Exxon Corporation sued its insurers (the "Underwriters") in a Texas state court, seeking reimbursement for losses and liabilities arising from the tragic grounding of its oil tanker, the Exxon Valdez. Though Exxon was headquartered in New York when the Valdez ran aground, it relocated to Texas before filing suit. The several Underwriters, with their various headquarters around the globe, doubtless feared the home-court advantage that Texas would give Exxon. So, the Underwriters filed their own suit against Exxon in the United States District Court for the Southern District of New York (Loretta A. Preska, Judge ), seeking a declaration that they were not liable on the insurance policies.

Invoking the abstention doctrine of Colorado River Water Conservation District v. United States, 424 U.S. 800, 96 S.Ct. 1236, 47 L.Ed.2d 483 (1976), the district court dismissed the Underwriters' action. The district court concluded that abstention was appropriate because (1) a parallel action was already proceeding in Texas state court, (2) states have a strong interest in insurance regulation, and (3) the case was controlled predominantly by state law.

On appeal, the Underwriters argue that Colorado River abstention was improper. Because this marine insurance dispute presents a federal question of first impression, and because the dual proceedings will cause no serious adverse consequences, we agree with the Underwriters, and reverse the district court's dismissal of their complaint.

The Underwriters comprise a group of more than 100 underwriters and insurance companies, headquartered in London, New York, Scandinavia, Japan, and elsewhere. All of them subscribed to a series of Global Corporate Excess insurance agreements (collectively, the "Agreement") with Exxon for the period from November 1988 through October 1989.

The Agreement insures Exxon and some of its subsidiaries against certain risks incurred during their worldwide activities. Specifically: Section I of the Agreement covers all physical loss or damage to the insured's own property; Section IIIA covers marine liabilities, such as fines and penalties for pollution; Section IIIB covers nonmarine liabilities, such as damage claims by third parties. Both Sections I and IIIA broadly cover losses that are "the unforeseen result of an intentional act." The Agreement also requires arbitration of claims arising under Section IIIB.

One of the subsidiaries covered by the Agreement owned the ill-starred Exxon Valdez oil tanker. In March 1989, the Valdez ran aground in a rocky shoal 25 miles south of Valdez, Alaska, and spilled 10.8 million gallons of oil into Prince William Sound. The master of the Valdez was Captain Joseph Hazelwood. The Underwriters allege that Hazelwood was drunk when the Valdez grounded, and that Exxon knew he was an alcoholic.

Exxon incurred colossal liabilities from the oil spill. Besides the damaged tanker and irreclaimable oil, Exxon faced a host of federal environmental violation charges, as well as federal and state civil claims for clean-up costs and other damages. In addition, private individuals such as commercial fishermen sued Exxon to recover for damage caused by the oil. (As of this appeal, Exxon has resolved its disputes with most of these claimants.)

In April 1989, Exxon tried to recoup some of its massive losses. It sent the Underwriters notice of its loss, and sought reimbursement for the lost oil, clean-up costs, and various liabilities to third parties. Exxon's demand exceeded $2 billion. The Underwriters denied coverage, claiming that the disaster was caused by Exxon's own misconduct, viz., placing a known alcoholic at the helm of the Valdez.

When negotiations collapsed, Exxon sued the Underwriters in Texas state court. Exxon's complaint alleged that the Underwriters had breached the Agreement and the covenant of good faith and fair dealing, and had also violated Texas insurance law. The complaint conceded that one of its insurers, Qatar National Reinsurance Company, is an instrumentality of a foreign state, and, as such, is not subject to suit in a state court. Accordingly, the complaint "release[d] and waive[d] all claims [Exxon] may have against [Qatar National] arising out of or in any way related to the [Agreement]."

Disquieted with the Texas forum, the Underwriters twice tried to remove the action to a federal court in Texas. Both attempts failed. The Underwriters also filed the present lawsuit in the United States District Court for the Southern District of New York.

