dissenting, in which BEEZER, Circuit Judge, joins in parts I, II, and III:

I write in dissent for four reasons. First, in my judgment, the majority’s opinion does not adequately differentiate between conduct that causes other conduct, and an agreement to violate or to disregard the law. The latter, of course, is the essence of a conspiracy; the former is *1548 not. I am concerned, with all due respect, that although the majority is clearly saying “conspiracy,” it may nevertheless be thinking “causation.”

Second, the majority, in arriving at its conclusion, gives too much importance to a handful of words spoken at a post-riot meeting attended by representatives of Phelps Dodge, command level of law enforcement, and the County Attorney’s office. Phelps Dodge, in the shadow of considerable unrest and violence surrounding its attempt to keep its business open, not surprisingly asked at this meeting whether arrests would be made. When the answer was in the affirmative, Phelps Dodge opined that if so, bond should be set high enough to keep the arrestees off the street. Even viewed in the context of all the facts and circumstances surrounding this case, which apparently aren’t enough collectively to carry the day, these words are hardly proof of a conspiratorial agreement.

Third, I am concerned that we improvidently create difficulties for people honestly attempting to work with public authorities to restore the rule of law during a public disorder.

And fourth, I am troubled by the manner in which we review the decision of the trial court, according no deference whatsoever to his view of the facts and the inferences that might be legitimately drawn therefrom.

I

Conspiracy v. Causation

Phelps Dodge is a private entity. It can be brought into this cause of action — which requires a showing of state action — only by proving that Phelps Dodge was part of a conspiracy, not merely that Phelps Dodge caused something to happen.

This is the teaching of Fonda v. Gray, 707 F.2d 435 (9th Cir.1983), in which we stated:

Fonda, 707 F.2d at 437-38 (citations omitted) (emphasis added). Although the majority opinion pledges fidelity to Fonda , in its analysis it gives the appearance of marching to the beat of a different drummer.

The source of this problem is found in Arnold v. International Business Machines, 637 F.2d 1350, 1356-57 (9th Cir.1981), where the analysis of the private party-state action issue was handled in terms of whether the private defendants “proximately caused” the constitutional deprivation. In so doing, the court in Arnold came to the correct result, but I believe respectfully that it did so by posing and answering the wrong question. The question was not whether IBM had “proximately caused” the constitutional deprivation, but whether IBM had conspired with the police to deprive the plaintiff of a constitutional right. The court in Arnold viewed the term “proximate cause” as virtually synonymous with “conspiracy,” but the two are markedly different and are not susceptible of such a marriage. What gets lost in this union is the requirement of a conspiracy that it be shown that the parties entered into an agreement to violate or to disregard the law. The use of the term “proximate cause” in this context, therefore, may tend to lead one down the wrong *1549 path with the wrong flashlight looking for the wrong quarry.

Under similar circumstances, the shared unlawful intent requirement was emphasized in Adickes v. Kress & Co., 398 U.S. 144 , 90 S.Ct. 1598 , 26 L.Ed.2d 142 (1970):

Adickes, 398 U.S. at 152 , 90 S.Ct. at 1605-06 (citations omitted) (emphasis added).

E. Devitt, C. Blackmar & M. Wolff, Federal Jury Practice and Instructions § 103.23 (4th ed. 1987) (Devitt, Blackmar & Wolff) sets forth the test for determining whether certain evidence proves the existence of a conspiracy under these circumstances. This is a test carefully distilled from controlling case law. The pertinent jury instruction reads in relevant part as follows:

Webster’s Third New International Dictionary (1976) defines “concerted” as something that is “mutually contrived or planned: agreed on.”

The term “proximate cause” is nowhere to be found in this instruction, and the reason for its absence is simple: It doesn’t belong there. 1

The record in this case amply and without ambiguity supports the proposition that law enforcement in Arizona reacted directly to perceived pressure from Phelps Dodge, and they may have so reacted because this was a “company town.” Cause and effect are shown, but that is all. As I assess them, the so-called inferences of an “agreement to violate or to disregard the law” in this case are wholly insubstantial and fall into the category of surmise and conjecture. Notwithstanding the array of facts set forth by the majority, the evidence of such an “agreement” is merely colorable. The facts simply do not reveal “a unity of purpose or a common design and understanding, or a meeting of minds in an unlawful arrangement.” American Tobacco Co. v. U.S., 328 U.S. 781, 810 , 66 S.Ct. 1125, 1139 , 90 L.Ed. 1575 (1946) (describing the term “conspiracy” in an antitrust context). But, if one is looking for proximate came, one might surely believe, when examining the inferences in this case, that a conspiracy has been spotted.

