or would hope that the law would cause so much libertarian resistance that the U.S. war effort would be undermined -- possible, but not particularly likely.  If the Court gives the salutary effects of speech every possible benefit of the doubt, and views the harmful effects of speech with maximal skepticism, any speech restriction can be condemned as unnecessary or even counterproductive.  I'm not sure, however, that this is an honest way of evaluating the likely effectiveness of a speech restriction; and it certainly is not the approach the Court took in Burson, Austin or Buckley. 57

3.  Narrow Tailoring -- Overinclusiveness

       For the same reasons, the law is not overinclusive.  All the forbidden speech can jeopardize the government interest; all statements by a presidential candidate suggesting that he'd offer mild terms can indeed encourage the enemy to keep fighting.
       True, we can't identify precisely the harm that flows from any one statement; it could be that, viewed in retrospect, some isolated statements did not by themselves have any appreciable effect.  But that doesn't make the law overbroad.  Not every campaign contribution or coordinated expenditure leads to corruption. 58  Not every instance of electioneering near polling places intimidates voters; all one can say is that some such electioneering "may . . . drive the voter[s] away," and that this represents a "potential deficienc[y] in the electoral process." 59  Not every independent expenditure by a corporation on behalf of a political candidate has a corrosive effect on the political process -- all they have is a "potential for distortion." 60  Strict scrutiny does not demand (and cannot demand) that every restricted statement have a provable, identifiable harmful effect.  If it did, no restrictions would ever pass strict scrutiny.

4.  Narrow Tailoring -- Least Restrictive Alternative

       Nor are there any less restrictive alternatives that will still further the interest equally well and thus avoid "unnecessarily interfering with First Amendment freedoms." 61  Any alternative seems quite likely to "fall short of serving [the] compelling interests." 62
       Counterspeech is pointless -- leaders of the current Congress might say "No, no, we'll continue to offer harsh terms to the Serbs no matter what," but how much can they do if the dove indeed gets elected?  Congressional leaders might say, "No, no, don't elect this traitor," but so long as he seems viable in the polls -- and viability does not mean that a current majority of voters support him, only that he has enough support that his election remains a possibility -- the Serbs will keep fighting and Americans will keep dying.  Finally, the Administration might redouble its military efforts and try to force the Serbs to surrender before the election; but that might not work, and might in any event lead to even more casualties.

5.  Narrow Tailoring -- Underinclusiveness

       It's also hard to see how the law is underinclusive.  It bars a class of statements that is especially likely to make the enemy hold out; few other statements pose a comparable danger.  Perhaps one might argue that similar statements by congressional candidates are equally dangerous -- if so, then assume a law that does go that far.     
       So the law, it seems to me, is narrowly tailored to a compelling interest.  It advances the interest, it is not overinclusive, it is not underinclusive, and there are no less restrictive alternatives.  But it still seems clearly unconstitutional. 63

B.  Praising Rioters

       Consider another example:  The government bars speech that praises rioting, on the grounds that the speech makes it more likely that people will riot in the future. 64  It seems clear that such a law would be unconstitutional. 65  But can one reach that conclusion by applying strict scrutiny candidly?

1.  Compelling Interest

       The government interest in preventing death, injury and destruction of property is surely compelling.  Like the interest in the first example, it is not an interest in "prohibit[ing] the ex-pression of an idea simply because society finds the idea itself offensive or disagreeable." 66  The interest is in preventing violence, not simply suppressing communication (though, as in the first example and as in Buckley, Austin and Burson, the law accomplishes the interest by means of suppressing communication).

2.  Narrow Tailoring

       The theory of the law is that praise or even moral defense of rioting -- for example, people and especially community leaders calling the riots "rebellion," or suggesting that rioting actually does good by shaking up the oppressive power structure, or arguing that the oppressed are morally justified in burning, looting and assaulting -- will make future riots more likely and more damaging.  And this seems eminently plausible.
       When people know the police aren't there, the main things that stop them from rioting (other than the possibility of individual, usually violent, self-defense by their victims) 67 are their own consciences and the judgments of those about whose opinions they care.  Both these factors are deeply affected by what people hear from respected voices in their community.
       Speech, as some have pointed out, constructs social reality. 68  It can turn conduct that society otherwise marks "evil" into conduct marked "good" or "tolerable" or at least "contested."  If we believe that advocacy of good behavior makes it more likely that people will behave well, then it seems probable that advocacy of bad behavior, or even making excuses for bad behavior, makes it more likely that people will behave badly:

Every denunciation of existing law tends in some measure to increase the probability that there will be violation of it.  Condonation of a breach enhances the probability.  Expressions of approval add to the probability.  Propagation of the criminal state of mind . . . increases it.  Advocacy of lawbreaking heightens it still further. 69

       Thus, the law advances the government interest, 70 and it does this in a way that's not overinclusive; if the factual theory outlined above is correct, all the banned statements have the potential to undermine the interest. 71  Nor are there any less restrictive alternatives that will still avoid "unnecessarily interfering with First Amendment freedoms" 72 without "fall[ing] short of serving [the] compelling interests." 73  Prohibiting rioting certainly won't do the job by itself:  Laws aren't always followed, especially when order has already broken down.
       The government, or others, could try to counterspeak, but how effective would that be?  Speech that advocates bad behavior -- be it rioting, racism, bombing draft offices, bombing abortion clinics or killing police officers -- is dangerous because it makes certain behavior thinkable.  The very fact that someauthority figures condone a form of conduct may embolden those who want to act that way.  Rather than being a horrible thing that no good person would do, rioting or bombing becomes an acceptable alternative, one of many approaches to problem-solving that various people advocate.  Perhaps the counterspeech might undo some of the harm, but it seems quite unlikely that it will undo all or even most of it. 74
       One could, of course, argue that the likely effectiveness of proriot advocacy is overstated, as Holmes and Brandeis argued in some (though not all) of their 1910s and 1920s free speech opinions. 75  One could also argue that suppression of the speech is unlikely to do much good in any event, and that counterspeech is indeed the most practically useful response.  But these are only guesses, and they strike me as guesses that are hard to support.
       If simply articulating such a guess were enough to condemn a speech restriction as unnecessary and therefore unconstitutional, it's hard to imagine how any restriction would be permissible.  In fact, the Court has upheld speech restrictions under strict scrutiny in the face of such guesses.  The Court was not, for instance, willing to speculate that disclosure requirements alone would deter corruption as much as would a contribution ban. 76  Nor was the Court willing to speculate that laws barring interference with voting would prevent such interference as well as would the ban on electioneering near polling places. 77
       Sometimes, the Court does make commonsense judgments that a particular less restrictive means will indeed be equally effective. 78  But common sense suggests that bad speech can indeed lead to bad results, that speech restrictions might to some degree prevent such results, and that counterspeech standing alone will not always do an equally good job.
       Finally, the law is not underinclusive -- it bars all statements that countenance and thus indirectly encourage rioting.  The law might not discourage other crimes, but advocacy of rioting is particularly troublesome because the threat of punishment is particularly ineffective when order has broken down.  The construction of attitudes that tolerate rioting is thus especially dangerous.  Perhaps one might say that this isn't enough of a justification for singling out praise of rioting, and that the law should also ban praise of rape and murder and robbery and such.  If so -- and this would make the underinclusiveness requirement toothy indeed, probably toothier than the cases warrant 79 -- the hypothetical can be modified accordingly.

