dissenting:
I agree with my colleagues that maintaining public confidence in the integrity of government is an interest of paramount importance. “[A] democracy is effective only if the people have faith in those who govern, and that faith is bound to be shattered when high officials and their appointees engage in activities which arouse suspicions of ... corruption.” United States v. Mississippi Valley Generating Co., 364 U.S. 520, 562, 81 S.Ct. 294, 315, 5 L.Ed.2d 268 (1961). Actions by public employees that suggest they are using their public offices for personal gain risk shaking public confidence, and that fact may in appropriate instances provide a justification for curtailing the right of employees to speak about job-related topics. But it is also settled constitutional doctrine that to justify infringing on the First Amendment rights of federal employees, the government must do more than nod its head in the direction of a significant interest it claims to protect. It must demonstrate that the specific regulation it proposes does not choke off more employees’ speech than is necessary to advance the government’s legitimate interest and that the regulation does not risk suppressing speech solely because of disagreement with the employees’ ideas. In my view, the Environmental Protection Agency’s (“EPA” or “agency”) interpretation, in its Ethics Advisory, of the Office of Government Ethics (“OGE”) regulation (the “OGE/EPA regulation”) does not pass that constitutional test.
The OGE’s regulation bars federal employees from “receiving compensation, including travel expenses, for speaking or writing on subject matter that focuses specifically on [their] official duties or on the responsibilities, policies, and programs of [their] employing agency.” 56 Fed.Reg. 1724-25 (1991) (to be codified at 5 C.F.R. § 2636.202(b)). “Travel expenses” are defined in the regulation to include the “actual and necessary cost of transportation, lodging and meals incurred while away from the employee’s residence or principal place of employment in connection with [a] ... speech or article.” 56 Fed.Reg. 1726 (1991) (to be codified at 5 U.S.C. § 2636.-203(g)). The EPA, in its Ethics Advisory, construed this prohibition to apply only to *452“non-official” travel expenses. EPA Ethics Advisory 91-1, at 3 (April 2, 1991).
My colleagues acknowledge that any appearance of impropriety arising from an employee receiving reimbursement for legitimate expenses involved in giving a speech on agency-related matters must be linked to an impression that the employee is receiving a valuable benefit from a private party. See Majority Opinion .(“Maj. Op.”) at 445. While an employee who receives lodging and meals at a deluxe vacation resort in return for giving a speech on matters related to her job might reasonably be thought to have improperly benefited from her public position, much reimbursement does not partake of “fun in the sun”; rather, it simply insures that the speaker will not be financially worse off because of her speechmaking. For example, a private group may simply repay the employee for the costs of gas and tolls incurred in traveling to the site of the speech. Because the OGE/EPA regulation is deaf to such distinctions, it is far too blunt an instrument for diminishing federal employees’ First Amendment rights in order to achieve the end of preserving public confidence in government.
Conversely, in many instances where expense reimbursements should raise eyebrows, the regulation is bafflingly oblivious. So long as EPA employees speak “officially,” they may continue to receive extravagant vacations to exotic locales on private parties’ tabs. Why the continued acceptance of such largesse does not raise the same problems that the government is apparently seeking to remedy through the total ban on privately reimbursed expenses for unofficial speech is inexplicable to me. The absence of any parallel provision concerning privately supported “official” speakers raises grave questions as to whether the government’s “significant” interest in avoiding the appearance of impropriety is served by this regulation at all, and at best impugns the force of the government’s interest in it. Additionally, the patently differential treatment accorded “officially” approved speakers and “unofficial” ones suggests that the regulation may indeed have the purpose or effect of suppressing speech based solely on the agency's disagreement with particular ideas. Under the OGE/EPA rule, the government’s “mouthpieces” (the majority’s term, see Maj. Op. at 448) — those who speak “officially” — are eligible to receive reimbursements from private sources. But people who speak only “unofficially,” who may have different views on current agency policies, are systematically dissuaded from speaking by the denial of any reimbursements even for barebones out-of-pocket expenses like bus fare or an overnight at the YMCA.
In sum, a total ban on privately reimbursed expenses for unofficial speeches on agency-related topics accompanied by a bland acceptance of privately reimbursed expenses for “approved” speeches that toe the current agency line is not a constitutionally tailored response to the problem of the appearance of impropriety arising from federal employees taking private benefits for work-related speeches. Such a regulation burdens the First Amendment rights of federal employees without bestowing comparable benefits on the public at large.
I. Discussion
A. Legal Standard
Pickering v. Board of Education, 391 U.S. 563, 88 S.Ct. 1731, 20 L.Ed.2d 811 (1968), the time-honored yardstick for measuring restrictions on government employees’ speech, is the governing standard here. Under that familiar rubric, we weigh EPA employees’ interest in speaking on matters of public concern against the EPA’s interest, as an employer, “in promoting the efficiency of the public services it performs through its employees.” Id. at 568, 88 S.Ct. at 1734.
