KATZMANN, Chief Judge:
This case calls upon us to determine whether federal admiralty jurisdiction extends to tort claims arising from a physical altercation among recreational visitors on and around a permanent dock surrounded by navigable water. We hold that federal admiralty jurisdiction does not reach the claims at issue here, because this type of incident does not have a potentially disruptive effect on maritime commerce.
Petitioners-Appellants Sapna Tandon and Robert Doohan, III, are the owners of the Up and Over, a thirty-nine-foot fiberglass powerboat designed for recreational purposes. On May 28, 2010, visitors on the Up and Over were involved in a fistfight on a floating dock operated by Claimant-Appellee Captain’s Cove Marina of Bridgeport, Inc. (“Captain’s Cove”). At least one person was seriously injured in the fight. Tandon and Doohan subsequently filed a petition for limitation of liability1 in the United States District Court for the District of Connecticut (Hall, [241]*241/.), seeking to limit their tort liability for the incident. The district court dismissed their petition for lack of subject matter jurisdiction, holding that this case falls outside the general grant of admiralty jurisdiction in 28 U.S.C. § 1333. We now affirm.
BACKGROUND
A. Factual Background
Captain’s Cove operates a marina in Bridgeport, Connecticut, on the waters of Black Rock Harbor and Cedar Creek, which open onto Long Island Sound. The marina facilities include a dockside restaurant, several docks extending from the dry land into the harbor, and a floating dock (the “South Dock”) accessible only by water. A water taxi runs from the South Dock to the restaurant and other facilities.
On May 28, 2010, Tandon and Doohan took several passengers2 on the Up and Over on a social trip to Captain’s Cove. They docked the Up and Over by the marina restaurant, and proceeded inside for food and drinks. Claimant-Appellee Ryan Ulbrick, who had also been invited along, arrived at Captain’s Cove by car and met the others there.
At about the same time, Third-Party Defendant-Appellee Frank Genna and two companions also made a social visit to Captain’s Cove. They arrived in a boat owned by one of Genna’s companions, moored at the South Dock, and then took a water taxi to the marina restaurant. Genna and his companions were not previously acquainted with Tandon, Doohan, or their companions.
Both parties left the restaurant at about the same time. As Tandon, Doohan, and their passengers were boarding the Up and Over; one of those passengers fell into the water and injured himself. Genna and his companions laughed at the mishap, leading the passengers on the Up and Over to yell unspecified but presumably unfriendly comments in response. Genna and his companions then boarded the water taxi to return to the South Dock, and both the Up and Over and the water taxi left the main docks.
At that point, the parties’ accounts diverge somewhat. According to an affidavit filed by Ulbrick, the water taxi headed slightly northeast, toward the north end of the South Dock, while the Up and Over headed southwest down the channel toward Long Island Sound. As the Up and Over was making its way down the channel, Tandon noticed that the passenger who fell while boarding the Up and Over was bleeding from a scalp wound. She therefore asked Doohan, who was piloting [242]*242the boat, to pull over and moor so that she could examine the passenger’s injuries. According to the state court complaint filed by Genna, on the other hand, the Up and Over followed in hot pursuit of the water taxi toward the South Dock. Meanwhile, its passengers yelled and screamed at Genna and his companions, and at one point threw a beer bottle at them.
The parties agree that both the Up and Over and the water taxi docked at the South Dock, where a fistfight broke out between Genna’s party and the passengers of the Up and Over. During the fight, one passenger from the Up and Over hit Gen-na, knocking him off of the South Dock into the water. According to Ulbrick, Genna landed face-down in the water and appeared unconscious; according to Gen-na, he was physically held underwater to the point of asphyxia. Genna claims that he suffered severe injuries from the lack of oxygen, including “cardiac arrest, respiratory failure, hypoxic encephalopathy resulting in permanent brain damage[,] and multi-organ failure.” J.A. 40.
B. Procedural Background
Genna and his wife Donna Genna (together, “the Gennas”) filed suit in Connecticut state court against Captain’s Cove and several persons affiliated with it (together, “the Captain’s Cove defendants”).3 They alleged that the Captain’s Cove defendants were liable for Genna’s injuries, and for Donna Genna’s resulting loss of consortium, under theories of negligent supervision, negligence, and reckless dispensing of liquor, and also under the Connecticut Dram Shop Act.4 The Captain’s Cove defendants responded by filing a third-party complaint against Tandon, Doohan, and their passengers on the Up and Over (including Ulbrick). In that third-party complaint, the Captain’s Cove defendants sought contribution and indemnity for any damages they might be required to pay the Gennas. The Gennas then filed a second amended complaint adding Tandon, Doohan, and the passengers on the Up and Over as third-party defendants, and asserting claims against them for negligence, recklessness, assault and battery, and conspiracy.5
Tandon and Doohan proceeded to file a petition for limitation of liability in the United States District Court for the District of Connecticut, initiating the present case. That petition asked the district court to either exonerate Tandon and Doo-han from liability for the incident at Captain’s Cove, or else limit their liability to the value of the Up and Over (appraised at $285,000). In accordance with the normal rules governing limitation proceedings, the district court stayed the pending state court proceedings and ordered that notice be sent to all persons asserting claims with [243]*243respect to the incident. The Gennas, the Captain’s Cove defendants, and Ulbrick all filed claims in the limitation proceeding; two other passengers on the Up and Over, Michael Hermann and Robert Barbieri, also filed notices of potential claims.
