Is machine translation translation?

The discussion startsAt a 2018 translation studies conference in Turku, Finland, two of the guest editors of this special issue, Félix and Mary, gave back-to-back talks.We did not know each other, nor were we aware that our research interests concerned the opposite ends of a spectrum of translation modalities: fully (and specifically) human translation and raw, unedited machine translation (MT).Our presentations did not exactly oppose each other but nevertheless caused enough friction to start a long conversation that would last several years and lead to wider interest, a fullday conference panel, and this special issue.One of the first points of discussion concerned Félix's definition of translation as "an efficient process by which a text's communication effectiveness is improved by transforming it into a new text in a different language" (do Carmo 2018, 323).Mary's focus on MT gisting, or "the process of knowingly consuming raw machine translation with the aim of understanding as much of it as needed for a specific purpose" (Nurminen 2021, 30) led to discussions and further questions about the definition: would MT gisting fulfil the requirement of improving a text's effectiveness and could it therefore be considered a full form of translation?Does a translation that is of lesser quality than human translation nevertheless still integrate enough elements of what defines translation to be justifiably called that?The conversation continued, touching on human communication, the fragile notion of meaning, the intentions of source message authors, the dimensions of human communication that are perhaps not covered by MT, and the possible assumption of those dimensions by the reader in MT gisting.We oscillated between feeling that our views were so far apart as to be incompatible and noticing that we are on the same page in many ways.