Critical International Relations Theory (CIRT) is in ‘crisis’. Some argue for a recovery of ‘the inspirational quality’ of Horkheimer and Adorno’s first-generation negative critique. Certainly the challenge of right-wing populism begs questions of CIRT’s ‘consolatory’ cosmopolitanism. I have two concerns however. First, these proposals underplay the reasons why first-generation theorising failed; secondly, CIRT risks throwing the second-generation Habermas–Linklater ‘baby’ out with the ‘bathwater’ at the moment it has particular value. I do two things. I look back to pre-Habermasian Critical Theory, but I set a future agenda based on the Pragmatism of John Dewey. This helps CIRT realise the emancipatory potential in IR’s recent ‘practice turn’, addressing concerns that CIRT is disengaged. It also brings balance to negative and positive critiques, offering a novel challenge to critical/problem-solving binaries in ways that speak to real-world challenges like climate change. Second, I look forward from Habermasian-inspired theory to the third-generation (and Pragmatist-inspired) ‘recognition theory’ of Honneth. This brings a critical edge to IR ontological security studies, further develops the praxeological branch of CIRT, and better informs the political left’s response to the alienating effects of the liberal international order and the rise of right-wing populism.

Voluntary firearm safety actions avoid Second Amendment scrutiny, but rely on individuals recognizing their own risks. This could be aided by a network of healthcare professionals that have received proper training and information about all available tools to help prevent firearm-related suicide attempts, and combining the trust of clinicians and firearm owners could represent an opportunity to inform and educate in a manner that will engage patients.

Let Fn be the free group on $n \geq 2$ generators. We show that for all $1 \leq m \leq 2n-3$ (respectively, for all $1 \leq m \leq 2n-4$), there exists a subgroup of ${\operatorname{Aut}(F_n)}$ (respectively, ${\operatorname{Out}(F_n)}$), which has finiteness of type Fm but not of type $FP_{m+1}(\mathbb{Q})$; hence, it is not m-coherent. In both cases, the new result is the upper bound $m= 2n-3$ (respectively, $m = 2n-4$), as it cannot be obtained by embedding direct products of free noncyclic groups, and certifies higher incoherence up to the virtual cohomological dimension and is therefore sharp. As a tool of the proof, we discuss the existence and nature of multiple inequivalent extensions of a suitable finite-index subgroup K4 of ${\operatorname{Aut}(F_2)}$ (isomorphic to the quotient of the pure braid group on four strands by its centre): the fibre of four of these extensions arise from the strand-forgetting maps on the braid groups, while a fifth is related with the Cardano–Ferrari epimorphism.