Ulbrick then moved to dismiss the petition under Federal Rule of Civil Procedure 12(b)(1), asserting that the district court lacked subject matter jurisdiction. The district court agreed. Applying the jurisdictional analysis laid out in Jerome B. Grubart, Inc. v. Great Lakes Dredge & Dock Co., 513 U.S. 527, 115 S.Ct. 1043, 130 L.Ed.2d 1024 (1995), the district court held that the alleged torts at issue in this case failed both the “location” test and the “connection” test for federal admiralty jurisdiction. It held the location test was not met because the fight in which Genna was injured took place primarily on the South Dock, and this floating dock was properly considered an extension of land because it remained permanently in a fixed location. It also held that the connection test was not met because the type of incident involved — which the district court characterized as “[a] fight on a dock” — did not have a potentially disruptive impact on maritime commerce. J.A. 130. The district court therefore dismissed the petition for lack of subject matter jurisdiction.
Tandon and Doohan then filed the present appeal. We have jurisdiction under 28 U.S.C. § 1291, and now affirm.
DISCUSSION
A. Standard of Review
‘When reviewing a district court’s determination of subject matter jurisdiction pursuant to [Rule] 12(b)(1), we review factual findings for clear error and legal conclusions de novo.” Close v. New York, 125 F.3d 31, 35 (2d Cir.1997). In resolving a motion to dismiss under Rule 12(b)(1), the district court must take all uncontroverted facts in the complaint (or petition) as true, and draw all reasonable inferences in favor of the party asserting jurisdiction. See Amidax Trading Grp. v. S.W.I.F.T. SCRL, 671 F.3d 140, 145 (2d Cir.2011) (per curiam). But “[w]here jurisdictional facts are placed in dispute, the court has the power and obligation to decide issues of fact by reference to evidence outside the pleadings, such as affidavits.” APWU v. Potter, 343 F.3d 619, 627 (2d Cir.2003) (quoting LeBlanc v. Cleveland, 198 F.3d 353, 356 (2d Cir.1999)). In that case, the party asserting subject matter jurisdiction “has the burden of proving by a preponderance of the evidence that it exists.” Makarova v. United States, 201 F.3d 110, 113 (2d Cir.2000).
B. Admiralty Jurisdiction
Under our Constitution, the federal judicial power extends “to all Cases of admiralty and maritime Jurisdiction.” U.S. Const, art. Ill, § 2, cl. 1. Congress has codified that jurisdiction at 28 U.S.C. § 1333(1), which gives federal district courts original jurisdiction over “[a]ny civil case of admiralty or maritime jurisdiction.” See Vasquez v. GMD Shipyard Corp., 582 F.3d 293, 298 (2d Cir.2009). “The primary purpose of federal admiralty jurisdiction is to protect commercial shipping with uniform rules of conduct.” MLC Fishing, Inc. v. Velez, 667 F.3d 140, 141-42 (2d Cir.2011) (per curiam) (quoting Vasquez, 582 F.3d at 298).
In this case, Tandon and Doohan invoked the district court’s admiralty jurisdiction by filing a petition for exoneration from or limitation of liability. That petition is a form of action peculiar to the admiralty and maritime context. It seeks the protection of the Limitation of Liability Act, first enacted by Congress in 1851 “to encourage ship-building and to induce [244]*244capitalists to invest money in this branch of industry.” Lewis v. Lewis & Clark Marine, Inc., 531 U.S. 438, 446, 121 S.Ct. 993, 148 L.Ed.2d 931 (2001) (quoting Norwich Co. v. Wright, 80 U.S. (13 Wall.) 104, 121, 20 L.Ed. 585 (1872)). Under the present version of this statute, “the liability of the owner of a vessel for any claim, debt, or liability [covered by the Act] shall not exceed the value of the vessel and pending freight.” 46 U.S.C. § 30505(a). It applies broadly, limiting the owner’s liability for any claims arising from “embezzlement, loss, or destruction of any property, goods, or merchandise shipped or put on board the vessel, any loss, damage, or injury by collision, or any act, matter, or thing, loss, damage, or forfeiture, done, occasioned, or incurred, without the privity or knowledge of the owner.” Id. § 30505(b). The Act thus protects the owner of a vessel from unlimited vicarious liability for damages caused by the negligence of his captain or crew. See In re City of N.Y., 522 F.3d 279, 283 (2d Cir.2008).