3.  Typicality

       The praise-of-rioters example is far from unusual.  I use it because I think it's particularly apt factually -- it seems more obvious that legitimization of rioting would lead to bad results today than would legitimization of, say, Communist revolution -- but the Court has had to deal with laws like the one I describe for decades.  In fact, the arguments for suppression discussed above were largely what the Court accepted in Abrams v. United States 80 with regard to speech that hinders the war effort, and in Gitlow v. New York 81 with regard to advocacy of violent revolution.
       In more recent times, many who advocate restrictions on racist and sexist speech have argued that these restrictions pass muster under the formal parameters of strict scrutiny. 82  Preventing discrimination is, they say, a compelling interest (and the Court has generally, outside the free speech context, agreed).83  The restrictions on bigoted speech serve this interest, and are neither overinclusive nor underinclusive in serving it.  And the less restrictive alternatives -- prohibiting discrimination or racial violence as such, and counterspeaking -- might not be very effective. 84

C.  Sexually Explicit Speech and Minors

       Finally, consider a third example:  The government tries to ban all distribution of sexually oriented materials that are supposedly harmful to minors but not obscene for adults.  The law is clearly unconstitutional -- the Court struck down such a law in 1957 in Butler v. Michigan, saying the law "burn[ed] the house to roast the pig." 85  The Justices held that the restriction was "not reasonably restricted to the evil with which it is said to deal," pointing out that other laws already barred distribution of such materials to minors. 86
       But if we reconsider this under the strict scrutiny framework that supposedly governs today, why isn't the law permissible?  The interest in "shielding minors from the influence of [indecent] literature that is not obscene by adult standards," the Court has since held, is compelling. 87  I'm not sure that's right, but that's what the Court has said.
       In Sable Communications v. FCC, the Court did strike down a ban on dial-a-porn, on the grounds that there were less restrictive alternatives, such as a requirement that the service ask for the caller's credit card number before continuing with the call. 88  But these alternatives mattered because "[t]here [was] no evidence . . . that enterprising youngsters could and would evade the rules." 89  "For all we know from this record," the Court asserted, "[these alternatives] would be extremely effective, [and i]f this is the case, it seems to us that [the law] is not a narrowly tailored effort to serve the compelling interest of preventing minors from being exposed to indecent telephone messages." 90  The pregnant negative -- and this is consistent with the other least restrictive means cases -- is that if the alternatives were likely to be ineffective, the law would be narrowly tailored.
       Unlike phone sex services, sexually explicit books and magazines don't have credit card verification mechanisms.  The only way to shield minors even halfway effectively from such publications is to ban them outright; if you let them be distributed to adults, they'll inevitably become accessible to children (even if distribution to children is outlawed).  Certainly the alternative suggested by Butler v. Michigan would "fall short of serving [the] compelling interest[]"; 91 it would be a less restrictive alternative, but also a much less effective one.
       Nor can the Butler law be condemned as overinclusive in the sense in which overinclusiveness is used in the cases.  Each sexually explicit book sold to an adult might fall into a child's hands.  Of course, many such books won't actually end up accessible to children, but then again many campaign contributions won't actually lead to corruption, and much electioneering near polling places won't actually intimidate voters.  So long as each instance of the prohibited speech might implicate the interest -- so long as we cannot tell whether it will or it won't -- it's hard to condemn the prohibition as overinclusive.

D.  Possible Defenses of Strict Scrutiny

1.  Strict Scrutiny as Balancing

       Some people with whom I've discussed the above examples have suggested that there might be something to strict scrutiny that I'm missing.  The narrow tailoring inquiry, they say, is more than just an inquiry into advancement of the interest, overinclusiveness, underinclusiveness and less restrictive means.  It also contains an element of balancing.  Under it, the Court should ask whether the degree to which the law serves the interest, compelling though the interest may be, is nonetheless outweighed by the degree to which the law abridges valuable speech.
       Alternatively, they say, the compelling interest inquiry is more than just an inquiry into whether the government interest is strong enough:  It is also an inquiry into whether the benefit to the government interest is outweighed by the harm to the equally compelling interest in free speech itself.  Under either approach, the Court might say, for instance, that the benefit to the war effort (and to the lives of our soldiers) which would be gained by the speech restriction is outweighed by the cost to deliberative democracy that the restriction imposes.
       I sympathize with the normative theory behind this view, but I do not think this approach is "strict scrutiny" as traditionally understood.  To begin with, this isn't what the Court says when it talks of strict scrutiny.  The phrases "compelling state interest" and "narrow tailoring" do not on their face include either of the "balancing" inquiries described above.  The interest inquiry seems to focus only on the importance of the government's ends:  Is the interest compelling?  If the answer is yes, the interest inquiry appears to stop.  Likewise, the tailoring inquiry seems to focus only on the means-ends fit:  Is the law the best practicable way of serving the interest?
       The cases don't go into the balancing approaches identified above either.  Austin v. Michigan Chamber of Commerce 92 and the plurality in Burson v. Freeman, 93 for instance, engage in the compelling-interest/narrow-tailoring inquiry I outlined in Part I:  They look at how compelling the interest is; they ask whether the law serves the interest and whether the law is overinclusive, underinclusive, or not the least restrictive means; and they don't do any "balancing" beyond this, even though election-related speech is generally seen as being at the core of Free Speech Clause protections. 94  A lower court judge or a government official who reads these cases for guidance would, I think, read them as demanding only the conventional compelling-interest/narrow-tailoring analysis. 95
       Strict scrutiny can be described as a "balancing test" only in the sense that it -- like most constitutional tests -- tries to strike a balance between the claims of the opposite sides.  Government interests that are not compelling are outweighed by the free speech interest; if the interest, however, is compelling and the law is narrowly tailored to it, the balance comes out in the other direction.  At least this is what is suggested by the language of the test and the Court's description of how it applies the test.
       If one departs from this conventional model, and injects an extra balancing element -- an inquiry into whether the need to serve the compelling government interest is outweighed by the need to protect free speech -- then this element will become the true focus of the inquiry.  This is where a court would have broad discretion to strike down a law.  Silently packing the discretion into either the "compelling state interest" prong or the "narrow tailoring" prong will only confuse.  Better to come out and say that narrow tailoring to a compelling interest isn't the whole story, and that there is an additional test the law must meet.

