Opinion

JUSTICE O'CONNOR, with whom JUSTICE SCALIA and JUSTICE GINSBURG join, and with whom JUSTICE THOMAS joins as to Parts I and III, concurring in part and dissenting in part.

There are only so many channels that any cable system can carry. If there are fewer channels than programmers who want to use the system, some programmers will have to be dropped. In the must-carry provisions of the Cable Television Consumer Protection and Competition Act of 1992, Pub. L. 102-385, 106 Stat. 1460, Congress made a choice: By reserving a little over one-third of the channels on a cable system for broadcasters, it ensured that in most cases it will be a cable programmer who is dropped and a broadcaster who is retained. The question presented in this case is whether this choice comports with the commands of the First Amendment.

I A

The 1992 Cable Act implicates the First Amendment rights of two classes of speakers. First, it tells cable operators which programmers they must carry, and keeps cable operators from carrying others that they might prefer. Though cable operators do not actually originate most of the programming they show, the Court correctly holds that they are, for First Amendment purposes, speakers. Ante, at 636. Selecting which speech to retransmit is, as we know from the example of publishing houses, movie theaters, bookstores, and Reader's Digest, no less communication than is creating the speech in the first place.

Second, the Act deprives a certain class of video programmers-those who operate cable channels rather than broadcast stations-of access to over one-third of an entire medium. Cable programmers may compete only for those channels that are not set aside by the must-carry provisions. A cable programmer that might otherwise have been carried may well be denied access in favor of a broadcaster that is less appealing to the viewers but is favored by the mustcarry rules. It is as if the Government ordered all movie theaters to reserve at least one-third of their screening for films made by American production companies, or required all bookstores to devote one-third of their shelf space to nonprofit publishers. As the Court explains in Parts I, II-A, and II-B of its opinion, which I join, cable programmers and operators stand in the same position under the First Amendment as do the more traditional media.

Under the First Amendment, it is normally not within the government's power to decide who may speak and who may not, at least on private property or in traditional public fora. The government does have the power to impose contentneutral time, place, and manner restrictions, but this is in large part precisely because such restrictions apply to all speakers. Laws that treat all speakers equally are relatively poor tools for controlling public debate, and their very generality creates a substantial political check that prevents them from being unduly burdensome. Laws that single out particular speakers are substantially more dangerous, even when they do not draw explicit content distinctions. See, e. g., Minneapolis Star & Tribune Co. v. Minnesota Comm'r of Revenue, 460 U. S. 575, 584, 591-592 (1983); see also Leathers v. Medlock, 499 U. S. 439, 447 (1991).

I agree with the Court that some speaker-based restrictions-those genuinely justified without reference to content-need not be subject to strict scrutiny. But looking at the statute at issue, I cannot avoid the conclusion that its preference for broadcasters over cable programmers is justified with reference to content. The findings, enacted by Congress as § 2 of the Act, and which I must assume state the justifications for the law, make this clear. "There is a substantial governmental and First Amendment interest in promoting a diversity of views provided through multiple technology media." § 2(a)(6). "[P]ublic television provides educational and informational programming to the Nation's citizens, thereby advancing the Government's compelling interest in educating its citizens." § 2(a)(8)(A). "A primary objective and benefit of our Nation's system of regulation of television broadcasting is the local origination of programming. There is a substantial governmental interest in ensuring its continuation." § 2(a)(10). "Broadcast television stations continue to be an important source of local news and public affairs programming and other local broadcast services critical to an informed electorate." § 2(a)(11).

Similar justifications are reflected in the operative provisions of the Act. In determining whether a broadcast station should be eligible for must-carry in a particular market, the Federal Communications Commission (FCC) must "afford particular attention to the value of localism by taking into account such factors as... whether any other [eligible station] provides news coverage of issues of concern to such community or provides carriage or coverage of sporting and other events of interest to the community." §4, 47 U. S. C. § 534(h)(1)(C)(ii) (1988 ed., Supp. IV). In determining whether a low-power station is eligible for must-carry, the FCC must ask whether the station "would address local news and informational needs which are not being adequately served by full power television broadcast stations." § 4, 47 U. S. C. § 534(h)(2)(B) (1988 ed., Supp. IV). Moreover, the Act distinguishes between commercial television stations and noncommercial educational television stations, giving special benefits to the latter. Compare § 4 with § 5. These provisions may all be technically severable from the statute, but they are still strong evidence of the statute's justifications.