To take advantage of this statute, “[t]he owner of a vessel may bring a civil action in a district court of the United States for limitation of liability.” 46 U.S.C. § 30511(a). The owner may also seek total exoneration from liability in the same action (for instance, by asserting an affirmative defense that bars potential claims). See Supplemental Rules for Admiralty or Maritime Claims and Asset Forfeiture Actions (“Supp. R.”) Rule F(2). Once the owner files a petition for limitation, “all [other] claims and proceedings against the owner related to the matter in question shall cease.” 46 U.S.C. § 30511(c). The district court then “issue[s] a notice to all persons asserting claims with respect to which the [petition] seeks limitation,” instructing such claimants to file their claims in the limitation proceeding before a specified deadline. Supp. R. Rule F(4). Claimants may also file an answer challenging the petitioner’s right to exoneration from or limitation of liability. Id. Rule F(5). If the petition for limitation of liability is granted, the owner can be liable on the covered claims only up to the total value of his vessel and its pending freight; that amount will then be distributed pro rata among the proven claims. Id. Rule F(8).
Although the Limitation of Liability Act provides a federal cause of action for a vessel owner seeking exoneration or limitation, it “does not provide an independent foundation for federal admiralty jurisdiction.” MLC Fishing, 667 F.3d at 143. That is, the fact that a vessel owner may file a petition for limitation does not mean the district court necessarily has jurisdiction to hear it. Instead, the district court will only have admiralty jurisdiction to hear a petition for limitation if it already has admiralty jurisdiction over the underlying claims that the petition seeks to limit. See id. at 143-44. We therefore ask whether the underlying claims raise a “civil case of admiralty or maritime jurisdiction” that the district court could hear under 28 U.S.C. § 1333(1).
Here, the petition seeks to limit liability on underlying claims that sound in tort. We therefore turn to examine the scope of federal admiralty jurisdiction over maritime tort claims.
1. Legal Standard
“The traditional test for admiralty tort jurisdiction asked only whether the tort occurred on navigable waters. If it did, admiralty jurisdiction followed; if it did not, admiralty jurisdiction did not exist.” Jerome B. Grubart, Inc. v. Great Lakes Dredge & Dock Co., 513 U.S. 527, 531-32, 115 S.Ct. 1043, 130 L.Ed.2d 1024 (1995). The location of the tort normally depended on where the plaintiff was harmed — or to [245]*245use a more lawyerly phrase, where “the substance and consummation of the injury” took place. The Plymouth, 70 U.S. (3 Wall.) 20, 33, 18 L.Ed. 125 (1866). This could occasionally lead to odd results. For instance, the Supreme Court held on multiple occasions that when negligently piloted ships rammed structures on the land, the resulting claims were outside the law of admiralty, because the structures harmed were on the land and not in the water. See, e.g., Martin v. West, 222 U.S. 191, 195-97, 32 S.Ct. 42, 56 L.Ed. 159 (1911) (collision between a steamship and the pier of a drawbridge); Cleveland Terminal & Valley R.R. Co. v. Cleveland S.S. Co., 208 U.S. 316, 319-21, 28 S.Ct. 414, 52 L.Ed. 508 (1908) (collision involving multiple vessels and causing damage to a shore dock, bridge, protection piling, and pier); Johnson v. Chi. & Pac. Elevator Co., 119 U.S. 388, 389, 397, 7 S.Ct. 254, 30 L.Ed. 447 (1886) (collision between a schooner and a warehouse, in which the jibboom of the schooner went through the wall of the warehouse and caused a large quantity of shelled corn to run out and be lost in the Chicago River).
These ship-to-shore collision eases were superseded in 1948 by the Extension of Admiralty Jurisdiction Act, which extended admiralty jurisdiction to all “eases of injury or damage, to person or property, caused by a vessel on navigable waters, even though the injury or damage is done or consummated on land.” 46 U.S.C. § 30101(a); see Grubart, 513 U.S. at 532, 115 S.Ct. 1043. But outside of cases covered by that Act, the jurisdictional test continued to focus solely on whether the tort at issue occurred on navigable water. See, e.g., Victory Carriers, Inc. v. Law, 404 U.S. 202, 203-05, 92 S.Ct. 418, 30 L.Ed.2d 383 (1971) (reiterating the “historic view ... that the maritime tort jurisdiction of the federal courts is determined by the locality of the accident,” and finding no admiralty jurisdiction over an accident suffered by a longshoreman driving a forklift on a pier).