2.  Strict Scrutiny as Presumption or as Non-Exclusive Guide

       Some other readers have suggested that strict scrutiny might have been meant only as a presumption:  Applying strict scrutiny yields the first cut at an answer, but the presumption might be rebutted, either by the challenger or by the state.  Or perhaps the presumption is conclusive one way but rebuttable the other -- being narrowly tailored to a compelling interest is necessary for a law to be valid but not sufficient.  Finally, perhaps strict scrutiny is not really a rule but is rather a guide for legislators and lower courts in exercising their judgment:  It reminds them that they should consider various things in their decisionmaking -- whether the law advances a compelling interest, whether it's overinclusive, whether it's the least restrictive means, and whether it's underinclusive -- but does not purport to set forth an exhaustive list.
       As with balancing, though, this is not what the Court says.  "To determine whether [a restriction is constitutional]," the Court tells us, "we must ascertain whether it burdens the exercise of political speech and, if it does, whether it is narrowly tailored to serve a compelling state interest." 96  Once that latter determination is made, the Court says the law is valid.  Many cases say the test is whether the law is narrowly tailored to a compelling state interest -- I've seen none that say this is just the start of the inquiry.
       Perhaps legislators, given their independent obligation to construe the Constitution, might conclude that a law is unconstitutional even though it passes strict scrutiny.  But nothing in the Court's opinions suggests that lower courts have the same option.  A strict-scrutiny-as-presumption approach might be better than the current one, though a court that accepts it would then have to decide what it takes to rebut the presumption.  It is not, however, the approach prescribed by the Court's decisions.
       My conclusion, then, remains:  Strict scrutiny, as the Court has described it, does not reach the correct result in the examples given above.

III.  The Harms Caused by Strict Scrutiny

       If the above criticisms are correct, the doctrine can cause three kinds of problems:

A.  Wrong Results

       First, decisionmakers might follow the doctrine and reach wrong results because of it.  A court (either a lower court or the Supreme Court itself) applying strict scrutiny might uphold a law even though the principles underlying the free speech doctrine -- those principles that lead us to condemn the laws described in Part II -- suggest that the law should be struck down. 97  Likewise, legislators and administrators might enact speech restrictions that seem to pass muster under strict scrutiny, but that conflict with deeper free speech principles.

B.  Stretching the Doctrine

       Second, courts might try to avoid the wrong results by ignoring or stretching the doctrine, striking down a law even though the law would pass strict scrutiny faithfully applied.  This seems to happen fairly often.
       Consider, for instance, the Court's discussion of least restrictive means in Riley v. National Federation of the Blind. 98  Riley struck down a requirement that charitable solicitors reveal what fraction of the funds they raise actually ends up in the charity's hands. 99  The State argued that the law was needed to better inform donors, but the Court held that there were less restrictive alternatives, such as "the State . . . itself publish[ing] the detailed financial disclosure forms it requires professional fundraisers to file." 100  Such a procedure, the Court said, "would communicate the desired information to the public without burdening a speaker with unwanted speech during the course of a solicitation," and would thus be "in keeping with the First Amendment directive that government not dictate the content of speech absent compelling necessity, and then, only by means precisely tailored." 101
       But it's not clear how publication by the state is even close to an effective substitute for requiring the fundraisers to reveal this information when they make their pitch.  Who's going to run to a state office to get these forms before contributing her money?
       The Court's inquiry into whether the expenditure limitation in Buckley v. Valeo advanced the interest in avoiding corruption seems similarly problematic.  The Court said that the ban on independent expenditures "fail[ed] to serve any substantial governmental interest in stemming the reality or appearance of corruption." 102  The lack of coordination with the candidate, it argued, made the independent spending less valuable to the candidate and thus "alleviate[d] the danger that expenditures will be given as a quid pro quo for improper commitments from the candidate." 103 
       This justification has some merit, but it hardly disposes of the government's argument.  Many officeholders would find it hard to say no to people who helped elect them, whether the help was independent or not.  Independent expenditures may have less of a corrupting tendency than direct contributions, but they have this tendency nonetheless.  It's hard to see how the expenditure ban "fail[ed] to serve" the interest in avoiding corruption.
       It's also odd to hold, as R.A.V. v. City of St. Paul did, that a law is not narrowly tailored to an interest because there is an alternative that would restrict more speech, including speech that doesn't implicate the interest.  R.A.V. held that a ban on racially hostile fighting words was not "narrowly tailored" to the concededly compelling interest in allowing racial minorities to "live in peace where they wish"; 104 it reached this conclusion because a general ban on all fighting words "would have precisely the same beneficial effect." 105
       Nonracially motivated fighting words, however, would not implicate the government interest, an interest limited on its face to protecting racial minorities.  The city ordinance was in fact neither over- nor underinclusive with respect to this interest, and the proposed alternative would have been more restrictive than the ordinance, not less.
       These judicial stretches are not cost-free.  When a test is sufficiently out of step with our intuitions that even the institution which articulated it often tries to wiggle out of it, the test loses what constraining and predictive qualities it might have.  It becomes much less of a guide for official behavior (by the Justices, by lower court judges, or by legislators or administrators), and it becomes a much less effective predictor of judicial decisions.
       Judicial stretching can also lead the losing side to conclude that its cause is being treated unfairly.  Restrictions on bigoted speech are a good example.  Like the restrictions on advocacy of rioting, they seem to me to be impermissible not because they aren't narrowly tailored to a compelling interest, but in spite of their being narrowly tailored.  If this is so, then it's much better for courts to say up front that strict scrutiny is the wrong test, rather than to claim strict scrutiny is the right test but then avoid candidly applying it.  The latter approach will only lead to all-too-plausible charges of judicial bias. 106

C.  The Bad Doctrine Drives Out the Good

       Worst of all, the focus on strict scrutiny distracts the Court from the important business of searching for a better framework.  Maybe Buckley was right to strike down the ban on independent expenditures, because a law that bars people from taking out an ad in the New York Times to speak their mind about a candidate is inconsistent with basic free speech principles.  Maybe the Constitution requires the government to tolerate whatever risk of corruption such independent spending might cause.  Maybe R.A.V. was right, because the Free Speech Clause prohibits viewpoint-discriminatory rules even if they pass strict scrutiny.
       But none of these conclusions really fits within the strict scrutiny framework.  Trying to shoehorn them into categories of "compelling interest" and "narrow tailoring" simply obscures the more important inquiry into what the rules ought to be.

