Glowacki's splendidly integrative and constructive analysis marks an important advance by assessing the evolution of peace and war as an entwined set of responses to competing dilemmas. His overall conclusion that long-term peaceful relationships among human groups must have depended on sophisticated systems of communication and norm enforcement is persuasive. It suggests three topics that could usefully be developed using his conceptual framework.

First, it is plausible that mechanisms for establishing peace between groups emerged earlier than Glowacki's proposal of around 100,000 ya. Glowacki argued that because archaeological evidence of wide-ranging trade networks is meager prior to ~100,000 ya, sustained peace was unlikely to have been enacted before then. In justification of that idea, he proposed that prior to that date, sociocognitive mechanisms would have been inadequate for maintaining peace. In contrast, I suggest such mechanisms would have been available by around 300,000 ya.

My claim is based on the assumption that a critical sociocognitive mechanism for making peace between groups would have been a sophisticated form of language, and that we can reconstruct language as reaching the necessary level prior to 300,000 years ago. The 300,000 ya date comes from the argument that by then, subelite males were conspiring to kill the alpha male. Evidence that they did so derives from domestication-like features in the earliest Homo sapiens, which mean that for the first time in our Homo lineage, there was then selection against the aggressive behavior that typically confers high fitness on tyrannical alpha males (such as those that characterize all Old World apes and monkeys, including bonobos). Since alpha males are defined by their being able to defeat subordinates in one-on-one interactions, the reduction in aggression-related anatomy is hard to explain unless it resulted from alliances of subelite males coordinating to kill the alpha (Wrangham, Reference Wrangham2019).

Nonhumans cannot conspire and therefore cannot plot to kill a resented alpha male (Wrangham, Reference Wrangham2021), whereas among humans the conspiratorial killing of excessively violent males is routine and was probably universal in societies without prisons (Boehm, Reference Boehm2012). Communally agreed executions undoubtedly depend on linguistic ability being sufficiently sophisticated to permit the development of careful plans, while minimizing the risk of being betrayed to the alpha. This suggests that by 300,000 ya, language (and the norms that it fosters) would have been capable of similar dynamics in the context of fostering peace between groups – at least if those groups spoke the same language. Therefore if peace was truly limited before 100,000 ya, as the archaeological evidence hints, the limits on its development should have come from something other than sociocognitive constraints. Here I assume that negotiations about intergroup peace are about as cognitively challenging as negotiations about intragroup conspiratorial killing. Evidence against that idea would undermine the hypothesis that intergroup peace was possible long before 100,000 ya.

Second, the occurrence of peaceful relationships between groups would presumably have been strongly structured by the distribution of different languages or dialects, to the extent that the ethnolinguistic social arrangements of nomadic hunter–gatherers provide a relevant model for H. sapiens (Singh & Glowacki, Reference Singh and Glowacki2021). Residential groups (bands) would typically have been members of a network of groups sharing a common language or dialect, and the entire network would have been neighbored by others using a different language or dialect. Andamanese societies offer a model of this system. On the one hand, “internal war” (i.e., war within a language network) alternated with peace, in a pattern that Glowacki described for small-scale societies in general. On the other, among the 11 Andamanese ethnolinguistic societies a state of “external war” devolved into peace so rarely that war was described by Kelly (Reference Kelly2000, pp. 118–119) as “unremitting… a condition of existence that defines the boundaries of the niches exploited by two populations… peace was unattainable in external war” (Wrangham & Glowacki, Reference Wrangham and Glowacki2012). Presumably peace would always have been harder to generate as language differences increased.

Theoretical advances on Glowacki's conclusions could therefore come from a more explicit consideration of how the distribution of language differences among residential groups would have evolved. In this respect a fascinating challenge for archaeology is to distinguish within-society trade from between-society trade. The fact that cultural systems tend to be uniform within but not between societies suggests that a history of the development of ethnolinguistic societies might eventually be discernible.

Third, peaceful relationships among residential groups would have had implications for promoting mutual benefits not only at the dyadic but also the polyadic level. Specifically, peace between two groups would often have allowed them to more effectively make war against a third. This reminds us that one benefit of peace is an alliance that increases both defensive and offensive power, and adds a further context to Glowacki's demonstration that peace and war have been intimately related to each other throughout their evolution.