The Supreme Court first turned away from the traditional location-based rule in Executive Jet Aviation, Inc. v. City of Cleveland, 409 U.S. 249, 93 S.Ct. 493, 34 L.Ed.2d 454 (1972). Executive Jet involved an airplane that took off from a Cleveland, Ohio airport bound for Portland, Maine. Shortly after takeoff, the plane hit a flock of seagulls and crashed into the navigable waters of Lake Erie. Id. at 250, 93 S.Ct. 493. The owners of the airplane alleged that the air traffic controller and other airport employees had been negligent in failing to keep the runway free of seagulls or to adequately warn the plane’s pilot about the birds. Id. at 251, 93 S.Ct. 493.
The Court declined to decide whether the alleged tort was consummated when the plane hit the birds (over land) or when it hit the water. Id. at 266-67, 93 S.Ct. 493. Instead, it held that the location of the tort alone was not enough to give rise to admiralty jurisdiction. At least in the aviation context, the Court declared, admiralty jurisdiction also required that the underlying incident bear a “significant relationship” to “traditional maritime activity involving navigation and commerce on navigable waters.” Id. at 272, 93 S.Ct. 493. The Court held that a flight between two airports in the continental United States had no such significant relationship to traditional maritime activity, and therefore found no admiralty jurisdiction over the case. Id. at 272-74, 93 S.Ct. 493.
In Foremost Insurance Co. v. Richardson, 457 U.S. 668, 102 S.Ct. 2654, 73 L.Ed.2d 300 (1982), the Court clarified that the “significant relationship” or “connection” requirement announced in Executive Jet was a general rule of admiralty juris[246]*246diction, not limited to the aviation context. Id. at 674, 102 S.Ct. 2654. Foremost involved a collision between two noncommercial vessels — pleasure boats — in navigable water on the Amite River in Louisiana. Id. at 669, 102 S.Ct. 2654. The Court found that the connection test was satisfied, even though neither boat was engaged in commercial maritime activity, because this type of incident had a significant potential effect on maritime commerce: “For example, if these two boats collided at the mouth of the St. Lawrence Seaway, there would be a substantial effect on maritime commerce, without regard to whether either boat was actively, or had been previously, engaged in commercial activity.” Id. at 675, 102 S.Ct. 2654. The Court determined that this potential effect on maritime commerce, “when coupled with the traditional concern that admiralty law holds for navigation,” demonstrated the significant relationship necessary to support admiralty jurisdiction. Id.
The Court next considered the scope of admiralty tort jurisdiction in Sisson v. Ruby, 497 U.S. 358, 110 S.Ct. 2892, 111 L.Ed.2d 292 (1990). In that case, a vessel owner filed a petition for limitation of liability after a fire broke out on his pleasure yacht while it was docked at a marina on Lake Michigan. The fire destroyed the yacht and damaged several nearby vessels, but no commercial vessels were affected. Id. at 360, 363, 110 S.Ct. 2892. Nevertheless, the Court held that this incident had a sufficiently significant relationship to traditional maritime activity to sustain admiralty jurisdiction.
The Court applied a two-part test in Sisson to determine whether the case before it had a significant connection to maritime affairs. First, the Court looked to whether the underlying incident had a potentially disruptive effect on maritime commerce. It described the underlying incident as “a fire on a vessel docked at a marina on navigable waters,” id. at 363, 110 S.Ct. 2892, and held that such a fire has a potentially disruptive effect, because it “can spread to nearby commercial vessels or make the marina inaccessible to such vessels,” id. at 362, 110 S.Ct. 2892. The Court explained, moreover, that this potential effect does not depend on “the particular facts of the incident in this case,” but on “the general features of the type of incident involved.” Id. at 363, 110 S.Ct. 2892. The jurisdictional inquiry therefore turned on whether fires at marinas would generally disrupt maritime commerce, rather than on whether the particular fire at issue actually disrupted maritime commerce.
Second, the Sisson Court looked to whether there was a “substantial relationship between the activity giving rise to the incident and traditional maritime activity.” Id. at 364, 110 S.Ct. 2892. Again, the Court emphasized that the relevant “activity giving rise to the incident” is defined by the general type of conduct from which the incident arose, not by the specific facts of the case. “In Executive Jet, for example, the relevant activity was not a plane sinking in Lake Erie, but air travel generally.” Id. Likewise, in Sisson, the Court described the relevant activity as “the storage and maintenance of a vessel at a marina on navigable waters.” Id. at 365, 110 S.Ct. 2892. The Court found this conduct was substantially related to traditional maritime activity, id. at 367, 110 S.Ct. 2892, and so concluded that admiralty jurisdiction was appropriate.
Finally, the Court restated and formalized the current test for admiralty tort jurisdiction in Jerome B. Grubart, Inc. v. Great Lakes Dredge & Dock Co., 513 U.S. 527, 115 S.Ct. 1043, 130 L.Ed.2d 1024 (1995). That case involved a tort claim arising from construction work in the Chi[247]*247cago River. A construction company had used a crane sitting on a barge in the river to drive wooden pilings into the riverbed; it thereby (allegedly) cracked a freight tunnel running under the river, causing water to pour down into the tunnel and flood buildings in downtown Chicago. The flood victims filed a number of tort actions in state court; in response, the construction company filed a petition for limitation of liability in federal district court, invoking the court’s admiralty jurisdiction. Id. at 529-31,115 S.Ct. 1043.