IV.  "The Theory of Our Constitution"

A.  The Pre-Strict Scrutiny Free Speech Cases

       As I mentioned before, if one looks at the majority opinions in Gitlow v. New York 107 and other cases that upheld bans on advocacy of illegal conduct (such as Dennis v. United States 108), one sees something not far from the strict scrutiny framework.  The opinions stressed that the government interest was compelling; the speech presented "a sufficient danger of substantive evil," "danger to the public peace and to the security of the State," and the threat of "breaches of the peace and ultimate revolution." 109  The danger was "real and substantial"; while some revolutionary advocacy might fall on deaf ears, other speech "may kindle a fire that, smoldering for a time, may burst into a sweeping and destructive conflagration." 110  (The world had just witnessed such conflagrations in the years before, in the Russian Revolution as well as smaller revolts in other European countries.)  And any less restrictive means would likely have been less effective -- waiting until the threat was "clear and present" would mean bearing the burden of "actual disturbances of the public peace or imminent and immediate danger of [the State's] destruction." 111
       The Court has generally repudiated the Gitlow approach, in Brandenburg v. Ohio 112 and in the cases leading to it and following it. 113  But Brandenburg's protection of the advocacy of bad conduct can't fit within the strict scrutiny framework.  The problem is not that the ultimate interests asserted by the government aren't compelling; they are.  Nor is the problem that the laws suppressing speech are somehow, as an empirical matter, not the most effective way of serving the interests.  Subversive speech is protected regardless of whether the restriction is narrowly tailored to a compelling interest. 114
       "[T]he best test of truth," Justice Holmes said, "is the power of the thought to get itself accepted in the competition of the market." 115  But he did not, and could not, assert this as empirically provable fact.  Rather, he followed his statement by saying:  "That at any rate is the theory of our Constitution." 116
       Likewise, in Gitlow v. New York, Holmes did not suggest that counterspeech would effectively suppress revolutionary advocacy, and that it therefore was in fact a less restrictive but equally effective alternative.  Rather, he asserted that "[i]f in the long run the beliefs expressed in proletarian dictatorship are destined to be accepted by the dominant forces of the community, the only meaning of free speech is that they should be given their chance and have their way." 117  And in Whitney v. California, 118 Justice Brandeis did not deny that bad speech could lead to bad results (he acknowledged it), nor did he argue that good counsels would always be adequate remedies for evil ones.  He argued only that they were "the fitting . . . ones," the ones the Framers judged appropriate, a judgment they enacted into the First Amendment. 119
       Holmes and Brandeis were not condemning the restrictions simply because the government's goals were not important enough, or arguing that the speech-restrictive means were unnecessary.  Rather, they were arguing that the means were impermissible (except where a clear and present danger was present).  Even if the speech was likely, through its persuasive powers, to undermine the most compelling of interests, the "meaning of free speech," 120 "the theory of our Constitution," 121 the judgment of "[t]hose who won our independence" about the "fundamental principle[s] of the American government," 122 required that the nation run the risk of the compelling interest being under-mined. 123
       Something similar is at work for sexually explicit speech.  In Butler v. Michigan, 124 the law banning the distribution to adults of explicit speech that could be harmful to minors was in fact the least restrictive means of serving the government's end, an end eventually declared compelling by Sable Communications v. FCC. 125  But the Butler Court struck down the restriction nonetheless, because the law "arbitrarily curtail[ed] one of those liberties of the individual, now enshrined in the Due Process Clause . . . that history has attested as the indispensable conditions for the maintenance and progress of a free society." 126  In the Court's view, the law constituted "burn[ing] the house to roast the pig." 127
       The trouble with burning the house to roast the pig is not just that there are better ways -- "less restrictive alternatives" -- to cook your dinner.  Rather, it's that even if you had no other stove, you would be better off leaving the pig unroasted.  It is this conclusion about the judgment of the Constitution and of history, and not strict scrutiny as applied in Sable, that is the true defense of sexually explicit speech.
       In none of these cases did the Justices claim that they found any unified theory that would by itself resolve all Free Speech Clause cases.  Justice Holmes wasn't arguing, for instance, that the "marketplace of ideas" model disposed of all free speech issues.  The Butler Court didn't assert that the sole test in free speech matters was whether the speech furthered the progress of a free society.
       Rather, the claims about theory and about the attestation of history were aimed at producing individual strands of doctrine -- the "clear and present danger" test, or the rule that the government may not "reduce the adult population . . . to reading only what is fit for children." 128  And the resulting doctrines, while narrow, were categorical; they called for no further inquiry into narrow tailoring or compelling interests.
       Strict scrutiny, then, is not the test that is doing the heavy lifting in free speech cases.  Through the force of precedents that emerged outside the strict scrutiny context, the Court has applied broader principles, principles that in the Court's view express the theory of free speech.  The Court may talk strict scrutiny, but what ultimately protects speech is something else.