Qatar National joined as a co-plaintiff; but four of the Underwriters had become insolvent (the "insolvent insurers"), and they did not join. The Underwriters' complaint sought a declaratory judgment that they were not liable to Exxon. Further, citing the Federal Arbitration Act, 9 U.S.C. Sec. 1, et seq., the complaint alleged that Exxon's Section IIIB claims were subject to mandatory arbitration. Exxon agreed thereafter to arbitrate the Section IIIB claims, rendering the Underwriters' Federal Arbitration Act argument moot. Thus, only the Section I and Section IIIA claims remain in the lawsuit, both raising the same question: what insurance coverage is there for "the unforeseen result of an intentional act"?

Exxon moved to dismiss the Underwriters' complaint, claiming that the federal court lacked subject-matter jurisdiction, and, alternatively, that it should defer to the Texas state court proceeding. The district court rejected Exxon's challenge to federal jurisdiction, holding that the dispute sounded in admiralty. Nevertheless, the court concluded that exceptional circumstances existed to warrant dismissal under Colorado River. Accordingly, it granted Exxon's motion, and dismissed the complaint. See Youell v. Exxon Corp., 1994 WL 376068 (S.D.N.Y. July 19, 1994).

As counseled by Colorado River, and the subsequent case Moses H. Cone Memorial Hospital v. Mercury Construction Corp., 460 U.S. 1, 103 S.Ct. 927, 74 L.Ed.2d 765 (1983), the district court examined: whether either court had obtained jurisdiction over a res; the relative convenience of the fora; the order in which the actions were filed; the desirability of avoiding piecemeal litigation; the source of controlling law; and the state court's ability to protect the federal plaintiff's interests. Finding that the suit did not involve a res, and that the two fora were equally convenient, the district court eliminated these two factors. The court also noted that the Texas action was filed first, and had progressed slightly further than the federal action.

The district court then found that piecemeal litigation was particularly undesirable under the circumstances because it could lead to conflicting interpretations of the same insurance policy. The court also held that Texas law predominantly supplied the rules of decision for the dispute because states have a strong policy interest in regulating insurance and because no applicable admiralty rule existed. Finally, the district court determined that the Texas court was fully capable of protecting the Underwriters' interests.

Balancing the six factors, the district court concluded that abstention was proper under Colorado River and Moses H. Cone, and dismissed the complaint. The Underwriters now appeal. We reverse.

Abstention jurisprudence begins with the principle that federal courts have a "virtually unflagging obligation" to exercise jurisdiction over all cases properly before them. Colorado River, 424 U.S. at 817-18, 96 S.Ct. at 1246. Only under "exceptional circumstances" relating to federal-state comity may a court eschew this duty, and leave the parties to fend for themselves in state court. Id. at 813, 96 S.Ct. at 1244.

There are four "extraordinary and narrow exception[s]" to a federal court's duty to exercise jurisdiction. Id. A federal court may decline to hear a case: (1) that poses a federal constitutional issue that may be mooted or altered by a state court ruling on state law, see Railroad Comm'n of Texas v. Pullman Co., 312 U.S. 496, 61 S.Ct. 643, 85 L.Ed. 971 (1941); (2) involving complex questions of state law that bear on important state policy issues, see Burford v. Sun Oil Co., 319 U.S. 315, 63 S.Ct. 1098, 87 L.Ed. 1424 (1943); (3) that was filed to enjoin state court proceedings, see, e.g., Younger v. Harris, 401 U.S. 37, 91 S.Ct. 746, 27 L.Ed.2d 669 (1971) (attempt to enjoin state criminal proceedings); or (4) when duplicative state court proceedings are pending, and exceptional "reasons of wise judicial administration" counsel dismissal, Colorado River, 424 U.S. at 817-18, 96 S.Ct. at 1246.

The Pullman abstention doctrine prevents federal courts from interpreting the Constitution where the case can be decided on alternative grounds; Burford, and Younger extricate federal courts from situations where the assertion of jurisdiction would intrude into the state courts' proper domain. Duplicative proceedings do not necessarily implicate these weighty concerns, and, thus, only "exceptional" circumstances permit Colorado River abstention. Id. at 817-18, 96 S.Ct. at 1246-47.