The Supreme Court held that this case fell within the scope of federal admiralty jurisdiction. It began by laying out its analytical framework:
[A] party seeking to invoke federal admiralty jurisdiction pursuant to 28 U.S.C. § 1333(1) over a tort claim must satisfy conditions both of location and of connection with maritime activity. A court applying the location test must determine whether the tort occurred on navigable water or whether injury suffered on land was caused by a vessel on navigable water. The connection test raises two issues. A court, first, must assess the general features of the type of incident involved to determine whether the incident has a potentially disruptive impact on maritime commerce. Second, a court must determine whether the general character of the activity giving rise to the incident shows a substantial relationship to traditional maritime activity.
Id. at 534, 115 S.Ct. 1043 (internal quotation marks and citations omitted).
The Court then proceeded to apply that analysis to the facts before it. It held that the location test was met because the alleged injury, though occurring on land, was proximately caused by a vessel on navigable water; the location of the tort was therefore within the bounds of admiralty as defined by the Extension of Admiralty Jurisdiction Act. Id. at 534-37, 115 S.Ct. 1043; see 46 U.S.C. § 30101(a). It also held that both parts of the connection test were met: i.e., that this type of incident had a potentially disruptive impact on maritime commerce, and that the general character of the activity giving rise to the incident showed a substantial relationship to maritime commerce. Grubart, 513 U.S. at 538-43, 115 S.Ct. 1043.
In assessing whether this type of incident had a potentially disruptive effect on maritime commerce, the Court explained that the type of incident should be described “at an intermediate level of possible generality,” neither too general to distinguish different cases nor too specific to the unique facts of the particular case. Id. at 538-39, 115 S.Ct. 1043. It thus characterized the type of incident at issue in Grubart as “damage by a vessel in navigable water to an underwater structure.” Id. at 539, 115 S.Ct. 1043. The Court found this type of incident could easily have a disruptive effect on maritime commerce, by disrupting the course of the waterway itself or by restricting the use of the waterway during necessary repairs. Id.
In applying the second part of the connection test, the Court recognized that there was “inevitably some play in the joints in selecting the right level of generality” in describing the general character of the activity giving rise to the incident. Id. at 542, 115 S.Ct. 1043. But that “inevitable imprecision,” it warned, “is not an excuse for whimsy.” Id. The Court therefore rejected the idea that the relevant activity could be described as simply “repair and maintenance” or “pile driving,” since those descriptions would artificially eliminate the maritime character of the construction company’s activity. Id. at 541, 115 S.Ct. 1043.
[248]*248Instead, the Court described the activity giving rise to the incident as “repair or maintenance work on a navigable waterway performed from a vessel.” Id. at 540, 115 S.Ct. 1043. It found that this activity did bear a substantial relationship to traditional maritime activity, since barges and similar vessels have traditionally been used to engage in similar repair work. Id. (citing cases involving repair work carried out from barges). The Court clarified that although other non-maritime activities might have contributed to the ultimate injury — for instance, the city of Chicago’s alleged failure to maintain and operate the tunnel system — the substantial relationship test requires only that “one of the arguably proximate causes of the incident originate[] in the maritime activity of a tortfeasor.” Id. at 541, 115 S.Ct. 1043. Consequently, as the location test and both parts of the connection test were met, the Court held that there was federal admiralty jurisdiction over the petition for limitation of liability arising from the incident.
The test established in Grubart remains the current test for admiralty jurisdiction over claims sounding in tort. See MLC Fishing, 667 F.3d at 142; Vasquez, 582 F.3d at 298. To restate: First, we ask whether the alleged tort meets the location test: that is, whether it occurred on navigable water or was caused by a vessel on navigable water. Second, we ask whether the alleged tort meets both subparts of the connection test: that is, whether the general type of incident involved has a potentially disruptive effect on maritime commerce, and whether the general character of the activity giving rise to the incident bears a substantial relationship to traditional maritime activity. See Grubart, 513 U.S. at 534, 115 S.Ct. 1043. Only if the location test and both subparts of the connection test are met will admiralty tort jurisdiction be proper under 28 U.S.C. § 1333(1).