B.  Analogies from Strict Scrutiny Outside the Free Speech Context

       The same seems to be true in other strict scrutiny contexts.  Consider Palmore v. Sidoti, which dealt with racial classifications. 129  Two white parents divorced, and the mother got custody of the couple's child; the mother married a black man.  A state court then awarded custody to the father, because living in an interracial household would have been against the child's best interest -- "`despite the strides that have been made in bettering relations between the races in this country, it is inevitable that [the daughter] will . . . suffer from the social stigmatization that is sure to come.´" 130  The Supreme Court reversed.
       The Court could have argued that preventing social stigmatization of the child was not a "compelling" enough interest to justify race-conscious action, or that the fears of social hostility were exaggerated.  But the Court didn't say this.  "The State," it said, "has a duty of the highest order to protect the interests of minor children." 131  "There is a risk that a child living with a stepparent of a different race may be subject to a variety of pressures and stresses," something the Court was willing to describe as an "injury." 132
       The reason the Court gave for invalidating the race-conscious custody decision is hard to reconcile with the formal framework of narrow tailoring and compelling interests.  "[T]he reality of private biases and the possible injury they might inflict," the Court said, is not a "permissible consideration[] for the removal of an infant child from the custody of its natural mother. . . . Private biases may be outside the reach of the law, but the law cannot, directly or indirectly, give them effect. . . . The effects of racial prejudice, however real," could not justify this racial classification. 133
       The Court wasn't saying that the means didn't advance the interest, were overinclusive, were underinclusive, or weren't the least restrictive.  It was saying that no means, no matter how narrowly tailored, would be permitted so long as their efficacy turned on "the reality of private biases" and gave effect to those biases.  The only way I can see to explain this decision is as a judgment that the theory of our Constitution requires a certain result, even in the face of a possibly compelling interest and government action that may, as an empirical matter, be narrowly tailored to it. 134
       Consider also a religious freedom example.  It's widely assumed that the Free Exercise Clause and the Religious Freedom Restoration Act bar the application of Title VII to race or sex discrimination by churches in their choice of clergy. 135  But why?
       The Court has already held, in the free exercise and expressive association contexts, that the interests in stopping race and sex discrimination in education and public accommodations are compelling; 136 lower courts agree that the same is true in employment. 137  What could be more narrowly tailored to these interests than a prohibition on such discrimination? 138
       Of course, the overall social harm of allowing discrimination in one small sector of the economy is less than the harm of allowing it everywhere, but that's true regardless of which small sector one identifies.  It's true of clubs such as the Jaycees, but the Court has held that antidiscrimination law passes strict scrutiny there; 139 it is true of employment by religious institutions generally (rather than just employment of clergy), but courts have upheld antidiscrimination law there too. 140  Moreover, the compelling interest is not just the interest in solving a broad social problem -- it is also the interest in protecting the economic and dignitary concerns of individuals. 141  These concerns are implicated even by discrimination in a small sector of the economy, especially when the sector covers all the opportunities that the applicant most wants. 142
       To protect the church's right to discriminate in its choice of clergy, courts must abandon the notion that infringements of religious freedom are allowed so long as they pass strict scrutiny.  In some situations, a court must hold -- as lower courts generally do in clergy discrimination cases -- that "the `inroad on religious liberty´ is too substantial to be permissible" 143 even though the law is narrowly tailored to a compelling interest.  What does the work here is not strict scrutiny, but an underlying theory of the autonomy of religious institutions.
       Even outside the strict scrutiny context, talk of means-ends analysis often hides the real theoretical basis for the opinion.  Consider Linmark Associates v. Willingboro, 144 which involved a ban on the posting of "For Sale" and "Sold" signs.  Today, this would be analyzed under the Central Hudson quasi-intermediate-scrutiny commercial speech framework.  At the time, the Court talked more generally about the "important governmental objective" -- in Linmark, the interest in preventing panic selling and "promoting stable, racially integrated housing" -- and whether the law was needed to serve the objective. 145
       But after examining the strength of the government interest and whether the law was necessary to serve the interest, the Court proceeded to state that "[t]he constitutional defect in this ordinance . . . is far more basic": 146  The government was trying to serve its ends by "deny[ing people] information that is neither false nor misleading." 147
       The problem with the law, according to the Court, was not just that there was no important interest at stake, or that there were other means that, as an empirically demonstrable matter, would serve the interest equally well.  The choice between, on the one hand, the "paternalistic approach" of serving government ends by restricting the flow of accurate information and, on the other, "assum[ing] that this information is not in itself harmful [and] that people will perceive their own best interests if only they are well enough informed," was not to be made on pragmatic grounds. 148 Rather, the choice was one "`that the First Amendment makes for us.´" 149  The perceived theory of the Constitution, not a consideration of the weight of the end or the practical efficacy of the means, was doing the work. 150
       Finally, this perspective might help explain why strict scrutiny has seemed much less strict in the pre-Smith Free Exercise Clause context than in Free Speech Clause and race discrimination cases. 151  One possible explanation for this result is that courts faced with religious freedom claims have not applied strict scrutiny honestly.  Another is that neutral laws restricting conduct might indeed be narrowly tailored to compelling interests more often than are race classifications or content-based speech restrictions.
       But a third piece of the puzzle might be that our underlying theory of religious freedom -- the principles that might lead us to protect religious conduct even when it jeopardizes compelling government interests -- is less rights-protective than our underlying theory of free speech or of racial equality.  Perhaps, setting aside institutional autonomy of churches and a few other areas, even people who are committed to strict scrutiny in religious freedom cases really do believe that religious claims must yield to laws that are narrowly tailored to compelling interests.
       If this is true, it might explain why the interest in stopping racial discrimination would override a religious university's free exercise rights to make race-based rules for its students, but why the same interest would not override the right to advocate racism.  It is not necessarily that strict scrutiny is being underapplied in religious freedom cases, but rather that something beyond strict scrutiny is protecting speech in many free speech cases.

C.  Reconciling Strict Scrutiny with the Doctrine

       What, then, would a court do if faced with one of the laws described in Part II?  For instance, say a legislature enacts the ban on praise of rioting.  The law should be struck down under Brandenburg, which holds that advocacy of lawlessness is protected unless it "is directed to inciting or producing imminent lawless action and is likely to incite or produce such action," 152 but, if I'm correct, under the strict scrutiny test the law would have to be upheld.  Given that Brandenburg is still good law, what could a court do to avoid applying strict scrutiny?
       The answer must be that strict scrutiny is in certain cases preempted, so to speak, by the "theory of our Constitution" as expressed in cases such as Brandenburg -- that, as I suggest above, some speech restrictions are unconstitutional under Brandenburg even though they are narrowly tailored to a compelling interest.  But how exactly does one draw the line between Brandenburg's zone of operation and the zone where the strict scrutiny test applies?
       Brandenburg might be read to suggest that the line is drawn based on the content of speech -- that advocacy which doesn't rise to the level of incitement is "immunized from governmental control" 153 -- but that can't be right.  Like all other forms of speech, protected advocacy of violence can still be restricted by some of the laws that pass strict scrutiny. 154  If I advocate violence near a polling place on election day, surely the law upheld in Burson v. Freeman prohibiting such speech would be as applicable to me as to anyone else.  If a corporation puts out an ad advocating violence in the course of endorsing a candidate, surely the law upheld in Austin v. Michigan Chamber of Commerce would cover it.
       The protection given by Brandenburg must turn on the nature of the law at issue, not just on the nature of the speech.  If the law achieves its goals by means of cutting off the persuasive effect of speech, the law is per se unconstitutional, unless the speech passes the intent-imminence-likelihood test. 155  As I suggest in Part II.B, this isn't a matter of the strength of the government interest or of narrow tailoring -- a law that tries to keep people from being persuaded by bad speech may effectively serve the compelling interest in preventing injury and death through the least restrictive means, and be neither overinclusive nor underinclusive.  There are certain means that, under Brandenburg's vision of the theory of the Free Speech Clause, are unconstitutional regardless of how well they serve a compelling interest.
       So the Brandenburg doctrine can be seen as one of two things.  One can view it as a categorical rule:  Speech restrictions that accomplish their ends by trying to stop people from persuading others are per se unconstitutional, regardless of whether they are narrowly tailored to a compelling interest (so long as the intent-imminence-likelihood threshold isn't crossed).  Or one can see it as a third prong to the strict scrutiny test:  Speech restrictions are constitutional only if they further a compelling interest, are narrowly tailored to that interest, and don't further the interest by stopping people from using speech to persuade.