B. Employees’ Interests
The majority significantly understates the burden that the OGE/EPA regulation places on public employees’ speech. Because the OGE/EPA regulation does not overtly gag anyone, they conclude that the burden is only “moderate.” See Maj. Op. at 441. I fear such a conclusion partakes *453of the “sheer formalism,” id. at 445, my colleagues elsewhere decry. Realistically, under this regulation, government employees — people of generally modest incomes— will not have the means to make speeches involving more than de minimis travel expenses; out-of-town speeches will be as surely eliminated as if the EPA prohibited them outright.1 See Brief of Amici Curiae Environmental and Civic Organizations in Support of Appellants 4 (noting that the regulation would effectively prevent environmental groups from obtaining live speakers from the EPA); see also Simon & Schuster, Inc. v. Members of the New York State Crime Victims Bd., — U.S. -, -, 112 S.Ct. 501, 511, 116 L.Ed.2d 476 (1991) (noting that application of statute preventing compensation — not mere reimbursement — for speech on certain subjects would have aborted many classic works, even though it is “hyperbole” to assume that law would have “prevented] publication of all” such works because some would have been written without compensation (emphasis in original)); cf. Minneapolis Star & Tribune Co. v. Minnesota Commissioner of Revenue, 460 U.S. 575, 585, 103 S.Ct. 1365, 1372, 75 L.Ed.2d 295 (1983) (noting that the threat of burdensome taxes “can operate as effectively as a censor to check critical comment”).
The regulation, moreover, effectively squelches speech of the highest value in the First Amendment hierarchy. Under Connick v. Myers, 461 U.S. 138, 103 S.Ct. 1684, 75 L.Ed.2d 708 (1983), employee speech is protected only when it deals with matters of “public concern,” i.e., when it contributes to debate on issues of community interest.2 See id. at 145, 103 S.Ct. at 1689 (noting Pickering’s roots in a mode of First Amendment analysis that reserves the “highest rung” for public issue speech). The OGE/EPA regulation restricts dissemination of information regarding current EPA policies, a matter on which EPA employees obviously have unique insights to share with the general populace, and, it must be noted, the regulation does so not any basis that these unofficial speeches are likely to be inaccurate, misleading, or expose confidential information. Depriving the general populace of a valuable and often unique perspective on important issues affecting their welfare requires a serious and carefully thought out justification. Cf. Cass R. Sunstein, Government Control of Information, 74 Cal.L.Rev. 889, 916 (1986) (noting that the “systemic consequences of government purchases of first amendment rights” must be considered even when there is a legitimate government interest in restriction of speech); Developments in the Law—Public Employment, 97 Harv.L.Rev. 1611, 1768 (1984) (‘‘Pickering recognized that public employees, by virtue of their expertise and experience, are often among the citizens who are best informed about the workings of government and that their opinions are thus especially valuable to the public.”); Kathleen M. Sullivan, The First Amendment Wars, The New Republic, September 28, 1992, at 35, 39 (“[Tjhose dependent on government funding may, for the reasons that created their dependency, have some distinctive contribution to make to the speech mix.”). A democratic nation’s interest in maintaining an informed citizenry, as well as an individual government employee’s interest in free expression, must be respected *454in striking an appropriate balance here. Cf. Minneapolis Star, 460 U.S. at 585, 103 S.Ct. at 1372 (“[A]n informed public is the essence of working democracy.”). The government cannot be allowed to limit speech of public import where its legitimate interest in guarding the integrity of its operations, is not directly implicated.
C. The Government’s Interest
The government asserts only one interest here to justify its ban on expense reimbursement: preventing the appearance of impropriety stemming from private subsidization of a public employee’s speechmaking on agency-related matters. Because the precise nature of the feared “impropriety” is somewhat nebulous, it is useful to examine exactly how the receipt of expense money from private parties might shake the public’s confidence in the integrity of its government and those who serve in it. Two possible scenarios come to mind: an appearance that the private party is currying favor from the federal employee and an appearance that the federal employee is using “public office for private gain.” EPA Brief at 24.
The first scenario plays out like this. When a private party pays for a special service by a government employee (or gives the employee a gift), people will immediately speculate about whether the private party expects an illicit favor in return. Such arrangements (especially ones between regulators and the regulated or the beneficiaries of regulation) raise obvious questions about the integrity and impartiality of government decisionmakers; put simply, the public may suspect that the employees are being “bought and paid for.” But this species of impropriety necessarily involves (1) a payor who cares about the future actions of the agency and (2) the receipt of a benefit from that payor by an agency employee. The subject matter of the speech or other service the employee provides to the private payor is not crucial; the private party doesn’t really care about the service that it is superficially paying for. Instead, it hopes to receive some future favor in exchange for its largesse. The nature of the service paid for is relevant only to the extent an outsider will think the employee’s service is worth the price; when the service appears overpriced, the inference that the private payor is actually buying something else is stronger.
