concurring in the result.
At the time the military retreated into the YMCA and the Masonic Temple, three alternative courses of action were open to the army commander. First, the troops could have continued their prior strategy and stood their ground in front of the buildings without returning the rioters’ hostile sniper fire; second, the troops could have stood their ground and attempted to repel the mob by the use of deadly force; third, the troops could have retreated from the entire area, leaving the mob temporarily in control. The petitioners argue that if the troops had adopted either of the first two of these alternative strategies, their buildings would not have suffered the damage which resulted from the military’s occupation.
But what if the military had adopted the third strategy open to it? If the army had completely abandoned the area to the rioters, and regrouped for a later counterattack, there can be little doubt on this record that the rioters would have subjected the buildings to greater damage than that which was in fact suffered. I believe this fact to be decisive. For it appears to me that, in riot control situations, the Just Compensation Clause *?may only be properly invoked when the military had reason to believe that its action placed the property in question in greater peril than if no form of protection had been provided at all.
I.
I start from the premise that, generally speaking, the Government’s complete failure to provide police protection to a particular property owner on a single occasion does not amount to a “taking” within the meaning of the Fifth Amendment. Every man who is robbed on the street cannot demand compensation from the Government on the ground that the Fifth Amendment requires fully effective police protection at all times. The petitioners do not, of course, argue otherwise. Yet surely the Government may not be required to guarantee fully effective protection during serious civil disturbances when it is apparent that the police and the military are unable to defend all the property which is threatened by the mob. If the owners of unprotected property remain uncompensated, however, there seems little justice in compensating petitioners, who merely contend that the military occupation of their buildings provided them with inadequate protection.
Petitioners’ claim that they may recover on a bare showing that they were afforded “inadequate” protection has an additional defect which should be noted. If courts were required to consider whether the military or police protection afforded a particular property owner was “adequate,” they would be required to make judgments which are best left to officials directly responsible to the electorate. In the present case, for example, petitioners could argue that it was possible for the troops to maintain their position in front of the buildings if they had been willing to kill a large number of rioters. In rebuttal, the Government could persuasively argue that the indiscriminate use of deadly force would have en*96raged the mob still further and would have increased the likelihood of future disturbances. Which strategy is a court to accept? Clearly, it is far sounder to defer to the other duly constituted branches of government in this regard.
It is, then, both unfair and unwise to favor those who have obtained some form of police protection over those who have received none at all. It is only if the military or other protective action foreseeably increased the risk of damage that compensation should be required. Since, in the present case, the military reasonably believed that petitioners’ property was better protected if the troops retreated into the buildings, rather than from the entire area, the property owners have no claim to compensation on the ground that the protection afforded to them was “inadequate.”
I must emphasize, however, that the test I have advanced should be applied only to government actions taken in an effort to control a riot. The Army could not, for example, appropriate the YMCA today and claim that no payment was due because the building would have been completely demolished if the military had not intervened during the riot. Once tranquility has been restored, property owners may legitimately expect that the Government will not deprive them of the property saved from the mob. But while the rioters are surging through the streets out of control, everyone must recognize that the Government cannot protect all property all of the time. I think it appropriate to say, however, that our decision today does not in any way suggest that the victims of civil disturbances are undeserving of relief. But it is for the Congress, not this Court, to decide the extent to which those injured in the riot should be compensated, regardless of the extent to which the police or military attempted to protect the particular property which each individual owns.
*97II.
While I agree with the Court that no compensation is constitutionally available under the facts of this case, I have thought it appropriate to state my own views on this matter since the precise meaning of the rules the majority announces remains obscure at certain critical points. Moreover, in deciding this particular case we should spare no effort to search for principles that seem best calculated to fit others that may arise before American democracy once again regains its equilibrium.
The Court sets out two tests to govern the application of the Just Compensation Clause in riot situations. It first denies petitioners recovery on the ground that each was the “particular intended beneficiary” of the Government’s military operations. Ante, at 92. I do not disagree with this formula if it means that the Fifth Amendment does not apply whenever the policing power reasonably believes that its actions will not increase the risk of riot damage beyond that borne by the owners of unprotected buildings. But the language the Court has chosen leaves a good deal of ambiguity as to its scope. If, for example, the military deliberately destroyed a building so as to prevent rioters from looting its contents and burning it to the ground, it would be difficult indeed to call the building’s owner the “particular intended beneficiary” of the Government’s action. Nevertheless, if the military reasonably believed that the rioters would have burned the building anyway, recovery should be denied for the same reasons it is properly denied in the case before us. Cf. United States v. Caltex, Inc., 344 U. S. 149 (1952).
Moreover, the Court’s formula might be taken to indicate that if the military’s subjective intention was to protect the building, the courts need not consider whether this subjective belief was a reasonable one. *98While the widest leeway must, of course, be given to good-faith military judgment, I am not prepared to subscribe to judicial abnegation to this extent. If a court concludes, upon convincing evidence, that the military had good reason to know that its actions would significantly increase the risk of riot damage to a particular property, compensation should be awarded regardless of governmental good faith.
While I accept the Court's “intended beneficiary” test with these caveats, I cannot subscribe to the second ground the majority advances to deny recovery in the present case. The majority analogizes this ease to one in which the military simply posted a guard in front of petitioners' properties. It is said that if the rioters had damaged the buildings as a part of their attack on the troops standing in front of them, the property damage caused would be too “indirect” a consequence of the military’s action to warrant awarding Fifth Amendment compensation. It follows, says the Court, that even if the military's occupation of the buildings increased the risk of harm far beyond any alternative military strategy, the Army’s action is nevertheless too “indirect” a cause of the resulting damage.
This argument, however, ignores a salient difference between the case the Court hypothesizes and the one which we confront. If the troops had remained on the street, they would not have obtained any special benefit from the use of petitioners’ buildings. In contrast, the military did in this instance receive a benefit not enjoyed by members of the general public when the troops were ordered to occupy the YMCA and the Masonic Temple. As the Court’s statement of the facts makes clear, the troops retreated into the buildings to protect themselves from sniper fire. Ordinarily, the Government pays for private property used to shelter its officials, *99and I would see no reason to make an exception here if the military had reason to know that the buildings would have been exposed to a lesser risk of harm if they had been left entirely unprotected.
On the premises set forth in this opinion, I concur in the judgment of the Court.