Preferences for diversity of viewpoints, for localism, for educational programming, and for news and public affairs all make reference to content. They may not reflect hostility to particular points of view, or a desire to suppress certain subjects because they are controversial or offensive. They may be quite benignly motivated. But benign motivation, we have consistently held, is not enough to avoid the need for strict scrutiny of content-based justifications. Simon & Schuster, Inc. v. Members of N. Y. State Crime Victims Bd., 502 U. S. 105, 117 (1991); Arkansas Writers' Project, Inc. v. Ragland, 481 U. S. 221, 228 (1987). The First Amendment does more than just bar government from intentionally suppressing speech of which it disapproves. It also generally prohibits the government from excepting certain kinds of speech from regulation because it thinks the speech is especially valuable. See, e. g., id., at 231-232; Regan v. Time, Inc., 468 U. S. 641, 648-649 (1984); Metromedia, Inc. v. San Diego, 453 U. S. 490, 514-515 (1981) (plurality opinion); Carey v. Brown, 447 U. S. 455, 466-468 (1980); Police Dept. of Chicago v. Mosley, 408 U. S. 92, 96 (1972); Cox v. Louisiana, 379 U. S. 536, 581 (1965) (Black, J., concurring); see also R. A. v: v. St. Paul, 505 U. S. 377, 386 (1992) ("The government may not regulate [speech] based on hostility-or favoritismtowards the underlying message expressed").

This is why the Court is mistaken in concluding that the interest in diversity-in "access to a multiplicity" of "diverse and antagonistic sources," ante, at 663 (internal quotation marks omitted)-is content neutral. Indeed, the interest is not "related to the suppression of free expression," ante, at 662 (emphasis added and internal quotation marks omitted), but that is not enough for content neutrality. The interest in giving a tax break to religious, sports, or professional magazines, see Arkansas Writers' Project, supra, is not related to the suppression of speech; the interest in giving labor picketers an exemption from a general picketing ban, see Carey and Mosley, supra, is not related to the suppression of speech. But they are both related to the content of speech-to its communicative impact. The interest in ensuring access to a multiplicity of diverse and antagonistic sources of information, no matter how praiseworthy, is directly tied to the content of what the speakers will likely say.

B

The Court dismisses the findings quoted above by speculating that they do not reveal a preference for certain kinds of content; rather, the Court suggests, the findings show "nothing more than the recognition that the services provided by broadcast television have some intrinsic value and, thus, are worth preserving against the threats posed by cable." Ante, at 648. I cannot agree. It is rare enough that Congress states, in the body of the statute itself, the findings underlying its decision. When it does, it is fair to assume that those findings reflect the basis for the legislative decision, especially when the thrust of the findings is further reflected in the rest of the statute. See Church of Lukumi Babalu Aye, Inc. v. Hialeah, 508 U. S. 520, 534-535 (1993) (relying on recitals in a city council resolution as evidence of the justifications for an ordinance).

Moreover, it does not seem likely that Congress would make extensive findings merely to show that broadcast television is valuable. The controversial judgment at the heart of the statute is not that broadcast television has some value-obviously it does-but that broadcasters should be preferred over cable programmers. The best explanation for the findings, it seems to me, is that they represent Congress' reasons for adopting this preference; and, according to the findings, these reasons rest in part on the content of broadcasters' speech. To say in the face of the findings that the must-carry rules "impose burdens and confer benefits without reference to the content of speech," ante, at 643, cannot be correct, especially in light of the care with which we must normally approach speaker-based restrictions. See Minneapolis Star & Tribune Co. v. Minnesota Comm'r of Revenue, 460 U. S. 575 (1983).