The second species of impropriety also involves the presence of a benefit, but, in contrast to the first, it does require that the subject matter of the federal employee’s speech (or other service) relate to her job. When a public employee gets additional benefits from work that the government is already paying for, she may be thought to be improperly using “public office for private gain” — or, in the majority’s terms, serving two masters. The receipt of money from private donors for information related to the federal employee’s work, even when there is no reason to suspect that the donor is trying to curry favor, may cause the public to lose faith in the single-minded dedication of government employees to' the public interest. Cf. Robert G. Vaughn, Conflict-of-Interest Regulation in the Federal Executive Branch 37 (1979) (“[Pjractices that enrich government employees beyond their appropriate compensation solely because of their status as government employees undermine public faith and confidence.”); Association of the Bar of the City of New York, Conflict of Interest and the Federal Service 211 (1960) (when private parties pay federal employees to do their jobs it creates “a generally unwholesome appearance that breeds suspicion and bitterness among fellow employees” and outsiders).3 Although both scenarios may be implicated by the OGE/EPA regulation, its exclusive focus on employee speech related to current EPA duties or policy suggests strongly that it is primarily motivated by the latter one.
*455D. Overinclusiveness
Assuming that to be the case,41 conclude that the total ban on reimbursement is an overinclusive restriction on EPA employees’ First Amendment rights. In a prior Pickering case, we required that a rule be “narrowly drawn” to “restrict speech no more than is necessary” to serve the government’s limited goal. See McGehee v. Casey, 718 F.2d 1137, 1143 (D.C.Cir.1983) (internal citation omitted). A similar tailoring requirement should apply here because of the exceptionally strong First Amendment interests on the other end of the balance. Compare id. (requiring “narrow tailoring" where government sought to classify and prohibit speech on matters of public concern) with Williams v. IRS, 919 F.2d 745, 747 (D.C.Cir.1990) (per curiam) (requiring only that regulations be “tailored” where court could not find any firm First Amendment interest on the other side and merely assumed the existence of one arguendo); see also EPA Brief at 22-23 (arguing that the government’s burden should vary with strength of employees’ speech interests).5
The overinclusiveness in this regulation results from the breadth of its definition of a “benefit,” the sine qua non of an appearance of impropriety in this context. See Maj. Op. at 445. The rule does not outlaw only travel reimbursements that may be perceived to be compensation; instead, it prohibits the receipt of all “travel expenses,” which include the “actual and necessary cost of transportation, lodging and meals.” 56 Fed.Reg. 1726 (1991) (to be codified at 5 C.F.R. § 2636.203(g)). Thus, the regulation curtails reimbursement in a wide range of cases in which the public *456would not believe that a benefit has been conferred. A box lunch and bus fare to Buffalo as well as a lobster and Lear jet to Lake Tahoe are equally verboten under this rule. Indeed, the box lunch alone would run afoul of the regulation.6 Yet it is almost incomprehensible that an ordinary citizen would look askance at a federal employee addressing a church supper in return for train fare and a few bites of casserole.7
But, I would emphasize, what is involved here is far more than a judicial quibble over the fine tuning of a rule. Contrary to my colleagues’ assertion that there are “comparatively few cases” where the regulation might be unconstitutionally applied, see Maj. Op. at 446, the wide gulf between the means — restricting all reimbursement whether or not the public could conceivably believe that a benefit has been received— and the end — avoiding the appearance of impropriety caused by the receipt of a benefit — imposes a wide-reaching and deleterious effect on federal employees’ free speech, as well as on potential listeners’ opportunity to become informed about public affairs. Many small nonprofit environmental organizations that reimburse expenses for “unofficial” speeches by EPA employees argue as amici against the regulation. The accommodations most such groups can provide are exceedingly modest, generally only enough to insure that the speaker will not lose money for his efforts. In such cases, no benefit has been bestowed and thus the reimbursement does not raise concerns about impropriety, but the regulation still curtails the dissemination of valuable information. A more tailored regulation8 could put a reasonable cap on reimbursement (akin to the federal employee’s per diem allowance) thereby assuring that the employee is not dissuaded from speaking because of financial loss but also avoiding any side benefits from his doing so. In truth, the OGE/EPA regulation is.something of a blunderbuss, and the government’s use of it is flatly inconsistent with the more precise targeting required by the First Amendment for regulating valuable speech interests. See, e.g., Simon & Schuster, — U.S. at ---, 112 S.Ct. at 511-12 (finding statute “not narrowly tailored” because it swept in a wide range of expression that did not serve the government’s interest in preventing criminals from profiting while victims go uncompensated); Carey v. Brown, 447 U.S. 455, 465, 100 S.Ct. 2286, 2292, 65 L.Ed.2d 263 (1980) (statute proscribing all nonlabor picketing of residences was overinclusive because “the statute makes no attempt to distinguish among various sorts of nonla-bor picketing on the basis of the harms they would inflict” on the state’s asserted interest in residential privacy).