Whether Colorado River abstention is justified turns on a balancing of six factors: (1) whether either the state or federal court has assumed jurisdiction over a res; (2) the relative inconvenience of the federal forum; (3) the desirability of avoiding piecemeal litigation; (4) the order in which the actions were filed; (5) whether state or federal law provides the rule of decision; and (6) whether the state action will protect the federal plaintiffs' rights. De Cisneros v. Younger, 871 F.2d 305, 307 (2d Cir.1989). See Moses H. Cone, 460 U.S. at 23, 26, 103 S.Ct. at 941, 942 (announcing factors five and six); Colorado River, 424 U.S. at 818, 96 S.Ct. at 1246 (announcing factors one through four). Not all factors are created equal. A court will accord greater weight to those factors that bear more significantly on the case, and should thumb the scale in favor of exercising jurisdiction. See Moses H. Cone, 460 U.S. at 16, 103 S.Ct. at 937.

We review the decision to abstain for abuse of discretion, but scrutinize the legal issues underlying that decision de novo. See Sheerbonnet, Ltd. v. American Express Bank Ltd., 17 F.3d 46, 48 (2d Cir.), cert. denied, --- U.S. ----, 115 S.Ct. 67, 130 L.Ed.2d 23 (1994).^[1]

Here, one of the Colorado River- Moses H. Cone factors heavily tilts against abstention, two disfavor it, and three weakly support it. We will review these factors in the order of their importance.

Of the six Colorado River- Moses H. Cone factors, the fifth, the source of the rule of decision, is the most prominent here because an unresolved question of federal admiralty law will profoundly affect whether the Underwriters are liable to Exxon.

It is indisputable that the presence of federal questions weighs heavily against abstention. See Moses H. Cone, 460 U.S. at 26, 103 S.Ct. at 942 ("[T]he presence of federal-law issues must always be a major consideration weighing against surrender."); Colorado River, 424 U.S. at 815 n. 21, 96 S.Ct. at 1245 n. 21 ("[T]he presence of a federal basis for jurisdiction may raise the level of justification needed for abstention."); De Cisneros, 871 F.2d at 308 ("When the applicable substantive law is federal, abstention is disfavored."). And, since the lawsuit here concerns a marine insurance policy, it falls within the embrace of federal admiralty law. See Wilburn Boat Co. v. Fireman's Fund Ins. Co., 348 U.S. 310, 313, 75 S.Ct. 368, 370, 99 L.Ed. 337 (1955).

That admiralty jurisdiction exists does not automatically assure that the applicable law will be federal, however. Unlike other areas of admiralty law, Congress has not chosen to legislate with regard to marine insurance. See id. at 313-14, 75 S.Ct. at 370-71. As a result of this lacuna, marine insurance policies are governed by state insurance regulations, unless the federal courts have fashioned an admiralty rule on point, or unless a need for such a federal rule exists. See id. at 314, 75 S.Ct. at 370; Pope & Talbot, Inc. v. Hawn, 346 U.S. 406, 409, 74 S.Ct. 202, 205, 98 L.Ed. 143 (1953) (where federal law provides a defense in an admiralty suit, it controls over state law); see also Kossick v. United Fruit Co., 365 U.S. 731, 742, 81 S.Ct. 886, 894, 6 L.Ed.2d 56 (1961) ("The application of state law in [Wilburn Boat ] was justified by the Court on the basis of a lack of any provision of maritime law governing the matter there presented."). We must therefore determine whether the federal courts have fashioned an applicable admiralty rule, and if not, whether one is necessary.

The plaintiffs suggest several rules that, they claim, provide defenses to marine insurance liability--among them, the rule that all-risk policies in marine insurance contracts only cover losses caused by fortuitous events (the "fortuity rule"). They argue that the fortuity rule is an admiralty rule, and that it exempts a marine insurer from having to cover losses caused by the insured's recklessness. Thus, the plaintiffs argue, if they can prove that Exxon recklessly allowed a known alcoholic to pilot the Valdez and that this recklessness caused the grounding, the fortuity rule relieves them of liability.

