Inferring a “Shoot and Be Shot At” Agreement from Seconds of Reciprocal Gunfire: Applying Gnango as Confined by Seed and ARU

3.1 Precedents Cited and Their Influence

(a) R v Gnango [2011] UKSC 59: liability arising from a “duel-like” agreement

The judgment treats Gnango as the starting point for cases where a bystander is harmed during reciprocal violence between combatants. The Court recalled that the Supreme Court majority upheld liability where the defendant and an adversary had a common purpose “to shoot at and be shot at”. The Court emphasised Lord Dyson’s distinction (as quoted) between:

• a combat analogous to a duel (requiring an agreement that encourages reciprocal fighting); and
• a “mere fight” where return fire is simply provoked rather than encouraged pursuant to agreement.

The Court also highlighted Gnango’s unresolved complexities about whether such liability is best analysed as principal liability or accessory liability, noting the causation discussion (including Lord Dyson’s reliance on R v Kennedy (No 2) [2008] AC 269 and reference to R v Latif [1996] 1 WLR 104).

(b) R v. Jogee [2016] UKSC 8 and [2016] UKPC 7; [2016] 1 Cr. App. R. 31: secondary liability after “wrong turn” joint enterprise

The Court situated Gnango historically: it was decided during the period when joint enterprise doctrine had “taken a wrong turn” before correction in Jogee. The Court used Jogee chiefly to underscore that the modern focus is on intentional assistance/encouragement (and the requisite intent), not mere foresight. This reinforced the need to identify, with precision, the agreement and the mental element connecting the appellant to the co-accused’s wounding of the bystander.

(c) The Court of Appeal’s post-Jogee synthesis: Morgan, Seed, and ARU

The Court treated these authorities as defining (and, in practice, limiting) the proper application of Gnango:

• R v Morgan [2021] EWCA Crim 895: stated (in the passage quoted) that the combined effect of Jogee and Gnango is that one who agrees to a “shoot out”, or intentionally assists/encourages shots during a gun battle, with intent that those in the line of fire should die or suffer really serious injury, is liable for resulting deaths (and, by parity of reasoning, serious injuries).
• R v Seed [2024] EWCA Crim 650: articulated a careful inference-based test—agreement to “shoot and be shot at” may be inferred where the participants shoot at each other each knowing it was a virtual certainty the others were armed and would open/return fire, and each intending to kill or cause really serious injury; the Court cautioned that “shoot out” is shorthand and the jury must analyse “reciprocity”.
• R v ARU [2024] EWCA Crim 1101: reaffirmed Seed, stressing (i) “virtual certainty” and (ii) reciprocity; and observed that reciprocity may be more readily inferred with exchange of gunfire than with knives, while warning that such cases remain rare.

In Durrell-Smith, these cases did not expand liability; rather, they provided the structured conditions under which a jury may infer the requisite agreement and mental element. The Court’s key move was to treat the trial judge’s directions as a faithful application of this “confined Gnango” framework.

(d) “Riley and Robinson”

The Court referred to “Riley and Robinson” (without citation in this judgment) as an example of “duel-like” facts, reinforcing that the paradigm is agreed reciprocal violence rather than unilateral aggression. The reference served to support the proposition (also drawn from ARU) that such cases are unusual and fact-sensitive.

3.2 Legal Reasoning

(a) The legal threshold: evidence capable of supporting the necessary inferences

The appeal was framed as an evidential sufficiency challenge: even accepting the legal framework, was there evidence capable of proving an agreement “to shoot and be shot at” (including reciprocity and “virtual certainty”), such that the appellant could be liable for the bystander’s wounding?

The Court’s answer turned on the role of the jury. It held that the appellant’s points—brevity, alleged ambush, and the possibility that one or both were initially unaware of the other’s weapon—were matters of evaluation for the jury, not grounds to withdraw the count. The CCTV itself was treated as capable of supporting the inferences required by Seed.

(b) Reciprocity and “virtual certainty” can arise in moments

A notable aspect of the reasoning is the Court’s acceptance that an agreement “formed in advance, but perhaps only in the moments before the shooting” can satisfy the requirement. This addressed the appellant’s core submission that “seconds” could not realistically allow an agreement to crystallise. The Court effectively treated the rapid, reciprocal presentation and discharge of firearms—at close range—as conduct capable of evidencing a mutual commitment to exchange gunfire.

(c) The bystander’s injury and the structure of liability

The bystander was hit by a bullet fired by the co-accused, after passing through the appellant. The Court did not treat this as legally disabling. Instead, it approached the case through the Gnango/Seed prism: if the jury were sure there was an agreed reciprocal gunfight with intent to kill or cause really serious injury (and the required reciprocal knowledge), the appellant could be liable as a party to the violent joint activity and/or as one who intentionally assisted or encouraged the co-accused’s firing.

Importantly, the Court did not uphold liability on a simple “exchange of fire = automatic Gnango” basis; it rested on (i) the correctness of the Seed-compliant directions and (ii) the sufficiency of evidence that the jury could accept to meet those conditions.

(d) Directions and adverse inference

The Court noted that the appellant did not give evidence and that the jury were directed they could draw an adverse inference. While not determinative, this mattered because the appellant’s case depended heavily on an innocent explanation (self-defence and lack of agreement) in circumstances where the CCTV depicted rapid reciprocal targeting and shooting.

(e) Procedural decisions: dismissal and no case to answer

The Court endorsed the trial judge’s approach to both procedural stages:

• Application to dismiss: properly refused because there was evidence on which the jury could convict, including facts capable of supporting the requisite inference-based agreement.
• No case to answer: properly refused because the evidence, taken at its highest, could support a conviction; the later evidence (including ballistics/CCTV analysis and the “gloves” point) reinforced that the issue remained one for the jury.

3.3 Impact

The practical significance of R v Durrell-Smith is less about creating a new doctrinal test and more about demonstrating how the Seed/ARU constraints operate in a fast-moving, non-“feud” fact pattern:

• Brevity is not a bar: an agreement to “shoot and be shot at” may be inferred even where the exchange lasts only seconds, provided the conduct supports reciprocity and “virtual certainty”.
• CCTV-driven inference: high-quality footage can itself provide a sufficient evidential platform to leave Gnango-type issues to the jury, reducing the need for express communications or prolonged prelude.
• Bystander liability remains viable but confined: the decision reaffirms that liability for third-party injury can arise from participation in a reciprocal armed confrontation, but only within the strict conditions articulated in Seed and ARU.

For prosecutors, the case illustrates the evidential building blocks likely to matter: reciprocal aiming, near-simultaneous firing, proximity, and contextual evidence supporting mutual awareness and readiness to use weapons. For defence teams, it underscores that challenges will often be won or lost on whether the evidence is capable of supporting the Seed inferences—not on the mere fact that the defendant did not fire the injuring shot.