The majority, of course, reaches a different conclusion. They base their approval of the regulation on the theory that the “appearance-of-impropriety reasoning applies with full force as long as the benefit the employee receives has some positive *457value.” Maj. Op. at 446 (emphasis in original); see also id. at 446 (citing the regulatory provision that defines “travel expenses” to include all actual and necessary costs of travel in support of the proposition that the regulation applies “only where the employee has received a benefit”). Thus, the majority apparently assumes that the public will view private reimbursement as suspect not only when the speaker receives the type of pleasure and benefit associated with premium transportation and accommodations — a situation exemplified by the majority’s hypothetical employee who “fly[s] first-class, stay[s] at an exclusive resort and dine[s] at five-star restaurants,” Maj. Op. at 445 — but also when a private party merely offsets some or all of the speaker’s actual and reasonable expenses, no matter how small the reimbursement or how pedestrian the accommodations. Because, as indicated above, see supra page 440, I cannot believe that the public will be suspicious — will lose faith in the integrity of its government — when a federal employee is only receiving reimbursement for reasonable expenses such as gas and tolls, I part company with my colleagues on this basic point.
Nor can I concur in their alternative argument that as the reimbursement decreases in value, the disincentive to speak and the government’s interest in preventing the appearance of impropriety decrease in lockstep. To a federal employee taking home several hundred dollars a week, train fare and similar expenses that the majority characterizes as “trivial,” see Maj. Op. at 446, are significant expenditures; thus, a requirement that the employee pay such expenses out-of-pocket creates a significant disincentive to out-of-town speaking. At the same time, however, reimbursement for such modest expenses does not raise an inference that the employee is traveling in luxury — receiving compensation in the form of fine dining and lush resort accommodations — and thus cannot reasonably be thought to raise appearance of impropriety problems. Because in the large number of cases that fit in this middle ground between de minimis and premium reimbursements there is a substantial disincentive to speak but no countervailing government interest, I conclude that the regulation is not narrowly tailored.
I could stop there. An ábysmal lack of fit between the perceived problem and the solution causes the reimbursement ban to flunk the constitutional test for facial validity set out in McGehee. However, I believe that the regulation has other significant defects as well.
E. Underinclusiveness
The majority believes that the underin-clusiveness of this regulation is, at most, of limited relevance in striking the Pickering balance in this case. See Maj. Op. at 443 n. 9. I cannot agree. The underinclusiveness of this speech-restricting rule is relevant in two ways. My first, although not my primary, point is that when a statute or regulation does not in fact cover a substantial part of the activity giving rise to the harm the government claims as its target, that fact “raises serious doubts about whether [the government] is, in fact, serving ... the significant interests which [it] invokes.” Florida Star v. B.J.F., 491 U.S. 524, 540, 109 S.Ct. 2603, 2612, 105 L.Ed.2d 443 (1989); id. at 541, 109 S.Ct. at 2613 (“Without more careful and inclusive precautions against alternative forms of dissemination, we cannot conclude that Florida’s selective ban on publication by the mass media satisfactorily accomplishes its stated purpose.”); see also id. at 541-42, 109 S.Ct. at 2613 (Scalia, J., concurring) (“[A] law cannot be regarded as protecting an interest ‘of the highest order’ ... when it leaves appreciable damage to that supposedly vital interest unprohibited.” (internal citation omitted)); FCC v. League of Women Voters, 468 U.S. 364, 396, 104 S.Ct. 3106, 3126, 82 L.Ed.2d 278 (1984) (assuming that government’s asserted interest in the law is legitimate, its “substantiality” is “dubious” because of underinclusiveness and overinclu-siveness). Although my colleagues insist that the Supreme Court has announced this principle only in cases outside the Pickering context, I see no logical reason why it should not apply to Pickering cases as well. While the government’s interest *458need not be as strong under Pickering as under the strict scrutiny standard applied in other First Amendment cases, under both tests we must evaluate the substan-tiality — the weight — of the proffered government concern: under strict scrutiny, we ask whether the government’s rule is serving a compelling state interest; under Pickering, we inquire whether the government’s interest is stronger than the private interest asserted in a particular case. And the Supreme Court has told us that the underinclusiveness of a regulation is definitely germane to the generic type of assessment involved in either inquiry. See, e.g., id.