2. Analysis
We begin our analysis with the location test, evaluating whether the underlying torts at issue in this case occurred on land or on navigable water. But the facts here present one of those “perverse and casuistic borderline situations,” Executive Jet, 409 U.S. at 255, 93 S.Ct. 493, that have always bedeviled the traditional location test. The fistfight at issue took place on and around a floating dock surrounded by navigable water, and so one might think the tort occurred on navigable water. That dock, however, was connected by pilings to the harbor floor, making it “technically land, through a connection at the bottom of the sea.” The Blackheath, 195 U.S. 361, 367, 25 S.Ct. 46, 49 L.Ed. 236 (1904). It also remained permanently situated in its existing location, according to the district court’s factual findings, and so it cannot be described as a vessel for admiralty purposes. See Stewart v. Dutra Constr. Co., 543 U.S. 481, 493-94, 125 S.Ct. 1118, 160 L.Ed.2d 932 (2005) (holding that a watercraft that “has been permanently moored” is no longer a “vessel” for admiralty purposes).6 Moreover, the fight occurred both on the floating dock and in the surrounding water. According to the facts recited in the district court’s opinion, Gen-na was struck on the dock and knocked into the water. He was then held under[249]*249water to the point of asphyxiation. He consequently suffered injury both on the dock, where he was struck, and in the water, where he landed and where he was suffocated.
Like the Supreme Court in Executive Jet, we see no reason to resolve the difficult question of where the underlying tort (or torts) here occurred. Even assuming arguendo that the location test is met, admiralty jurisdiction cannot attach because the connection test is not met. In particular, we conclude that the first sub-part of the connection test is not met, as this type of incident does not have a potentially disruptive effect on maritime commerce.
a. Potential Effect on Maritime Commerce
In assessing the potential effect of this type of incident on maritime commerce, we begin by describing the incident “at an intermediate level of possible generality.” Grubart, 513 U.S. at 538, 115 S.Ct. 1043. Our description should be general enough to capture the possible effects of similar incidents on maritime commerce, but specific enough to exclude irrelevant cases. We then determine whether that type of incident is “likely to disrupt [maritime] commercial activity.” Id. (quoting Sisson, 497 U.S. at 363, 110 S.Ct. 2892). In so doing, we look not to the particular facts of the case before us — i.e., whether maritime commerce was actually disrupted here— but to whether similar occurrences are likely to be disruptive. The overall purpose of the exercise is to determine “whether the incident could be seen within a class of incidents that pose[ ] more than a fanciful risk to commercial shipping.” Id. at 539, 115 S.Ct. 1043.
We conclude that the incident at issue in this case is best described as a physical altercation among recreational visitors on and around a permanent dock surrounded by navigable water. This description accurately captures the nature of the event giving rise to this suit, and the type of risks that the incident could pose to maritime commerce. Like the descriptions the Court has used in its cases, our description focuses on the direct and immediate cause of the injuries suffered, rather than the alleged negligence underlying the suit. See Grubart, 513 U.S. at 538-39, 115 S.Ct. 1043 (considering “damage by a vessel in navigable water to an underwater structure”); Sisson, 497 U.S. at 362-63, 110 S.Ct. 2892 (considering “a fire on a vessel docked at a marina on navigable waters”). It also takes into account the general location of the incident and the roles of the persons involved, both of which can be relevant to the potential effect on maritime commerce. See Sisson, 497 U.S. at 363, 110 S.Ct. 2892 (including in its description the general location of the fire); Vasquez, 582 F.3d at 300 (describing the type of incident as “the death of persons repairing and refitting a vessel ” (emphasis added)).
We conclude that this type of incident does not realistically pose a threat to maritime commerce. First, a fistfight on and around a dock cannot immediately disrupt navigation. Unlike a sinking plane (as in Executive Jet), a collision between vessels (as in Foremost), or a collision between a vessel and an underwater structure (as in Grubart), it does not create any obstruction to the free passage of commercial ships along navigable waterways. Nor can it lead to a disruption in the course of the waterway itself. See Grubart, 513 U.S. at 539, 115 S.Ct. 1043. Second, a fistfight on a dock cannot immediately damage nearby commercial vessels. The fire considered in Sisson threatened the safety of all other boats nearby; a fistfight threatens only its participants. As the district court correctly pointed out, “[a] fight is unlikely to [250]*250spread the entire length of a dock, as a fire would, and, therefore, there is little risk that a fight would make the marina inaccessible or impact other boats.” J.A. 130.
Third, the class of incidents we consider here includes only fights on permanent docks — that is, docks that are connected in a permanent fashion to the land underneath or beside navigable water, and that do not move relative to the shore (except perhaps by rising and falling with the tide). This type of incident does not pose the same risks to maritime commerce as a ■ fistfight occurring on a vessel on navigable water. A fight on a vessel may distract the crew from their duties, endangering the safety of the vessel and risking collision with others on the same waterway. If a fight injures someone on a vessel that is at sea, moreover, that vessel may be forced to divert from its course to obtain medical care for the injured person. By contrast, a fistfight on a permanent dock does not endanger the safety of the dock itself or risk a collision between that dock and nearby vessels. And it obviously cannot require the dock to move or change course.