II

The Non-Smoking Words

After two years of discovery, approximately 90 depositions, answers to numerous sets of interrogatories, and extensive argument on the issue, the district court concluded that the specific facts and circumstances adduced by the Steelworkers in response to Phelps Dodge’s motion for summary judgment failed to connect Phelps Dodge to a conspiracy. Not a shred of direct evidence of a conspiracy was found anywhere. It can only be concluded that the district court judge determined from the record as a whole that “a fair-minded jury could not return a verdict for the plaintiff on the evidence presented.” Anderson v. Liberty Lobby, 477 U.S. 242, 255 , 106 S.Ct. 2505, 2513 , 91 L.Ed.2d 202 (1986). Nevertheless, the majority finds a handful of very predictable words spoken by a party who viewed itself as the victim of continuing criminal conduct to have enough leverage to convert the relationship between Phelps Dodge and certain public officials into a conspiracy. I find this to be nothing more than a theory without proof.

Without passing on the constitutionality of his actions, the true import of the words “keep them off the streets” and the talk of high bail is illuminated when one reads Sheriff Clarence Dupnik’s explanation of the process by which high bonds were recommended for persons arrested. The following is but a small portion of his deposition that gives his view of the challenge with which he was confronted:

[ER].

The “off the street” language from Phelps Dodge to which the majority attaches so much importance loses its conspiratorial impact when read in the context of the Sheriff’s explanation of his actions and when viewed in the light of his affidavit filed as Exhibit 7 in support of the motion for summary judgment. In that affidavit, Sheriff Dupnik stated:

[ER 094].

Three words end up being the difference in this case. In the view of the majority, these words — “off the streets” — uttered in connection with a request for high bail become the ultimate basis for an inference of an intent unlawfully to punish the strikers and an “agreement to violate or to disregard the law.” The majority even adds an interpretation to these words — “all streets” — that with all respect is unsupported and strikes me as grasping at straws.

As to the other circumstances to which the majority points to substantiate its view that somewhere during all of this Phelps Dodge and some unidentified state actor consummated an invisible agreement amounting to a conspiracy, these too amount to little more than fodder for speculation. Yes, these circumstances existed, *1553 but where’s the agreement? Where’s the beef?

Each of these circumstances is admittedly capable of innocent interpretation, yet they are added up so that the whole assumes greater probative value than its aggregate parts. This appears to be rationalized with the frequently stated but mistaken belief that direct evidence of a conspiracy is “seldom available.” It is correct to say that such evidence is frequently not available, but it is inaccurate to say that it is “seldom available.” The problem with the “seldom available” analysis is that it begins to sound like a rule of necessity that forgives a litigant’s failure to prove his case; and as I point out in Section III, this is not an appropriate setting in which to invite such a relaxed approach to the proof of critical facts.

It is with good reason that we do not permit conjecture, surmise, or speculation to determine the results in lawsuits. Such adventures of the mind have a tendency to be unreliable. They may serve many valid purposes in everyday life, as does hearsay. But when it comes to the law and proof of an allegation, we require more, and correctly so. Conjecture in the context of proof of facts is outside the channels of careful, rational, discriminating thinking — which is the tool by which lawyers, judges and fact-finders should be guided. This is a lawsuit we are dealing with, not the interpretation of a painting. We should not be inclined to arrive at or adopt conclusions based on theory rather than demonstration. This is precisely what motions for summary judgment are designed to weed out: lawsuits based on theory without proof. Does all of this evidence add up to “state action” on the part of Phelps Dodge? I think not.

Ill

The Fallout

The crucial distinction between being part of a concatenation and being a member of a conspiracy is of great importance in this particular setting. We are dealing not with a group of drug peddlers, bank robbers, swindlers, or klansmen, but with a legal entity that viewed itself — rightly or wrongly — as the victim of continuing criminal conduct.

First, this setting itself makes it questionable at best to infer an agreement “to violate or to disregard the law” from anything other than reasonably clear evidence. Such an inference as to the required unlawful state of mind is usually not difficult where the object of the conspiracy is patently illegal, such as in the case of drug peddling, etc.; but when the object of the agreement has an equally plausible lawful objective — and indeed one that is laudable — great care should be taken before imputing a conspiratorial state of mind to the parties.