Because the OGE/EPA regulation leaves so much undone, I conclude that the government’s interest in the regulation is not as strong as my colleagues assert and thus that the weight on the government’s end of the scale is lighter than they contend.9 An EPA employee may still receive significant benefits from private parties — “fly first-class, stay at an exclusive resort and dine at five-star restaurants,” Maj. Op. at 445— for giving “official” speeches relating to her duties or EPA policy. Indeed, the record in this case demonstrates that such privately funded benefits have flowed freely at the EPA. To cite a few examples, EPA employees have gone “officially” to Santa Barbara on Chevron, U.S.A.’s tab, to Scottsdale, Arizona on the Water Utilities Executive Council's account, and to Ft. Lauderdale on the American Coal Ash Association’s account. Why those reimbursements do not raise the same appearance of impropriety problems as the receipt of expenses from “unofficial” speeches totally escapes me (and I would assume others as well).10 In both cases, a benefit — if indeed it is assumed that every reimbursement is a benefit — is received from a private party for receipt of information related to the employees’ work. In fact, it would seem logical that officially sanctioned benefits from private sources would create a greater appearance of systemic governmental impropriety than the unsanctioned indiscretions of individual bureaucrats and, of the two, would produce the more deleterious effect on public confidence in the workings of our government.
The only defense that the majority gives for the critical relevance of the official/unofficial distinction is the assumption that, when speaking officially, an agency employee is “under the watchful eye of the EPA.” Maj. Op. at 448. Even if this totally unsupported empirical assertion were correct, it is unclear why it is even relevant. If it is the receipt of expense reimbursement from a private source that creates an appearance of impropriety, no amount of official scrutiny can dissolve that problem. Whether or not the agency is “watching” the employee, she still gets the so-called “benefits” of the enjoyable trip, those benefits still have a nexus to her federal employment, and they are still being paid by a private party (and possibly even a private party with a conflict of interest with the EPA, see EPA Ethios Advisory 91-10, at 3 (August 30, 1991) (allowing reimbursement from a “conflicting source” under certain circumstances)). In fact, it was the perception of impropriety from the receipt of authorized benefits such as honoraria from *459private parties that led to the passage of the Ethics Reform Act.
Perhaps agency supervision would be relevant if employees had no control over their official travel because the agency fielded all requests for speakers and par-celled out the assignments. But that is not the case here. While an EPA employee may not solicit invitations, the employee may receive them personally and decide if she wants to seek official approval to accept them. See id. at 2. Given, then, that the employee herself makes the initial determination whether a particular trip confers a benefit worth pursuing, it is hard to see why official “oversight” is likely to make the public less suspicious of such private sponsorship. But even if such oversight were somehow relevant because it brought the transaction “out in tbe open,” it would be constitutional overkill to ban all unofficial reimbursement on that basis when a sensible requirement of public documentation represents a far more tailored response to the perceived problem.
In the end, I can find no basis for concluding that the official/unofficial distinction has any relevance to the government’s interest in preventing the appearance of impropriety. Cf. Simon & Schuster, — U.S. at -, 112 S.Ct. at 510 (rejecting a proffered government interest in regulating a subset of speech because the state could not “explain why [it] should have any greater interest in compensating victims from the proceeds of ... ‘storytelling’ than from any of the criminal’s other assets”).
That brings me to my second, and primary, concern, which is perhaps better described as about viewpoint discrimination, rather than underinclusiveness per se.11 See R.A.V. v. City of St. Paul, — U.S. -, -, 112 S.Ct. 2538, 2545, 120 L.Ed.2d 305 (1992). The Supreme Court has repeatedly noted that when a statute is “underinclusive” in a way that turns on content (even more so for viewpoint), the risk of idea suppression is almost always present. See, e.g., id.; see also Erznoznik v. City of Jacksonville, 422 U.S. 205, 215, 95 S.Ct. 2268, 2275, 45 L.Ed.2d 125 (1975) (presumption of validity of underinclusive legislation “has less force when a classification turns on the subject matter of expression,” and statute could not be upheld absent a justification for differentiating between activities which would appear to cause similar harms); cf. City of Lakewood v. Plain Dealer Publishing Co., 486 U.S. 750, 762-63, 108 S.Ct. 2138, 2146-47, 100 L.Ed.2d 771 (1988) (noting that the greater includes the lesser syllogism “is blind to the radically different constitutional harms inherent in the ‘greater’ and ‘lesser’ restrictions”).