V.  Alternatives to Strict Scrutiny

A.  Amending Strict Scrutiny:  A New "Permissible Tailoring" Prong

       Strict scrutiny, as I argue in Parts II and III, is flawed:  If taken seriously, it is hard to reconcile with our intuitions and with some important precedents, such as Brandenburg v. Ohio and Butler v. Michigan.  As Part IV suggests, these precedents, among others, look to something beyond compelling interests and narrow tailoring; they articulate per se rules that, in the Justices' views, embody the theory of free speech.
       The existence of these per se rules suggests two things:  First, whatever doctrine the Court uses must somehow include these rules; and, second, the doctrine must take into account the possibility that other, similar rules might be created in the future.  After all, if Brandenburg identifies one type of law that is unconstitutional even if narrowly tailored to a compelling interest, perhaps there might also be other types.  Rather than focusing the courts solely on compelling interests and narrow tailoring, the doctrine should also focus on searching for these new categorical rules.
       One way of repairing strict scrutiny doctrine, 156 then, is to add a third prong -- the "permissible tailoring" inquiry -- that captures this possibility.  Under this modified doctrine, a law would be constitutional if it is (1) narrowly tailored to, (2) a compelling state interest, and (3) serves the interest in a way consistent with the theory of our Constitution.
       Thus, the ban on praise of rioters might be "impermissibly tailored" because under Brandenburg v. Ohio a law may not try to cut off the persuasive effect of political advocacy, an effect that may be the very reason why free speech is protected.  The ban on distribution to adults of material that's harmful to minors would be impermissibly tailored because, as Butler v. Michigan holds, it's per se improper to entirely deprive adults of this sort of speech. 157
       If a ban on dovish speech by wartime candidates were enacted, a court could conclude that it's impermissible for a law to serve even the most compelling interest in a way that denies the sovereign people the ability to choose national policy.  This would be a new per se principle that the Court could enunciate, and that would become part of the "permissible tailoring" inquiry. 158  Similarly, outside the free speech area, a ban on discrimination by churches in clergy selection might be impermissibly tailored because it interferes with a core theoretical principle of church autonomy.  The race-conscious custody decision might likewise be impermissible because it operates in a way that gives effect to private prejudices.

B.  Abandoning Strict Scrutiny

       It seems to me, though, that the permissible tailoring inquiry would often end up wagging the strict scrutiny dog.  If part of the answer to the question "Is this law constitutionally permissible?" is "Does the theory of our Constitution permit it?" or "Is the law permissibly tailored to the interest?," it doesn't seem useful to view the framework -- here the new three-prong strict scrutiny framework -- as much of an answer at all.  It would at most be a way of reframing the question.
       It may therefore be better to shift away from the strict scrutiny approach altogether, as Justice Kennedy, concurring in Simon & Schuster v. Members of the New York State Crime Victims Board, suggested several years ago. 159  Instead, the Court should see its task as being the development of a system of categorical rules and categorical exceptions that reflect its view of the meaning of the Free Speech Clause.  Such a system would not be that different from what the Court has done (admittedly, with varying success) for many other constitutional provisions, such as the Ex Post Facto Clause, the Double Jeopardy Clause, the Criminal Jury Trial Clause and the Assistance of Counsel Clause. 160  And it would not be that different from what the Court has done in other areas of free speech jurisprudence, for instance in libel cases, incitement cases and obscenity cases. 161
       In fact, many of the most important free speech principles were themselves developed outside the strict scrutiny framework.  Content-based restrictions are viewed with more skepticism than content-neutral restrictions because (rightly or wrongly) the Court believes that "above all else, the First Amendment means that government [presumptively] has no power to restrict expression because of its message, its ideas, its subject matter, or its content." 162  The intermediate level of protection given to commercial speech is premised on the Court's vision (again, rightly or wrongly) of the "choice . . . that the First Amendment makes for us," 163 and of the "commonsense differences" between commercial speech and noncommercial speech. 164
       Likewise, the government has a freer hand when acting as employer than when acting as sovereign because of the Court's normative judgments about when it's reasonable to expect employees to give up some part of their rights in exchange for a government salary. 165  Similar theoretical judgments underlie the Court's conclusions that "fighting words," obscenity, child pornography and threats are unprotected; that speech restrictions in nonpublic fora must be viewpoint-neutral; and that libel recoveries involving public figures and matters of public concern are allowed only if actual malice is proven. 166  To be sure, in formulating many of these principles the Court did consider the magnitude of the government interest in restricting the speech and the degree to which the restrictions served the interest; but it didn't limit itself to these inquiries.
       By abandoning strict scrutiny of content-based restrictions -- or perhaps even intermediate scrutiny of commercial speech restrictions -- the Court can shift its focus to creating, as best it can, rules that capture its theory about the proper role of such restrictions.  It could, for instance, say that content-based regulations of high-value speech imposed by the government acting as sovereign are simply per se unconstitutional (subject to the recognized exceptions).  Content-based restrictions on nonmisleading commercial speech concerning lawful activities might also be declared per se unconstitutional (subject to a perhaps larger set of exceptions).
       This sort of doctrinal change would be substantial but hardly unthinkable.  As I mentioned above, Justice Kennedy has already called for the abandonment of strict scrutiny of content-based speech restrictions.  Justice Thomas has argued that Buckley v. Valeo should be partly overruled because "[t]he First Amendment . . . constrain[s] Congress's ability to accomplish certain goals," even when those goals (such as elimination of corruption) are compelling and when any alternatives would be less effective. 167  Justice O'Connor, while willing to apply strict scrutiny in the free speech context, has shown some unease with it in the associational freedom context. 168  And Justice Stevens, while no fan of categorical rules such as the one I propose, has long been no fan of strict scrutiny, either. 169
       And because so few cases -- only Buckley, Austin and the Burson plurality -- have actually used strict scrutiny to uphold content-based speech restrictions, shifting to a per se rule of invalidity (with room for categorical exceptions) would not substantially upset the existing regime.  If the Court believes that these three cases were rightly decided, it would just have to identify why it believes this, and formulate categorical exceptions that embody its theory.
       The Court might, for instance, conclude that corporate speakers are different from ordinary speakers, at least where candidate elections are involved. 170  Likewise, it might conclude that the per se bar should not apply where the speech has a reasonable likelihood of intimidating voters, or where "the justification for the speech restriction is to protect another constitutional right." 171  I'm not persuaded that it would be right to carve out these exceptions, but then again I'm unpersuaded by the arguments the Court gave for reaching these results under strict scrutiny.
       In each case, the Court would ask the familiar questions:  Does some interpretive theory -- whether tied to the constitutional text, to broader constitutional or moral values, to the caselaw, or to something else -- support this distinction?  Is the proposed rule likely to lead to good results in most cases?  Is the rule likely to be properly administered by courts and other government officials? 172
       Sometimes the Court's answers to these questions might lead it to demand that a restriction serve a compelling government interest, or that the law not be overinclusive or underinclusive, or that it be the least restrictive means.  Sometimes, it might demand less:  Consider the Court's willingness to uphold copyright law and, to some extent, right of publicity law 173 -- both of these are content-based restrictions on high-value speech imposed by the government acting as sovereign, but it's not clear that they are narrowly tailored to a compelling state interest, and the Court certainly didn't analyze them this way. 174  Sometimes though, the categorical framework might lead the Court to strike down even those restrictions that would otherwise pass strict scrutiny.
       I make no claims here about what the particular rules and exceptions should be.  I'm arguing that the Court should try to decide what is demanded by the "theory of the Constitution," but I'm not arguing in favor of any particular theory.  The Justices already have their own views about the theory of the Free Speech Clause; all I suggest is that the law will develop better if they implement these views through the categorical approach rather than through a strict scrutiny framework.
       Such a proposal, of course, leaves the inquiry somewhat indeterminate.  I hope the categorical rules the Court sets forth will be fairly precise, but in every case there will be a possible argument for a new rule or a new exception, which will leave room for a good deal of subjective and unpredictable judgment.  The Court, though, already has the power to create new categorical rules and to carve out new exceptions, whether or not the strict scrutiny framework is retained; the decision in New York v. Ferber 175 that child pornography lacks First Amendment value is one example.  And if I am correct, strict scrutiny as the Court applies it -- and as strict scrutiny must be applied in order to avoid results that many would condemn as clearly incorrect -- must contain an equally indeterminate and subjective "permissible tailoring" component.
       It is better to make the nature of the inquiry explicit, and to make the Justices' constitutional choices more visible, than to leave them hidden under a false veneer of a more determinate- and objective-sounding test.  I am not certain about what "the theory of the Constitution" should be, or whether one should figure it out by consulting original meaning, moral principle, pragmatic judgment, or what have you.  My suggestion is only that strict scrutiny is not a helpful approach.

