We thank the respondents for their thoughtful replies to our debate article (Price & Jaffe Reference Price and Jaffe2023). Our main objective was to start a dialogue on failure and, in that, we have happily succeeded. The comments and critiques highlight the need for more discussion and thinking if we are to place failure in the archaeological interpretive toolbox. That said, the range of definitions, analytical perspectives, and unanswered questions will, we hope, provide a bulwark against turning ‘failure’ into yet another archaeological buzzword.

Kitching and Witcher (Reference Kitching and Witcher2023) remind us that archaeologists fail. A lot. The changing interpretations of Hadrian's Wall provide an excellent example—one that reminds us that failure is a part of scientific progress. Indeed, provocative theories or allegedly definitive interpretations often succeed most in pushing the discipline forward exactly when they fail. Think, for example, of the ‘Clovis First’ consensus model for the peopling of the Americas and its spectacular failure at Monte Verde, Paige Ladson, White Sands and other sites (Braje et al. Reference Braje, Dillehay, Erlandson, Klein and Rick2017).

Definitions of social phenomena are important but difficult. The inability, unwillingness or, well, failure of social scientists to agree on basic terms of debate often leads to a situation that Zeder and Smith (Reference Zeder and Smith2009: 681)—in reference to discussions concerning the origins of agriculture—characterise as “talking past each other in a crowded room”. We therefore take seriously Swenson's (Reference Swenson2023) contention that ‘failure’ may not be an appropriate term for small-scale ‘mistakes’. We agree so long as such mistakes are isolated and inconsequential. As Joyce (Reference Joyce, Bille and Sorensen2016) shows, purchasing the wrong type of cat litter can lead to a nuclear disaster. Here, the issue is not so much the scale of the error, per se, but its social and historical impact.

Scale is, however, of critical importance. Kitching and Witcher draw attention to the many alterations to Hadrian's Wall, where ‘big-I’ intentions (e.g. the Wall's masterplan) ran up against ‘little-i’ intentions (the daily construction of the Wall). Similarly, Swenson reiterates that trial-and-error is a path to success in mastering of crafts or skills. Such issues necessarily complicate the analysis of failure. We hope that archaeologists will continue to struggle with these scalar problems. We are explicitly not advocating for archaeologists to invoke failure as a deployable and catchy explanans. What we hope for instead are more debates, onsite and during analysis, of whether particular patterns represent successes, failures, minor mistakes, or something else.

Swenson (Reference Swenson2023) and Van Oyen (Reference Van Oyen2023) both stress the need to consider context. In this, we are in complete agreement. Indeed, failure has meaning only within and through its social context. In no way do we eschew an emic perspective (how to do this archaeologically is another matter). What we critique is the assumption of success. Why is it that cultural relativist approaches in archaeology and anthropology have tended to be biased against failure?

This brings up Van Oyen's Gramscian observation: “What counts as failure is decided by whatever is the hegemonic view of society.” There is an important caveat: hegemony has limits in cold reality. Indeed, exposing those limits is among the main ways by which one wages a “war of position”, to use Gramsci's terminology, and overcome ideological blinkers (Gramsci Reference Gramsci, Hoare and Smith1971: 108–20). Thus, while what counts as failure is dictated by those in power, it is never entirely so. It is precisely when that failure to control the definition of success/failure becomes obvious that hegemony cracks.

Perhaps contentiously, we maintain that despite all the complexities there are objective, or at least intersubjectively reasonable, ways to assess failure in the past and the present. Both Swenson and the article by Kitching and Witcher raise flags of caution; the latter pointing out that identifying failure is a “moral judgement” and the former rightly counselling against “modernist values and sensibilities”. But we argue that assessing failure is no more a moral judgement or imposition of modernist values than the tacit assumption of success.

Finally, Van Oyen puts us on the right track by urging us to think of failure as a social phenomenon. Who fails, in what way, the ramifications and meanings of failure, and the politics of failure—these are the key questions. For instance, failure may not be immaterial to some, particularly in highly unequal societies (think of those who ‘fail up’ in the corporate world). Failure can also, paradoxically, reproduce power; not only can some people afford to fail but they can sometimes succeed through failure. Van Oyen cites the ‘fail fast and fail forward’ credo in Silicon Valley; we add Zuckerberg's now infamous ‘move fast and break things’ approach (see Taplin Reference Taplin2017). In this case, the relationship between failure and success is not so much learning from one's mistakes, but rather having sufficient capital and credit to accept risk, of which failure is a correlate, and being able to avoid paying for one's mistakes.

In short, we are pleased that this brief exchange has sparked interesting questions and we hope it will continue to inspire debate and discussion. If we have led some archaeologists to look at their trenches and be able to entertain the possibility that someone, in the past, messed up, then we have succeeded. If not, well…