That concern is apt here. The OGE/ EPA regulation does not merely draw an irrelevant, but benign, distinction. The majority acknowledges that the rule allows an EPA employee to receive reimbursement for travel expenses when the agency has “sent him to present its views,” and he is acting “much like a mouthpiece,” Maj. Op. at 448 (emphasis added), but not when he is free to present his own perspective on issues of current EPA policy (which would comport with the official line only coincidentally).12 Thus, an EPA employee’s re*460ceipt of the “benefit” of private reimbursement of travel expenses is conditioned upon her speech presenting the EPA’s point of view, “touting its policies,” id. at 442 n. 7. My colleagues assert, however, that the OGE/EPA rule’s distinction is not “real” viewpoint discrimination because the government must always be able to exempt its own spokespeople from any speech regulation it imposes on other employees. See id. If the EPA merely used its own funds to pay its “mouthpieces” and did not interfere with speech by others in any way, my colleagues would be correct that viewpoint bias concerns were off base. See Rust v. Sullivan, — U.S. -, -, 111 S.Ct. 1759, 1772, 114 L.Ed.2d 233 (1991) (in selectively funding activities it believes to be in the public interest, the “Government has not discriminated on the basis of viewpoint”). But EPA has gone far beyond such an unexceptional arrangement. First, the disputed regulation relates only to privately funded speech; that fact by itself raises different concerns than a Rust-type case. See FCC v. League of Women Voters, 468 U.S. 364, 400, 104 S.Ct. 3106, 3128, 82 L.Ed.2d 278 (1984) (invalidating content-based restriction on speech that was funded from both public and private sources). More importantly, the rule controls only some privately funded speech; it curtails only speech that may not present the agency’s views. It is clear that, under Pickering, such regulations raise viewpoint bias concerns. See Rankin v. McPherson, 483 U.S. 378, 384, 107 S.Ct. 2891, 2897, 97 L.Ed.2d 315 (1987) (“Vigilance is necessary to ensure that public employers do not use authority over employees to silence discourse, not because it hampers public functions but simply because superiors disagree with the content of employees’ speech.”). While we have approved differential treatment of speech critical of a public employer in specific instances, we have done so because of the significance of the government’s countervailing interest, not because the regulation did not implicate concerns about viewpoint bias. See supra note 10 (noting that in some contexts the government may have a significant interest in preventing critical speech by policymaking employees but noting that that interest, even if asserted, would not justify a regulation applicable to all EPA employees).
Thus, the viewpoint-based distinction inherent in this regulation implicates core First Amendment concerns because it risks skewing debate; the government’s side has open access to particular audiences and forms of expression but employees who seek to speak from other perspectives must overcome a financial disincentive.13 See, e.g., R.A.V. v. City of St. Paul, — U.S. at ---, 112 S.Ct. at 2547-48 (facially invalidating a law that in practice allowed *461“ ‘fighting words’ that do not themselves invoke race, color, creed, religion, or gender ... in the placards of those arguing in favor of racial, color, etc. tolerance and equality” but not in the placards of those who opposed those ideas (emphasis in original)); id. at -, 112 S.Ct. at 2543 (even in the extreme case where an entire category of speech may be proscribed because of its negligible value, such regulation may not be “made the vehicle[] for content [much less viewpoint] discrimination unrelated to th[at type of speech’s] distinctively pro-scribable content”). To escape invalidation, the government must, at the very least, demonstrate a significant interest in regulating this subclass of speech. See Hall v. Ford, 856 F.2d 255, 264 (D.C.Cir.1988) (finding restriction on public employee’s critical speech justified by a significant interest because the relationship between the employee and the government fell “ ‘into that narrow band of fragile relationships requiring for job security loyalty at the expense of unfettered speech’ ” (quoting Gonzalez v. Benevides, 712 F.2d 142, 150 (5th Cir.1983)); see also R.A.V., — U.S. at -, 112 S.Ct. at 2546 (within presumptively proseribable class of speech “differential treatment [of] ... a content-defined subclass of proseribable speech” is valid if that “subclass happens to be associated with particular ‘secondary effects’ of the speech.”). Because the only interest that my colleagues assert to justify the differential treatment of this subclass — the unique interest in avoiding the appearance of impropriety where the EPA’s watchful eyes are not present — is, at best, insubstantial, I conclude that the EPA has not justified this blanket restriction of private reimbursement for “nonofficial” speeches by all EPA employees with a sufficiently weighty countervailing government concern. Thus, the EPA rule has created a process that impermissibly risks suppression of ideas based on their viewpoint. Even if the other defects in the law were insufficient to merit invalidation, that fact in my view would put the kibosh to the regulation.
II. Conclusion
Government employee speech is protected by the First Amendment, and can be restricted only where and to the extent the government can show its legitimate interests require restriction. See Pickering v. Board of Education, 391 U.S. 563, 568, 88 S.Ct. 1731, 20 L.Ed.2d 811 (1968). This regulation throttles much valuable speech that does not implicate the EPA’s interests in avoiding an appearance of impropriety from private reimbursement of the expenses incurred in giving a speech about the agency. At the same time, the reimbursement of “official” speechmaking by the same private parties, which might be expected to raise even greater concerns about impropriety, is left unchecked. Finally, as the majority astutely recognizes, the regulation allows private support of agency “mouthpieces” but not of individuals who might voice contrary opinions. Taking account of all these factors, the Pickering balance tips decisively against the government. I therefore dissent.