Exxon, for its part, recognizes the fortuity rule but disputes that the rule is federal, claiming instead that it arises under state law. Exxon also argues that, in any event, losses caused by recklessness are still fortuitous, and are therefore covered.

As an initial matter, we agree with plaintiffs that the fortuity rule is indeed a creature of federal law. Federal courts sitting in admiralty have been applying some variation of the fortuity rule in marine insurance cases for over a hundred years. See Orient Mut. Ins. Co. v. Adams, 123 U.S. 67, 8 S.Ct. 68, 31 L.Ed. 63 (1887). In so doing, these courts have predominantly looked to federal law as precedent. See, e.g., Morrison Grain Co. v. Utica Mut. Ins. Co., 632 F.2d 424, 431 (5th Cir.1980) (relying on Second, Fifth, and Ninth Circuit cases); Atlantic Lines Ltd. v. American Motorists Ins. Co., 547 F.2d 11, 12-13 (2d Cir.1976) (relying on various federal district and circuit court decisions); Western Assur. Co. of Toronto v. Shaw, 11 F.2d 495, 496 (3d Cir.) (relying on federal precedent), cert. denied, 273 U.S. 698, 47 S.Ct. 93, 71 L.Ed. 846 (1926). And, to the extent that admiralty courts have looked to other sources, they have relied on British as much as state case law authority. See, e.g., Zurich Ins. Co. v. Wheeler, 838 F.2d 338, 340-41 (9th Cir.1988) (looking to British case law); Goodman v. Fireman's Fund Ins. Co., 600 F.2d 1040, 1042 (4th Cir.1979) (relying on cases from Great Britain, North Carolina, and federal courts). That admiralty courts do not look exclusively to state law on the issue of fortuity indicates to us that the rule must have an independent existence in federal maritime law.

Less clear, however, is how the fortuity rule regards losses caused by the insured's own recklessness. The fortuity rule excludes from coverage losses that arise from inherent defects, wear and tear, or from the insured's wilful misconduct; losses that arise from acts of nature, or the insured's negligence, are covered. See Ingersoll Milling Mach. Co. v. M/V Bodena, 829 F.2d 293, 307 (2d Cir.1987); Grant Gilmore, Charles L. Black Jr., The Law of Admiralty Sec. 2-9, at 72 (2d ed. 1975) ("What is covered is not any loss that may happen on the sea, but fortuitous losses occurring through extraordinary action of the elements at sea, or any accident or mishap in navigation.") (emphasis omitted). While negligent losses are fortuitous, and wilful ones are not, admiralty courts have yet to assign losses caused by the insured's own reckless conduct to either camp.

The concept of recklessness does not lend itself to easy characterization as either negligent or wilful conduct. Recklessness implies that "the actor has intentionally done an act of an unreasonable character in disregard of a known or obvious risk that was so great as to make it highly probable that harm would follow, and which thus is usually accompanied by a conscious indifference to the consequences." See W. Page Keeton et al., Prosser and Keeton on the Law of Torts Sec. 34, at 213 (5th ed. 1984) (citations omitted). As such, recklessness falls somewhere "between intent to do harm, which ... includes proceeding with knowledge that the harm is substantially certain to occur, and the mere unreasonable risk of harm to another involved in ordinary negligence." Id. at 212 (citation omitted). In other words, recklessness lies in the mists between mere negligence and wilful misconduct.

The hallmark fortuity decision is Orient Mutual Insurance Company v. Adams, 123 U.S. 67, 8 S.Ct. 68, 31 L.Ed. 63 (1887). In Orient, the master, or captain, of a steamer failed to follow a "custom of the river," and, as a consequence, his vessel was carried over falls and dashed upon a pier. The master had obtained an insurance policy upon the vessel, and sued to recover upon it.

