Reception study has a theory richer than its data. Jauss (1982) proposed that a work is understood against a horizon of expectations formed by earlier reading, and that literary history should be written as the history of such horizons. The proposal is difficult to carry out because horizons leave few traces. When scholars have gone to living readers, as Radway (1984) did in her study of women who read romances, the gain in understanding has been considerable, but such studies are laborious and small. Fan fiction offers reception evidence that is unsolicited. Nobody asked these readers a question. A story that gives a minor character an inner life, extends a narrative beyond its ending, or retells an episode from the antagonist's side is an interpretation carried out in practice, and it shows what its writer took the source to have left unsaid. Pugh (2005) distinguishes writers who want more of a source from those who want more from it, and the second group in particular is engaged in something that looks very like criticism.

Literary theory has long held that meaning is produced within communities of readers who share assumptions and procedures. In most settings those procedures have to be inferred. In fan communities they are written down. Stories are discussed in comment threads, disputed readings of a character are argued over with citation of the source, and terms of art distinguish what the source establishes from what a community has come to agree. Busse (2017) shows that fan stories are composed for, and in conversation with, a particular community, and cannot be fully understood apart from that conversation. For a scholar interested in how interpretive norms arise, are enforced and change, this is an unusually well-documented case. It is also a corrective: readers in these communities attend closely to consistency of character and to textual detail, which complicates any easy contrast between professional and lay reading.

The suspicion that writing about another author's characters is inherently minor does not survive acquaintance with literary history. Virgil continued Homer. Medieval romance is a web of continuations. Cervantes wrote the second part of Don Quixote partly in answer to an unauthorised sequel. Jean Rhys's Wide Sargasso Sea gives a history to the first Mrs Rochester of Jane Eyre, and it is taught as a classic. Genette (1997) devoted a treatise to the relations by which one text is grafted onto another, and Derecho (2006) has argued that fan fiction belongs to this long tradition of what she calls archontic literature, in which a source is treated as an archive to which new entries can always be added. To study Rhys and ignore the same operation performed by great numbers of contemporary writers is to let the mode of publication decide what counts as literature.

Genre history is normally reconstructed after the fact from surviving print. In fan archives conventions can be observed as they form. Tropes are named by their users, attached to stories as tags, combined, parodied and abandoned within a few years, and the record of each stage remains. Coppa (2017) presents fan fiction as a set of genres with their own forms and histories, and computational studies have begun to compare fan writing with its sources at scale, examining which characters receive attention and how readers respond (Milli and Bamman, 2016). Work on particular communities has shown how such writing lets young writers explore matters of sexuality and identity outside the limits of the published books (Tosenberger, 2008). For the theory of genre, which has always had to work with incomplete series, a corpus in which the series is nearly complete and the classifying terms are supplied by participants is a considerable resource.

Much fan fiction is clumsy, formulaic or unfinished. The same is true of most surviving writing from any period, and literary historians do not on that account ignore the minor verse of the eighteenth century or the three-volume novels of the circulating libraries. The objection confuses two questions. Whether a text rewards aesthetic attention is one; whether it is evidence about reading, genre or reception is another. Guillory (1993) argued that canons are made by institutions, above all by the school and its syllabus, and that judgements of value cannot be separated from that history. A department can continue to teach judgement while accepting that the material on which claims about readers rest need not itself be distinguished. Some fan fiction is, in any case, accomplished by any standard, and the judgement is better made after reading than before.

It may be said that fan fiction is already studied, in media studies and sociology, and that English has no need to duplicate the effort. But the questions differ. Fan studies has been concerned chiefly with communities, identities and media industries. Thomas (2011), surveying the field, observed that the stories themselves had received less analysis than the practices around them and called for closer attention to them as narratives. That is work for which literary scholars are trained: attention to form, to intertextual relation, to voice and to the history of genres. A division of labour in which other disciplines study the people and nobody studies the texts serves no one.

The most serious objection is ethical. Fan writers publish under pseudonyms for audiences of fellow enthusiasts, often about intimate subjects, and some are young. Material that is technically public may have been written with a reasonable expectation that it would not be quoted in a lecture or a journal. This is a real constraint, and it is one that internet researchers have thought about carefully. The guidelines of the Association of Internet Researchers recommend judging each case by the expectations of the people concerned and the potential for harm, not by a simple test of public availability (Franzke et al., 2020). In practice this means seeking permission before quoting where it is feasible, avoiding links that expose individuals, taking special care with material by or about minors, and reporting patterns in aggregate where particular stories are not needed. Historians of reading face a version of the same problem with private diaries and letters. The difficulty calls for good practice, not for abstention.

Finally, the archives are enormous, and works can be revised or deleted by their authors at any time. No scholar can read more than a sliver, and a citation may point to something that no longer exists. Both problems are familiar from other fields. Sampling has to be principled and declared, as in any study of a large corpus, and conclusions should be scoped to what was read. Instability is a condition of all born-digital sources, and the remedy is the same: record the date of access, keep a private copy for verification where the archive's terms and the author's wishes allow, and describe as well as cite.