. My colleagues claim, see Maj. Op. at 441 n. 5, that their conclusion as to the weight of the employee’s interest accords with the Supreme Court’s admonition in Rust v. Sullivan, — U.S. -, 111 S.Ct. 1759, 114 L.Ed.2d 233 (1991), that ‘"[t]here is a basic difference between direct state interference with a protected activity and state encouragement of an alternative activity.'” Id. at -, 111 S.Ct. at 1772 (quoting Maher v. Roe, 432 U.S. 464, 475, 97 S.Ct. 2376, 2383, 53 L.Ed.2d 484 (1977)). That statement, however, points in the other direction. The Rust Court was emphasizing that the government’s use of its own funds to promote certain activities (childbirth) and not others (abortion) does not raise the same concerns that exist when the state directly regulates protected private activity. In this case, the government is doing the latter — engaging in "direct state interference with a protected activity’’ — by regulating the availability of private funding for protected speech.
. My colleagues acknowledge that the speech at issue here easily clears the public concern hurdle. See Maj. Op. at 440.
. Of course, not all private gain from federal employment raises concerns. Generally, we do not begrudge former federal employees the opportunity to profit from experience gained in public service, as long as their post-employment behavior itself does not directly contradict ethical norms. See Beth Nolan, Public Interest, Private Income, 87 Nw.Univ.L.Rev. 57, 80 (1992).
. This assumption in no way prejudices the government’s case. The regulation is significantly less tailored to address the problems caused by the first scenario. In any case, both scenarios require a benefit, and the absence of a benefit is the key to the overinclusive nature of the rule.
. My colleagues assert that traditional over-breadth doctrine allows challenges to a rule’s fit only by plaintiffs who admittedly fall within the constitutionally permissible reach of a regulation on behalf of third parties not before the court. See Maj. Op. at 443-44. At the same time, however, they acknowledge that the Supreme Court has recognized that
... "overbreadth” is not used only to describe the doctrine that allows a litigant whose own conduct is unprotected to assert the rights of third parties to challenge a statute, even though "as applied" to him the statute would be constitutional [citation omitted]. “Over-breadth” has also been used to describe a challenge to a statute that in all its applications directly restricts protected First Amendment activity and does not employ means narrowly tailored to serve a compelling governmental interest. Schaumburg [v. Citizens for a Better Environment, 444 U.S. 620,] 637-39 [100 S.Ct. 826, 836-37, 63 L.Ed.2d 73] (1980); First National Bank of Boston v. Bellotti, 435 U.S. 765, 786 [98 S.Ct. 1407, 1421, 55 L.Ed.2d 707] (1978); Zwickler v. Koota, 389 U.S. 241, 250 [88 S.Ct. 391, 396, 19 L.Ed.2d 444] (1967). Cf. City Council of Los Angeles v. Taxpayers for Vincent, [466 U.S. 789, 797, 104 S.Ct. 2118, 2124, 80 L.Ed.2d 772 (1984)] (recognizing the validity of a facial challenge but suggesting that it should not be called "overbreadth”)....
.... Whether that challenge should be called "overbreadth” or simply a facial challenge, the point is that there is no reason to limit challenges to case-by-case "as applied" challenges when the statute on its face and therefore in all its applications falls short of constitutional demands.
Secretary of State v. Joseph H. Munson Co., 467 U.S. 947, 965 n. 13, 104 S.Ct. 2839, 2852 n. 13, 81 L.Ed.2d 786 (1984). Yet, my colleagues go on to cite a different passage from Munson for the proposition that to prevail on such a claim, a first- or third-party "plaintiff must carry the heavy burden of showing that ... ‘there is no core of easily identifiable and constitutionally proscribable conduct that the statute prohibits.’’’ Maj. Op. at 444 (quoting Munson, 467 U.S. at 965, 104 S.Ct. at 2852). Just last Term, however, the Supreme Court, after determining that a statute was not narrowly tailored, unanimously invalidated it without engaging in any such inquiry. See Simon & Schuster, - U.S. at ---, 112 S.Ct. at 511-12 ("The State’s interest in compensating victims from the fruits of crime is a compelling one, but the Son of Sam law is not narrowly tailored to advance that objective. As a result, the statute is inconsistent with the First Amendment.”). Also, despite my colleagues’ suggestion that appellants do not argue that the statute is overinclusive as applied to third parties, see Maj. Op. at 444 n. 10, it is evident to me that they have made just that claim. See Appellants’ Brief at 34 ("[T]he OGE/EPA regulation is unquestionably ‘overin-clusive.’ For example, employees could be fired for merely accepting a beverage at a speaking event where the employee talked about EPA policy. There is no question that such behavior does not present an 'appearance of impropriety.’ However, the regulation does not distinguish between minor [and] major reimbursements.”).
. Further, the regulation would apply even if the employee loses income because he has taken uncompensated leave to give the speech.
. Contrary to my colleagues’ assertion, see Maj. Op. at 441 n. 6, 446-47, the use of examples such as these (which are not very far afield from the facts of this case, i.e., NC WARN's offer to pay appellants’ actual travel expenses from Washington, D.C. to North Carolina) to dramatize the utter lack of fit between a rule and the government’s interest is familiar in facial challenge cases. For example, in Simon & Schuster, the Court argued that because the law would apply to, the autobiography of a prominent figure that included a brief reference to stealing a worthless item as a youthful prank — certainly a fact situation not before the Court, see id. at 447 — it was "to say the least, not narrowly tailored to achieve the State’s objective." — U.S. at -, 112 S.Ct. at 512.