It may well be that Congress also had other, contentneutral, purposes in mind when enacting the statute. But we have never held that the presence of a permissible justification lessens the impropriety of relying in part on an impermissible justification. In fact, we have often struck down statutes as being impermissibly content based even though their primary purpose was indubitably content neutral. See Arkansas Writers' Project, Inc., supra (striking down content-based exemptions in a general revenue measure); Regan v. Time, Inc., supra (striking down contentbased exemptions in a general anticounterfeiting statute); Metromedia, Inc. v. San Diego, supra (plurality opinion) (striking down on content discrimination grounds a general urban beautification ordinance); Carey v. Brown, supra, at 466-468 (striking down on content discrimination grounds an ordinance aimed at preserving residential privacy). Of course, the mere possibility that a statute might be justified with reference to content is not enough to make the statute content based, and neither is evidence that some legislators voted for the statute for content-based reasons. But when a content-based justification appears on the statute's face, we cannot ignore it because another, content-neutral justification is present.

C

Content-based speech restrictions are generally unconstitutional unless they are narrowly tailored to a compelling state interest. Boos v. Barry, 485 U. S. 312, 321 (1988). This is an exacting test. It is not enough that the goals of the law be legitimate, or reasonable, or even praiseworthy. There must be some pressing public necessity, some essential value that has to be preserved; and even then the law must restrict as little speech as possible to serve the goal.

The interest in localism, either in the dissemination of opinions held by the listeners' neighbors or in the reporting of events that have to do with the local community, cannot be described as "compelling" for the purposes of the compelling state interest test. It is a legitimate interest, perhaps even an important one-certainly the government can foster it by, for instance, providing subsidies from the public fiscbut it does not rise to the level necessary to justify contentbased speech restrictions. It is for private speakers and listeners, not for the government, to decide what fraction of their news and entertainment ought to be of a local character and what fraction ought to be of a national (or international) one. And the same is true of the interest in diversity of viewpoints: While the government may subsidize speakers that it thinks provide novel points of view, it may not restrict other speakers on the theory that what they say is more conventional. Cf. Metro Broadcasting, Inc. v. FCC, 497 U. S. 547, 612-613 (1990) (O'CONNOR, J., dissenting); Pacific Gas & Elec. Co. v. Public Util. Comm'n of Cal., 475 U. S. 1, 20 (1986) (plurality opinion).

The interests in public affairs programming and educational programming seem somewhat weightier, though it is a difficult question whether they are compelling enough to justify restricting other sorts of speech. We have never held that the Government could impose educational content requirements on, say, newsstands, bookstores, or movie theaters; and it is not clear that such requirements would in any event appreciably further the goals of public education.

But even assuming, arguendo, that the Government could set some channels aside for educational or news programming, the Act is insufficiently tailored to this goal. To benefit the educational broadcasters, the Act burdens more than just the cable entertainment programmers. It equally burdens CNN, C-SPAN, the Discovery Channel, the New Inspirational Network, and other channels with as much claim as PBS to being educational or related to public affairs.

Even if the Government can restrict entertainment in order to benefit supposedly more valuable speech, I do not think the restriction can extend to other speech that is as valuable as the speech being benefited. In the rare circumstances where the government may draw content-based distinctions to serve its goals, the restrictions must serve the goals a good deal more precisely than this. See Arkansas Writers' Project, Inc., 481 U. S., at 231-232; Erznoznik v. Jacksonville, 422 U. S. 205, 214-215 (1975).

Finally, my conclusion that the must-carry rules are content based leads me to conclude that they are an impermissible restraint on the cable operators' editorial discretion as well as on the cable programmers' speech. For reasons related to the content of speech, the rules restrict the ability of cable operators to put on the programming they prefer, and require them to include programming they would rather avoid. This, it seems to me, puts this case squarely within the rule of Pacific Gas & Elec. Co., 475 U. S., at 14-15 (plurality opinion); id., at 23-24 (Marshall, J., concurring in judgment); see also Miami Herald Publishing Co. v. Tornillo, 418 U. S. 241, 257-258 (1974).

II

Even if I am mistaken about the must-carry provisions being content based, however, in my view they fail contentneutral scrutiny as well. Assuming, arguendo, that the provisions are justified with reference to the content-neutral interests in fair competition and preservation of free television, they nonetheless restrict too much speech that does not implicate these interests.

Sometimes, a cable system's choice to carry a cable programmer rather than a broadcaster may be motivated by anticompetitive impulses, or might lead to the broadcaster going out of business. See ante, at 661-668. That some speech within a broad category causes harm, however, does not justify restricting the whole category. If Congress wants to protect those stations that are in danger of going out of business, or bar cable operators from preferring programmers in which the operators have an ownership stake, it may do that. But it may not, in the course of advancing these interests, restrict cable operators and programmers in circumstances where neither of these interests is threatened.

