Evolutionary approaches to culture have much to gain from studying children’s peer culture. Lew-Levy and Amir’s much-needed contribution does an outstanding job of unifying the wealth of anthropological data supporting the existence of an autonomous peer culture, chiefly transmitted from child to child. Once their existence is admitted, children’s peer cultures raise a series of questions that our field is yet to tackle – starting with the puzzle of their existence. How do children’s peer cultures manage to maintain themselves across time? In nearly all cultures, as Lew-Levy and Amir show, children form a sub-society that is not, on the face of it, well-suited for maintaining lasting traditions. Children’s peer societies reside in the midst of an adult culture that dominates their own in almost every way. Compared to adults, they only have limited time to acquire and then transmit their peer culture, because children do not stay children for very long – they have, in other words, unusually high turnover rates (Opie & Opie, Reference Opie and Opie1959, Morin, Reference Morin2016).

According to the “selection for proliferation” hypothesis (Morin, Reference Morin2016), the endurance of children’s peer culture is, somewhat paradoxically, due precisely to the high population turnover rate that should make it difficult to sustain. High turnover implies that children’s traditions must be constantly transmitted in order to survive. But, while this cultural transmission treadmill eliminates the greater share of the least appealing cultural practices, the « selection for proliferation » hypothesis claims that it allows some of them to survive longer. The traditions that survive frequent population turnover are (the hypothesis assumes) more likely than others to be frequently transmitted. High turnover selects those « proliferating » traditions, eliminates others, and thus makes it easier for proliferating traditions (lacking competition) to proliferate even more. The result is a limited repertoire of enduring traditions, detaching itself from a much larger cohort of ephemeral practices. The claim, in other words, is that high population turnover causes an increase in the variance of tradition’s longevity, causing short-lived traditions to disappear quickly but allowing long-lived traditions to endure more. For children’s peer culture, the most visible outcome is a set of traditions that can last as long as comparable adult traditions.

Assessing and explaining the longevity of children’s traditions is important if the field is to take children’s peer cultures seriously – to see them as repositories of genuine traditions on a par with adults’. Lew-Levy & Amir emphasise the contribution that children’s peer cultures can make to the cultural resilience of entire human societies, thanks to their capacity to serve as repositories for otherwise neglected knowledge. To back this up, they cite persuasive evidence for resilient and lasting juvenile traditions. Yet a question remains (one that I also address to myself): To what extent is the longevity of children’s peer culture really comparable to that of the most resilient adult traditions? Cultural longevity can be extreme, even in cultures that rely solely on oral transmission: claims of narratives being precisely transmitted for 7,000 years can be substantiated by compelling evidence (Nunn, Reference Nunn2020, Nunn & Reid, Reference Nunn and Reid2016).

Measuring and comparing the longevity of cultural traditions is difficult for two reasons, both having to do with granularity – with the precision we bring to recording and defining children’s traditions, and adults’.

First, to measure the lifespans of a set of traditions, we should ideally sample the historical record at regular intervals in time and detect every time whether the tradition is present or not – longevity being inferred from the length of time separating the two most extreme instances where a tradition was detected. With historical data, the times at which we can sample the record are never controlled by us, and they vary widely from one tradition to another. Typically, children’s traditions are less documented, and thus the rate at which we can sample the historical record is lower for them. This can be a problem for two reasons. First, more documented traditions are more likely to be recorded as present, biasing our longevity estimates in their favour. Second, cultural practices generally show greater rates of change when observed over shorter timescales (Perreault, Reference Perreault2012, Reference Perreault2020). Second, adult traditions are likely to be classified in a more granular way compared to equivalent practices common in children – again because adults pay more attention to their own culture and document it better. Data on Renaissance games in children and adults differentiates 15 games of card but only 4 games of tag (Morin, Reference Morin2016, appendix). This time, the bias would lead us to overestimate the longevity of children’s games: the same game can appear as one long-lasting game when described with low granularity or as three short-lived games under a finer grain of description.

Can such difficulties be remedied? Perhaps, because something unprecedented is happening to children’s peer cultures: older children (in rich countries) are now writing to one another without adult supervision, often publicly, and with considerable linguistic creativity. Online language data escapes many of the biases that affect anthropological or folklore material. Comparable adult and children’s innovations can be sampled at identical intervals and with the same grain of description. Time depth will necessarily be inferior to that of historical data, but online communication might have a decently long future, and in any case, the proliferation hypothesis makes a prediction that can be tested on short timeframes: there should be more short-lived linguistic innovations among children than among adults. Thus, I join Lew-Levy and Amir’s invitation to study children’s online worlds, equipped with cultural evolutionary hypotheses.