Fourth, the class of incidents we consider here involves only physical altercations among recreational visitors, not persons engaged in maritime employment. This type of incident therefore cannot have a potential effect on maritime commerce by injuring those who are employed in maritime commerce. Cf. Vasquez, 582 F.3d at 300 (holding there is “little question” that “the death of persons repairing and refitting a vessel” can potentially disrupt maritime commerce); Gruver v. Lesman Fisheries Inc., 489 F.3d 978, 982-83 (9th Cir.2007) (finding a potential effect on maritime commerce from “an assault on a seaman by his former maritime employer aboard a vessel in navigable waters”); Coats v. Penrod Drilling Corp., 61 F.3d 1113, 1119 (5th Cir.1995) (“Without a doubt, worker injuries, particularly to those involved in repair and maintenance, can have a disruptive impact on maritime commerce by stalling or delaying the primary activity of the vessel.”).7
We therefore conclude that the type of incident involved here — a physical altercation among recreational visitors on and around a permanent dock surrounded by navigable water — presents no realistic threat to maritime commerce.
b. Counterarguments
Tandon and Doohan raise several counterarguments, none of which we find persuasive. First, they claim that the “type of incident” involved should include the actions leading up to the fistfight — including their alleged negligence in piloting the Up and Over in hot pursuit of the water taxi. That argument confuses the first and second parts of the connection test. The first part of the connection test looks [251]*251to the nature of the incident that immediately caused the underlying injury; the second part, by contrast, looks to the nature of the broader activity giving rise to that incident. To take a few concrete examples: the type of incident at issue in Executive Jet was an airplane crash in navigable water, while the nature of the activity giving rise to that incident was air travel generally. See Sisson, 497 U.S. at 363-64, 110 S.Ct. 2892 (describing Executive Jet). The type of incident at issue in Foremost was a collision between two boats on navigable water, while the nature of the activity giving rise to that incident was the navigation of vessels generally. Id. (describing Foremost). The type of incident at issue in Sisson was a fire on a vessel docked at a marina on navigable waters, while the activity giving rise to that incident was the storage and maintenance of a vessel at a marina on navigable waters. Id. at 363-65, 110 S.Ct. 2892. In each case, the first part of the connection test looked more narrowly at the event that directly caused the injury at issue; the second part of the connection test looked more broadly at the proximate causes of that event. Cf. Grubart, 513 U.S. at 541, 115 S.Ct. 1043 (asking, in applying the second part of the connection test, “whether one of the arguably proximate causes of the incident originated in the maritime activity of a tortfeasor”). Here, the event that immediately caused Genna’s injuries was the physical altercation on and around the South Dock. It is that “incident” to which we look in determining the “type of incident” at issue and its potential effect on maritime commerce.
Tandon and Doohan next emphasize that the fight took place not only on the dock itself, but also in the water beside the dock. We agree that in considering the type of incident involved, the location of the incident may be relevant. And we agree that on the facts found by the district court, it appears some harm was done in the water as well as on the dock. That is why we characterize the type of incident at issue as a physical altercation on and around a permanent dock surrounded by navigable water.
But we are not convinced by the conclusion Tandon and Doohan seek to draw: that because the incident involved some harm suffered on navigable water, it necessarily had a potential effect on maritime commerce. Not all torts that happen on or over navigable water have the potential to disrupt commercial shipping. Otherwise, there would be no need for the potential effect test at all; we could simply apply the location test in its place. Cf. Foremost, 457 U.S. at 675 n. 5, 102 S.Ct. 2654 (noting that the substantial relationship test is necessary because “[n]ot every accident in navigable waters that might disrupt maritime commerce will support federal admiralty jurisdiction”).
Tandon and Doohan speculate that when a fight occurs partly in navigable water, the struggling bodies could themselves pose a navigational hazard. Because the class of incidents we are considering is limited to fights on and around permanent docks, however, we do not worry that the combatants might present an obstacle to commercial navigation in open sea lanes. Unlike the “collision between two pleasure boats in navigable waters” considered in Foremost, the type of incident described here could not disrupt commerce by occurring at the mouth of the St. Lawrence Seaway, because there are no permanent docks stationed in the middle of that major shipping route. (If there were, it would be the docks themselves rather than the altercation that would threaten commercial navigation.) We thus think that the scenario Tandon and Doohan pose presents only a fanciful risk to commercial shipping. [252]*252See Grubart, 513 U.S. at 539, 115 S.Ct. 1043.
At worst, an incident of this sort might temporarily prevent commercial vessels from mooring at the permanent dock around which the fight occurred. Cf. Sisson, 497 U.S. at 362, 110 S.Ct. 2892 (noting that a fire on a boat at a marina can “make the marina inaccessible to [commercial] vessels”). But the potential impact of such a temporary disruption is simply too meager to support jurisdiction. The fire considered in Sisson might have damaged a marina enough to close it for days or weeks, or even permanently; a fistfight presents no similar danger. At worst, it might prevent commercial ships from using part of a dock for a few hours. We do not think that this slight possibility of a temporary inconvenience is the “potentially disruptive impact on maritime commerce” envisioned by the Supreme Court’s test. Grubart, 513 U.S. at 538, 115 S.Ct. 1043.