"A regulation is not 'narrowly tailored' -even under the more lenient [standard applicable to content-neutral restrictionsJ-where... a substantial portion of the burden on speech does not serve to advance [the State's contentneutral] goals." Simon & Schuster, 502 U. S., at 122, n. (internal quotation marks omitted). If the government wants to avoid littering, it may ban littering, but it may not ban all leafleting. Schneider v. State (Town of Irvington), 308 U. S. 147 (1939). If the government wants to avoid fraudulent po litical fundraising, it may bar the fraud, but it may not in the process prohibit legitimate fundraising. Schaumburg v. Citizens for a Better Environment, 444 U. S. 620 (1980); see also Edenfield v. Fane, 507 U. S. 761, 776-777 (1993). If the government wants to protect householders from unwanted solicitors, it may enforce "No Soliciting" signs that the householders put up, but it may not cut off access to homes whose residents are willing to hear what the solicitors have to say. Martin v. City of Struthers, 319 U. S. 141 (1943). "Broad prophylactic rules in the area of free expression are suspect. Precision of regulation must be the touchstone.... " NAACP v. Button, 371 U. S. 415, 438 (1963) (citations omitted).

The must-carry provisions are fatally overbroad, even under a content-neutral analysis: They disadvantage cable programmers even if the operator has no anticompetitive motives, and even if the broadcaster that would have to be dropped to make room for the cable programmer would survive without cable access. None of the factfinding that the District Court is asked to do on remand will change this. The Court does not suggest that either the antitrust interest or the loss of free television interest are implicated in all, or even most, of the situations in which must-carry makes a difference. Perhaps on remand the District Court will find out just how many broadcasters will be jeopardized, but the remedy for this jeopardy will remain the same: Protect those broadcasters that are put in danger of bankruptcy, without unnecessarily restricting cable programmers in markets where free broadcasting will thrive in any event.

III

Having said all this, it is important to acknowledge one basic fact: The question is not whether there will be control over who gets to speak over cable-the question is who will have this control. Under the FCC's view, the answer is Congress, acting within relatively broad limits. Under my view, the answer is the cable operator. Most of the time, the cable operator's decision will be largely dictated by the preferences of the viewers; but because many cable operators are indeed monopolists, the viewers' preferences will not always prevail. Our recognition that cable operators are speakers is bottomed in large part on the very fact that the cable operator has editorial discretion. Ante, at 636-637.

I have no doubt that there is danger in having a single cable operator decide what millions of subscribers can or cannot watch. And I have no doubt that Congress can act to relieve this danger. In other provisions of the Act, Congress has already taken steps to foster competition among cable systems. § 3(a), 47 U. S. C. § 543(a)(2) (1988 ed., Supp. IV). Congress can encourage the creation of new media, such as inexpensive satellite broadcasting, or fiber-optic networks with virtually unlimited channels, or even simple devices that would let people easily switch from cable to overthe-air broadcasting. And of course Congress can subsidize broadcasters that it thinks provide especially valuable programming.

Congress may also be able to act in more mandatory ways.

If Congress finds that cable operators are leaving some channels empty-perhaps for ease of future expansion-it can compel the operators to make the free channels available to programmers who otherwise would not get carriage. See PruneYard Shopping Center v. Robins, 447 U. S. 74, 88 (1980) (upholding a compelled access scheme because it did not burden others' speech). Congress might also conceivably obligate cable operators to act as common carriers for some of their channels, with those channels being open to all through some sort of lottery system or time-sharing arrangement. Setting aside any possible Takings Clause issues, it stands to reason that if Congress may demand that telephone companies operate as common carriers, it can ask the same of cable companies; such an approach would not suffer from the defect of preferring one speaker to another. But the First Amendment as we understand it today rests on the premise that it is government power, rather than private power, that is the main threat to free expression; and as a consequence, the Amendment imposes substantiallimitations on the Government even when it is trying to serve concededly praiseworthy goals. Perhaps Congress can to some extent restrict, even in a content-based manner, the speech of cable operators and cable programmers. But it must do so in compliance with the constitutional requirements, requirements that were not complied with here. Accordingly, I would reverse the judgment below.