At trial, the district judge instructed the jury that the insurer of the vessel would not be liable for damages if the ship master "designedly or recklessly exposed" the vessel to danger. See Orient, 123 U.S. at 72, 8 S.Ct. at 70. The judge added, however, that "the [insurer] is not relieved from liability by showing that the loss was remotely ascribable to the negligence of the captain or the other officers or [employees]." Id. The jury found for the insured.

On appeal to the Supreme Court, the insurer argued that the district judge should have charged that losses caused by the master's "misconduct" were also outside the realm of fortuity. Orient rejected the insurer's argument, stating: "the court properly said, in effect, that the misconduct of the master, unless affected by fraud or design, would not defeat a recovery on the policy." Id. at 73, 8 S.Ct. at 71. Orient continued that "the insured, so long as he acts with fidelity, is answerable neither for his servants nor for himself." Id. (citation omitted). Thus, Orient concluded, the jury charge was proper.

We find that Orient is equivocal as to the precise issue of whether a marine insurer is liable under a policy for losses caused by an insured's recklessness. See 123 U.S. at 72-73, 8 S.Ct. at 70-71. On one hand, as the Underwriters point out, Orient affirmed the district court's jury charge that the insurer is not liable for damages that occur because the master acted "designedly or recklessly." Id. On the other, as Exxon notes, Orient also stated the insurer is liable for all losses except those caused by the master's "fraud or design." Id. We cannot reconcile these two statements, for, at least, in modern legal parlance, "recklessness" is different from both fraud and design.

Admiralty cases after Orient similarly shed little light on the issue. For example, Ingersoll stated that an insurer is liable for "all losses which are fortuitous no matter what caused the loss, including the insured's negligence." See 829 F.2d at 307. Ingersoll also defined fortuity to exclude any loss caused by the insured's intentional misconduct. See id. Nowhere does Ingersoll explain what happens when the loss stems from the insured's conscious disregard of a known risk, however. Other admiralty cases discussing fortuity have likewise not addressed the point. See, e.g., Zurich, 838 F.2d at 340-41 (loss is fortuitous if it is "due to the sea"); Morrison, 632 F.2d at 431 (fortuitous losses are brought about by chance); Goodman, 600 F.2d at 1042 (losses caused by negligence are fortuitous; losses caused by intentional misconduct are not); Atlantic, 547 F.2d at 12 (mysterious loss of goods is fortuitous); Western, 11 F.2d at 496 (only losses caused by "abnormal causes and extraordinary occurrences" are covered by marine insurance policy).

It is clear, then, that whatever court adjudicates Exxon's insurance dispute with the Underwriters must decide, on first impression, whether losses caused by the insured's recklessness are fortuitous. This being an issue of admiralty law, a federal court should not shy away from it. Indeed, if the mere presence of a federal law issue is "a major consideration weighing against" abstention, Moses H. Cone, 460 U.S. at 26, 103 S.Ct. at 942, a fortiori, a federal question of first impression must all but demand that the federal court hear the case.

Obviously, if the federal court refuses to abstain, the inevitable result will be piecemeal litigation. Just as obviously, that result cannot command that the federal court abstain. Considerations of sound judicial administration require a pragmatic evaluation of the relative costs and benefits of permitting parallel litigation to proceed apace. See Colorado River, 424 U.S. at 819, 96 S.Ct. at 1247.

Apart from the expense of mounting separate lawsuits in New York and Texas, the major problem in pursuing parallel litigation is the risk of inconsistent determinations. See De Cisneros, 871 F.2d at 308. That risk is particularly acute in insurance disputes, where the possibility of inconsistent determinations may leave the insured with insufficient coverage. See Lumbermens Mut. Cas. Co. v. The Connecticut Bank & Trust Co., 806 F.2d 411, 414-15 (2d Cir.1986). This problem is largely dissipated, however, when it is recalled that a judgment by either a federal or a state court will ordinarily be res judicata in the other. See Arizona v. San Carlos Apache Tribe, 463 U.S. 545, 567, 103 S.Ct. 3201, 3213, 77 L.Ed.2d 837 (1983); Milltex Indus. Corp. v. Jacquard Lace Co., 922 F.2d 164, 166 (2d Cir.1991).