Conclusion

       Contrary to the Court's announced test, some content-based speech restrictions are unconstitutional even though they are narrowly tailored to a compelling state interest.  The same goes for some race classifications and some restrictions on religious freedom.  The Court says that strict scrutiny is the test, but what it does -- correctly, in my view -- is something quite different.
       If I'm right, the Court has several options:
       1.  Take strict scrutiny seriously and reverse the decisions that have struck down those restrictions.
       2.  Continue saying that it's applying the traditional strict scrutiny framework, but nevertheless continue striking down these restrictions.
       3.  Add a third prong to the strict scrutiny framework that makes clear that the challenged law must not only be narrowly tailored to the compelling state interest, but must also be permissibly tailored -- must comport with the theoretical foundations of the right involved.
       4.  Reject strict scrutiny, and operate through categorical rules and categorical exceptions.
       I believe the fourth option is the best one, but at the very least either the third or fourth approach is necessary.  It's hardly healthy for free speech law -- or individual rights law more broadly -- to have the announced rules be as inconsistent as they are with the decided cases.

---

 
    *    Acting Professor, UCLA Law School (volokh@law.ucla.edu). Many thanks to Daniel Bussel, Evan Caminker, David Cruz, Michael Kent Curtis, Julian Eule, Robert Goldstein, Tom Grey, Jerry Kang, Kenneth Karst, Doug Laycock, Sanford Levinson, Daniel Lowenstein, Sidney Rosenzweig, Gary Schwartz, David Sklansky, David Sobelsohn, Jonathan Varat, James Weinstein, Stephen Yeazell and the members of the UCLA American Politics Discussion Group for their advice.
    1.    See, e.g., Austin v. Michigan Chamber of Commerce, 494 U.S. 652, 655 (1990); Boos v. Barry, 485 U.S. 312, 334 (1988) (plurality); see also Burson v. Freeman, 504 U.S. 191, 198 (1992) (plurality); Board of Airport Comm'rs v. Jews for Jesus, Inc., 482 U.S. 569, 573 (1987); Cornelius v. NAACP Legal Defense and Educ. Fund, Inc., 473 U.S. 788, 800 (1985); United States v. Grace, 461 U.S. 171, 177 (1983); Perry Educ. Ass'n v. Perry Local Educators' Ass'n, 460 U.S. 37, 45 (1983).
       Strict scrutiny is the rule, of course, only if the restrictions are imposed by the government acting as sovereign, and not as, for instance, employer, landlord, or primary or secondary school educator, where it has more discretion.  Even when the government is acting as sovereign, it may have more latitude in particular areas such as broadcasting regulations, copyright law and the right of publicity.  But strict scrutiny is still the general rule, and is the highwater mark of speech protection -- in all the non-strict-scrutiny contexts the rule is more tolerant of speech restrictions than is strict scrutiny.
    2.    My criticism here is not that the strict scrutiny test can't produce predictable results, though there's a good deal of merit to this point.  See, e.g., Eu v. San Francisco County Democratic Cent. Comm., 489 U.S. 214, 234 (1989) (Stevens, J., concurring); Illinois Bd. of Elections v. Socialist Workers Party, 440 U.S. 173, 188 (1979) (Blackmun, J., concurring).  Rather, it is that the strict scrutiny test can in many cases produce the wrong results.
    3.    See supra note 1.
    4.    See, e.g., Sable Communications v. FCC, 492 U.S. 115, 126 (1989); Lillian BeVier, Campaign Finance Reform:  Special Arguments, Intractable Dilemmas, 94 Colum. L. Rev. 1258, 1278 (1994); see also Adarand Contructors v. Pena, 115 S. Ct. 2097, 2117 (1995) (race classification context); Randall Kennedy, The State, Criminal Law, and Racial Discrimination:  A Comment, 107 Harv. L. Rev. 1255, 1265 n.45 (1994) (same).
    5.    Some have suggested that "public purpose" is a better term than "government interest," see Stephen E. Gottlieb, Compelling Governmental Interests and Constitutional Discourse, 55 Alb. L. Rev. 549, 552 (1992), and much can be said for this view. But "government interest" is so firmly a part of the strict scrutiny test that I prefer to stick with it in this context.
    6.    I say "primarily empirical" because there are some subsidiary normative judgments involved in deciding, for instance, how much over- or underinclusiveness is too much, or whether a particular alternative is less burdensome.  Cf. Sable Communications, 492 U.S. at 131-32 (Scalia, J., concurring).
    7.    Carey v. Brown, 447 U.S. 455, 467 (1980); see also Consolidated Edison Co. v. Public Serv. Comm'n, 447 U.S. 530, 537-38 (1980). 
    8.    Carey, 447 U.S. at 467.
    9.    Simon & Schuster, Inc. v. Members of the N.Y. State Crime Victims Bd., 502 U.S. 105, 118 (1991) (quoting Texas v. Johnson, 491 U.S. 397, 414 (1989)).