Alternatively, Tandon and Doohan suggest that a fight on a dock surrounded by navigable water may require emergency responders to come to the dock by boat and leave by boat, potentially snarling naval traffic in nearby waters. We recognize that other courts have found the potentially disruptive impact of a maritime emergency response enough to satisfy the first part of the connection test in some cases. See, e.g., In re Mission Bay Jet Sports, LLC, 570 F.3d 1124, 1129 (9th Cir.2009); Ayers v. United States, 277 F.3d 821, 827-28 (6th Cir.2002); Sinclair v. Soniform, Inc., 935 F.2d 599, 602 (3d Cir.1991). Those cases, however, have generally dealt with incidents occurring either aboard a vessel or else in open water. See, e.g., Mission Bay, 570 F.3d at 1129 (considering “harm by a vessel in navigable waters to a passenger”); Ayers, 277 F.3d at 827 (considering “a drowning which occurred a short distance downstream from a lock on navigable waters”). Where such an incident takes place on a vessel or in open water far from the shore, the potential danger to commercial shipping posed by a maritime emergency response may be more significant. Cf. Roane v. Greenwich Swim Comm., 330 F.Supp.2d 306, 315 (S.D.N.Y.2004) (“[T]hose on board a boat ... giving their full attention to the saving of the life of a swimmer in difficulty may well be distracted from hazards posed by the approach of other boats unaware of the rescue in progress, or coming at speed in an effort to assist.”).
But the type of incident at issue in this case is a fight on and around a permanent dock, not a fight on a vessel or in open water. The risks to maritime commerce posed by a rescue operation at a dock are substantially lower than the risks to maritime commerce posed by a rescue operation at sea. Emergency responders may have to travel by boat to reach persons injured near a permanent dock, but they will never have to travel far. And once the emergency responders arrive at the scene, they can moor their vessel at the permanent dock, rather than having to focus simultaneously on navigating their vessel and rescuing the injured. An emergency response to an incident on and around a floating dock is consequently much less likely to “ensnarl maritime traffic,” Mission Bay, 570 F.3d at 1129, than an emergency response to an incident on a vessel or an incident in open water.8
[253]*253Indeed, accepting the argument that Tandon and Doohan advance would effectively eviscerate the first part of the connection test. If the possibility of a maritime emergency response is alone enough to show a potential impact on maritime commerce, then almost any tort occurring anywhere on or near navigable water would satisfy this requirement.9 We do not think this part of the connection test is so easily evaded.
Because the type of incident at issue in this case poses only “a fanciful risk to commercial shipping,” Grubart, 513 U.S. at 539, 115 S.Ct. 1043, it is outside the admiralty and maritime jurisdiction extended by 28 U.S.C. § 1333(1).10 The district court was therefore correct to dismiss the petition for lack of subject matter jurisdiction.
c. Coda
In developing the modern test for admiralty tort jurisdiction, the Supreme Court “aimed at keeping a ... class of odd cases out.” Grubart, 513 U.S. at 532, 115 S.Ct. 1043. It sought to realign the jurisdictional inquiry toward the primary purpose that supports admiralty jurisdiction — namely, “the federal interest in the protection of maritime commerce.” Sisson, 497 U.S. at 364 n. 2, 110 S.Ct. 2892. We are quite convinced that this case does not implicate that underlying purpose.
In the eloquent words of Justice Stewart:
The law of admiralty has evolved over many centuries, designed and molded to handle problems of vessels relegated to ply the waterways of the world, beyond whose shores they cannot go. That law deals with navigational rules — rules that govern the manner and direction those vessels may rightly move upon the waters .... Through long experience, the law of the sea knows how to determine whether a particular ship is seaworthy, and it knows the nature of maintenance and cure. It is concerned with maritime hens, the general average, captures and prizes, limitation of liability, cargo damage, and claims for salvage.
Executive Jet, 409 U.S. at 269-70, 93 S.Ct. 493. As that description shows, the scope of admiralty law has little or nothing to do with the issues that are likely to appear in cases like this one. On the contrary, state courts have long dealt with similar fistfights under state tort law, without any need for interference from federal courts sitting in admiralty. Given the traditional role of state courts in adjudicating such torts, “[w]e are not inclined at this juncture to disturb the existing precedents and to extend shoreward the reach of the maritime law further than Congress has approved.” Victory Carriers, Inc. v. Law, [254]*254404 U.S. 202, 211, 92 S.Ct. 418, 30 L.Ed.2d 383 (1971).11
CONCLUSION
For the reasons above, the judgment of the district court is AFFIRMED.