.For example, the Ethics Reform Act defines honoraria to exclude "actual and necessary travel expenses.” 5 U.S.C. app. § 505(3) (emphasis added). Similarly, in the EPA’s own evaluations of whether reimbursement for "official” speechmaking is allowable, it differentiates between allowable expenses and "premium” accommodations, for which an employee must provide additional explanation. Interestingly, however, the EPA does not put a firm cap on the amount of allowable reimbursement for "official” travel. First-class airfare and accommodations are acceptable as long as similar arrangements are available to other individuals attending a meeting or function. See EPA Ethics Advisory 91-10, at 3 (August 30, 1991).
. Of course, the weight on the government’s end of the scale is particularly important here because the regulation affects such significant speech interests and because so much of that effect falls on speech that does not implicate the government’s interests.
. The official/unofficial distinction may be relevant to some other government interests not asserted here. For example, the government may have a legitimate interest in employee loyalty that justifies preventing certain policymak-ing employees from criticizing the official position on certain matters. See Hall v. Ford, 856 F.2d 255, 265 (D.C.Cir.1988); see also Rankin v. McPherson, 483 U.S. 378, 390-91, 107 S.Ct. 2891, 2900, 97 L.Ed.2d 315 (1987) (noting that interest in punishing government employee speech based on content must be examined in relation to the employee’s authority and public accountability). But this regulation covers virtually all EPA employees, not just those who hold key policy positions. Also, the government might have an interest in keeping certain proprietary or sensitive information confidential, and so wish to prevent employees from disseminating important information without prior approval. That concern has also not been raised here.
. The majority is at most only partially correct in stating that underinclusiveness is significant only if it "clearly gives rise to an inference that the real motivation behind the statute or regulation is the suppression of speech." Maj. Op; at 446. The first reason I identified for looking at underinclusiveness — to assess the weight of the government’s interest in the particular regulation — has nothing to do with legislative intent. Additionally, the Supreme Court has often found underinclusiveness relevant without looking explicitly at motive. See, e.g., Minneapolis Star & Tribune Co. v. Minnesota Commissioner of Revenue, 460 U.S. 575, 592, 103 S.Ct. 1365, 1375, 75 L.Ed.2d 295 (1983); see also News America Publishing, Inc. v. FCC, 844 F.2d 800, 805 (D.C.Cir.1988) ("Accepting th[e] intuition [that singling out some subcategory of speech may signal an intent to censor] without making an actual determination of legislators’ motives, the Supreme Court has for the regulation of speech insisted on a closer fit between a law and its apparent purpose than for other legislation.”).
. The majority is between a rock and a hard place on this issue. In its attempt to distinguish the harms caused by "official” and "unofficial” speech reimbursements, it acknowledges that the regulation only allows speech by those who present the agency's views. See Maj. Op. at 448. But at other points the opinion is less forthright. *460See id. at 441-42 (arguing that we can only determine effect of the official/unofficial distinction after empirical investigation). The majority refers to a regulation that defines the term "official" in reference to speaking offers "tendered because of the employee’s EPA position," as opposed to the employee’s "individual knowledge or accomplishments." See id. at 442. That gloss leads me to the same conclusion. When one is speaking because of one's position, one is presumably giving the views associated with that position.
If “official” does not imply this meaning, then I do not know what the agency does mean. In contrast to prior cases, see, e.g., McGehee v. Casey, 718 F.2d 1137, 1144-45 (D.C.Cir.1983), there are no precise standards to cabin the EPA’s discretion and assure that some criteria whose application would be subject to review will be regularly applied. Absent such standards, the regulation allows the unfettered discretion that presents the “risk of suppression of speech” justifying a facial challenge. See City of Lakewood v. Plain Dealer Publishing Co., 486 U.S. 750, 757-59, 108 S.Ct. 2138, 2144-45, 100 L.Ed.2d 771 (1988). Also, I cannot agree with my colleagues that this argument has not been urged on this court. The appellants discuss the EPA’s “unfettered discretion” several times in their briefs. See Appellants Brief at 16, 35. While those discussions are not extensive, they do make my point: that the EPA has not directed our attention to any standards that will insure that its regulation will be applied neutrally.
. This fact differentiates this case from Ater v. Armstrong, 961 F.2d 1224 (6th Cir.), cert. denied, — U.S. -, 113 S.Ct. 493, 121 L.Ed.2d 431 (1992), which the majority finds persuasive. There, the regulation distinguished only between types of speech — soliciting funds was allowed on roadways, but other forms of speech were not. There was no classification based on content or viewpoint, and thus less risk that underinclusiveness might reflect discrimination on the basis of disagreement with ideas.