    10.    Compare Suzanne Sangree, Title VII Prohibitions Against Hostile Environment Sexual Harassment and the First Amendment:  No Collision in Sight, 47 Rutgers L. Rev. 461 (1995) (arguing that speech restrictions would be constitutional in such a case) with Kingsley R. Browne, Title VII as Censorship:  Hostile-Environment Harassment and the First Amendment, 52 Ohio St. L.J. 481 (1991) (arguing that such speech restrictions would be unconstitutional) and Eugene Volokh, Comment, Freedom of Speech and Workplace Harassment, 39 UCLA L. Rev. 1791 (1992) (taking a middle course); see also Kingsley R. Browne, Workplace Censorship:  A Response to Professor Sangree, 47 Rutgers L. Rev. 579 (1995); Eugene Volokh, How Harassment Law Restricts Free Speech, 47 Rutgers L. Rev. 563 (1995); Suzanne Sangree, A Reply to Professors Volokh and Browne, 47 Rutgers L. Rev. 595 (1995).
    11.    See Florida Star v. B.J.F., 491 U.S. 524, 541 (1989) (declining to decide whether the "disclosure of private facts" tort is constitutional).
    12.    See id. at 542 (Scalia, J., concurring in part and in the judgment); Carey, 447 U.S. at 465.
    13.    The case is Simon & Schuster, Inc. v. Members of the N.Y. State Crime Victims Bd., 502 U.S. 105 (1991).  The government argued that it had an interest in "ensuring that criminals do not profit from storytelling about their crimes before their victims have a meaningful opportunity to be compensated for their injuries."  Id. at 119.  The Court agreed that the government had a compelling interest in "ensuring that criminals do not profit from their crimes" generally.  Nonetheless, the Court rejected the government's proffered interest -- even though it was a subset of the interest the Court concluded was compelling -- because the government's interest swept too narrowly:  The government could offer no "justification for a distinction" between profiting from speech about one's crime and profiting from the crime in other ways.  Id. at 119-20.
    14.    Eu v. San Francisco County Democratic Cent. Comm., 489 U.S. 214, 226 (1989).
    15.    Simon & Schuster, 502 U.S. at 118-19.
    16.    R.A.V. v. City of St. Paul, 505 U.S. 377, 395 (1992).
    17.    See Burson v. Freeman, 504 U.S. 191, 199 (1992) (plurality).
    18.    See Buckley v. Valeo, 424 U.S. 1, 27 (1976).
    19.    Austin v. Michigan Chamber of Commerce, 494 U.S. 652, 666 (1990).
    20.    Id. at 667.
    21.    See Roberts v. United States Jaycees, 468 U.S. 609, 623 (1984) (sex discrimination); Bob Jones Univ. v. United States, 461 U.S. 574, 604 (1983) (race discrimination).
    22.    See United States v. Lee, 455 U.S. 252, 258-59 (1982).
    23.    Gillette v. United States, 401 U.S. 437, 462 (1971).
    24.    Buckley v. Valeo, 424 U.S. 1, 48-49 (1976).
    25.    Id. at 57.
    26.    Eu v. San Francisco County Democratic Cent. Comm., 489 U.S. 214, 228 (1989).
    27.    Riley v. National Fed'n for the Blind, 487 U.S. 781, 790 (1988).
    28.    See City of Richmond v. J.A. Croson Co., 488 U.S. 469, 508 (1989) (race classification context); Carey v. Population Servs. Int'l, 431 U.S. 678, 691 (1977) (contraceptive rights context); Police Dep't of Chicago v. Mosley, 408 U.S. 92, 102 n.9 (1972) (holding that, in free speech cases, "small administrative convenience" is not a compelling interest).
    29.    I speak here of narrow tailoring in the context of content-based restrictions.  Content-neutral restrictions are also supposed to be "narrowly tailored" to significant government interests, but narrow tailoring there is quite different than where content-based restrictions are involved.  See Ward v. Rock Against Racism, 491 U.S. 781, 799 n.6 (1989).  Content-neutral restrictions need not, for instance, be the least restrictive alternatives, and it seems they might be valid even if they are underinclusive.  See id. (as to least restrictive alternative); Members of the City Council v. Taxpayers for Vincent, 466 U.S. 789, 810-11 & n.28 (1984) (as to underinclusiveness).
    30.    See, e.g., Eu, 489 U.S. at 226, 228-29 (1989); Meyer v. Grant, 486 U.S. 414, 426 (1988); FEC v. Massachusetts Citizens for Life, Inc., 479 U.S. 238, 262 (1986); Globe Newspaper Co. v. Superior Court, 457 U.S. 596, 609-10 (1982); First Nat'l Bank v. Bellotti, 435 U.S. 765, 789-90 (1978); Buckley v. Valeo, 424 U.S. 1, 45-47, 53 (1976).
    31.    See, e.g., Burson v. Freeman, 504 U.S. 191, 211 (1992) (plurality) (pointing to a "long history, a substantial consensus, and simple common sense" as support for a finding of narrow tailoring); Austin v. Michigan Chamber of Commerce, 494 U.S. 652, 660 (1990); Buckley, 424 U.S. at 26-27 (looking to longstanding political practice).
    32.    See, e.g., Simon & Schuster, Inc. v. Members of the N.Y. State Crime Victims Bd., 502 U.S. 105, 120-21 (1991); FEC v. National Conservative Political Action Comm., 470 U.S. 480, 500-01 (1985); First Nat'l Bank, 435 U.S. at 794.
    33.    See, e.g., Rutan v. Republican Party, 497 U.S. 62, 74 (1990); Sable Communications v. FCC, 492 U.S. 115, 126 (1989); Florida Star v. B.J.F., 491 U.S. 524, 538 (1989); Boos v. Barry, 485 U.S. 312, 329 (1988); Minneapolis Star & Tribune Co. v. Minnesota Comm'r of Revenue, 460 U.S. 575, 586 (1983).
    34.    See, e.g., Meyer v. Grant, 486 U.S. 414, 425-28 (1988); Boos, 485 U.S. at 326-29.
    35.    See, e.g., Sable Communications, 492 U.S. at 129-30.
    36.    Burson v. Freeman, 504 U.S. 191, 206 (1992) (plurality); see also Buckley v