True, because they are not parties to both suits, res judicata may not apply to the four insolvent insurers or to Qatar National. These loose ends, although conceptually discomfiting, are of little practical significance. Exxon has no right to recover from Qatar National under the Agreement because Exxon has released Qatar National from liability. And, if the Underwriters should lose to Exxon, and seek contribution from Qatar National, they will have to bring suit in a federal court anyway. See U.S. Const. Art. III, Sec. 2; 28 U.S.C. Secs. 1332(a)(4), 1603(a) (federal courts have exclusive jurisdiction to hear suits against agencies of foreign governments). Thus, as far as Qatar National is concerned, the exercise of federal jurisdiction will not occasion additional unnecessary suits.

The four insolvent insurers are defendants in the Texas action, but have not joined the Underwriters as plaintiffs in the Southern District. If the Texas court enters judgment before the federal court does, that judgment will conclusively determine the insolvent insurers' liability by virtue of res judicata.

If, on the other hand, the federal court enters judgment first, the insolvent insurers would not appear to be bound by res judicata because they are not parties to the federal action. In theory, then, no matter which way the federal court ruled, Exxon will have to continue to litigate in Texas against the insolvent insurers.

As Exxon points out, this chain of events could lead to inconsistent interpretations of the Agreement, and, as Exxon reads Lumbermens Mutual Casualty Company v. The Connecticut Bank & Trust Company, this supports abstention. See 806 F.2d at 414-15. We find Lumbermens distinguishable.

In Lumbermens, various insurance companies had issued liability insurance policies for Raymark Industries, an asbestos manufacturer, for different periods over a 42-year span. About 30,000 personal injury lawsuits were pending against Raymark, and the insurers disagreed as to whether, for coverage purposes, the injuries occurred on the date of exposure to asbestos or the date the injury manifested itself. Zurich Insurance, one of Raymark's primary insurers, sued Raymark and all other primary insurers in Illinois state court, for a declaration that asbestos injury occurs at the time of exposure, not the date the injury is discovered. All Raymark's excess insurers eventually joined the suit. See Lumbermens, 806 F.2d at 412-13.

A second-layer excess insurer, Lumbermens Mutual, had agreed to cover Raymark for liabilities in excess of Raymark's primary coverage (provided by Zurich) and first-layer excess coverage (provided by American Home Insurance). Lumbermens Mutual sued Raymark in federal court, raising solely the issue about the timing of injury. Zurich and American Home, however, were not parties to the federal suit. The district court stayed the federal proceeding in light of the state court action, and this Circuit affirmed, finding that Colorado River abstention was proper. See id. at 414-15.

Integral to Lumbermens 's holding was that inconsistent determinations by the two courts as to when asbestos injuries occur could leave Raymark without sufficient coverage. See 806 F.2d at 415. That is, if the federal court entered judgment first, and the state court followed with a contrary judgment, Raymark would be reimbursed by either Zurich and American Home or Lumbermens Mutual, but not all three. As a result, Raymark would be stranded without full coverage, even though it had contracted for full coverage. Lumbermens also noted that inconsistent judgments would render the "allocation of indemnity and defense costs and the handling and administration of over 30,000 asbestos-related claims against Raymark absolutely impossible." Id.