Second, Phelps Dodge’s approach to the proper authorities, as opposed to relying on self-help, should be encouraged. The majority makes this same point very clearly. But, by attaching special significance to one person’s articulated wish at the resulting meeting with law enforcement to “keep them off the streets,” and “set high bail,” the majority effectively pays little more than lip service to this principle by creating a serious problem for anyone similarly situated who feels aggrieved enough to go to the police for help. There is hardly a victim in America who has not at some point uttered similar words. If the police, after hearing such words, go out and violate a protester’s rights, is this sufficient without more to support a jury verdict of conspiracy between the “victim” and the police? This is precisely the kind of danger warned against by the Supreme Court in Matsushita Elec. Ind. Co. v. Zenith Radio, 475 U.S. 574 , 106 S.Ct. 1348 , 89 L.Ed.2d 538 (1986) when it said:

475 U.S. at 593-94 , 106 S.Ct. at 1360.

I fear we chill the laudable goal of encouraging people to work appropriately with law enforcement in any matter involving a disturbance or continuing problem. If demonstrators take over a school, a government building, a hospital, a medical clinic, a church, a power plant or whatever, and the people responsible for the facility make what is, after this opinion, the mistake of saying “we want these people out of our building and carted off to jail with high bail,” will this without more be sufficient to justify a jury verdict of conspiring to violate the protesters’ rights if the police stumble and make a bad arrest or if a magistrate sets an improper amount on a bond? If a sheriff succumbs to pressure from a citizens’ group against drunk driving and sets up a road block that is later ruled to be unconstitutional, are the citizens co-conspirators? What will a sheriff do in a meeting with a similarly aggrieved group when one of the group utters “get ’em off the streets” language? To the extent that we are obliged to give guidance so citizens can follow the law, we may be creating quicksand where there should be a firm footing.

IV

The Standards of Review

Finally, I am troubled by the standard of review applied by the majority. It seems that in failing to grant any deference whatsoever to the trial judge, we effectively undermine the Supreme Court’s holdings in Celotex, 477 U.S. 317 , 106 S.Ct. 2548 , 91 L.Ed.2d 265 (1986), Liberty Lobby and particularly Matsushita. In Celotex , one reason the Supreme Court accepted the case was because of the view of a dissenting panel judge that the majority decision “undermines the traditional authority of trial judges to grant summary judgment in mer-itless cases.” Id. at p. 322, 106 S.Ct. at 2552.

In Liberty Lobby , the Supreme Court reprinted a similar message from Schuylkill & Dauphin Improvement Co. v. Munson, 14 Wall. 442, 448 (1872):

And, in Matsushita, the Court gives every appearance of rejecting second-guessing by an appellate court of a trial court’s view of the facts.

I do not advocate abdicating our responsibility to insure a litigant is not improperly deprived of his right to a trial by jury, but a trial judge in cases such as this is in a far superior position to make judgments that deal with the facts, and especially inferences. This judge lived with this case for a considerable period of time. I note that the record now fills up four large boxes and weighs 87 pounds, and this does not cover most of the depositions that by stipulation remained with the parties. Under traditional principles of standard of review jurisprudence, it would appear that the concerns of judicial administration favor the perspective of a trial judge with respect to facts and inferences. Matters of law and mixed matters of fact and law (weighing heavily on the law side) are one thing, but *1555 defects of fact in the sense that a party has no evidentiary support for an essential element of the case are another — even though constitutional rights are at stake. I am not offended by a more discerning multi-facet-ed standard of review that would find room for a trial court’s perspective under these circumstances. It may be that the “clearly erroneous” standard would be appropriate as applied to such matters.

I think we engage in a practice that is the equivalent of what Judge Norris warned against in U.S. v. McConney, 728 F.2d 1195 (9th Cir.1984) (Norris, J., dissenting), when he said, “What the majority does is what an appellate court has no business doing: curing a defect in a ... case by gratuitously filling in gaps in the evidentiary record.” Id. at 1213 .

I would affirm.

. I note that the Manual of Model Jury Instructions for the Ninth Circuit (1985 Edition) does not contain a similar instruction. Instead, at page 243 it refers textually to Arnold and its requirement to show proximate or legal causation. Certainly there is an appropriate role in these lawsuits for causation. But that role first requires that a conspiracy be established. Then the issue is whether the damages complained of were caused by that conspiracy. See Devitt, Blackmar & Wolff, § 103.22 and § 103.25.