Compared to Lumbermens, the possibility of inconsistent adjudications in our case pales to insignificance. First, Raymark's entitlement to full coverage was not in question in Lumbermens; the issue there was which insurers were liable for the coverage. Here, by contrast, the dispute centers on whether, under the Agreement, Exxon may recoup its losses at all. Thus, we do not face a situation where inconsistent adjudications may deprive Exxon of its due coverage. Second, the four insurers who are not federal plaintiffs are insolvent, and probably judgment-proof. See, e.g., In re New York City Asbestos Litigation, 151 Misc.2d 1, 572 N.Y.S.2d 1006, 1010 (Sup.Ct.1991) (noting that a defendant who is insolvent may be unable to contribute to a judgment), modified on a different issue, 188 A.D.2d 214, 593 N.Y.S.2d 43 (1st Dep't 1993). Third, inconsistent judgments will not complicate the administration of liability claims against Exxon because these claims will have already been resolved. Finally, we note that the four insolvent insurers constitute a minuscule fraction of the Underwriters, and are likely responsible for a trifling percentage of the total coverage provided by the Agreement. Given these important differences, the risk of inconsistent judgments here is inconsequential.

Finally, and most significantly, we note that here, unlike Lumbermens, resolution of the dispute hinges upon a novel determination of federal law. See Lumbermens, 806 F.2d at 415 ("state law supplies the rule of decision on the merits"). On balance, the possible annoyance that piecemeal litigation may cause here hardly offsets the commanding presence of a federal question of first impression.

C. The Remaining Colorado River-Moses Cone Factors

Colorado River requires us to look at whether either the federal or state court has obtained jurisdiction over a res, and whether the federal forum is more or less convenient than the state. Colorado River, 424 U.S. at 818, 96 S.Ct. at 1246. Here, the district court determined that these factors had no bearing on the abstention analysis because no res was involved, and because New York and Texas were equally convenient fora.

While the district court's factual findings on these two points were sound, the court erred by concluding that they were totally neutral on the abstention question. When the dispute lacks a res and the federal forum is just as convenient as the state, these two factors favor retention of the case in federal court. See De Cisneros, 871 F.2d at 307; Bethlehem Contracting Co. v. Lehrer/McGovern, Inc., 800 F.2d 325, 327-28 (2d Cir.1986).

The order in which the actions were filed must also be considered. Colorado River, 424 U.S. at 818, 96 S.Ct. at 1246. This factor requires consideration of "how much progress has been made in the two actions." Moses H. Cone, 460 U.S. at 21-22, 103 S.Ct. at 940. While neither action has progressed very far, some discovery has occurred in the Texas action. Also, Exxon expects to schedule a trial date in Texas for as early as the end of 1995; it is unclear when a federal trial could be held. The Texas action being further along, this factor slightly favors abstention.

The final Colorado River- Moses H. Cone factor looks at the capability of the state forum to safeguard the federal plaintiffs' rights. Moses H. Cone, 460 U.S. at 26, 103 S.Ct. at 942. The Underwriters have offered no reason for us to doubt the Texas court's ability to protect their rights. This factor somewhat favors abstention as well.

In balancing the six Colorado River- Moses H. Cone factors, we find that the federal law issue is truly a brooding omnipresence. As discussed, a major determinant of the Underwriters' liability to Exxon will be whether maritime losses caused by an insured's recklessness are fortuitous. This is a novel issue of federal admiralty law, and a federal court should decide it. In addition, this dispute lacks a res, and the fora are equally convenient. While the other three factors tilt towards abstention, they do so only tepidly. Accordingly, we see no exceptional circumstances here, and hold that the district court abused its discretion by abstaining.

While we loathe wasting judicial resources, it would be worse to cede federal review of an issue of federal law merely because Exxon won the race to judgment in state court. See Erwin Chemerinsky, Federal Jurisdiction Sec. 14.2, at 668 (1989) ("[T]he Supreme Court rightly has refused to allow those who prefer to litigate in state court to have the power to prevent federal court review simply by filing a suit in state court."). On balance, we choose the lesser evil.

Unlike Pullman, Burford, and Younger abstention, which implicate lofty notions of constitutionalism, the decision to abstain under Colorado River is driven by the more mundane problems of judicial efficiency. Accordingly, in "the absence of weightier considerations of constitutional adjudication and state-federal relations," the federal action should generally continue. Colorado River, 424 U.S. at 818, 96 S.Ct. at 1246.

We reverse the district court's judgment of dismissal. REVERSED.