Default Settings: What the Metadata of HCI 2009’s Papers Betrays About the Field

Most readers meet a conference paper as a citation, then as a document. The archive meets it as a file. Pull the HCI 2009 proceedings from the shelf — or, in 2026, from the server — and the first thing you encounter is not an argument but a filename: smith_final_v4.pdf, Paper_47_revised.pdf, HCI2009_camera_ready_FINAL(2).pdf. These small strings are the beginning of a material record that runs alongside the published text, and they deserve the same scrutiny we give to a keynote transcript or a delegate list.

This is not a study of the research itself. It is a study of the conditions under which the research was made legible. The observations below are drawn from the files held by this archive and should be read as indicative rather than exhaustive: metadata is editable, inconsistent and sometimes simply absent. Even so, the patterns are persistent enough to say something about institutional habits, software choices and the unglamorous labour behind the field’s published record.

The Name on the File

Before a paper was a paper, it was a working document with a name. The filenames in the HCI 2009 corpus show a discipline negotiating its own conventions in the absence of a shared one. A minority follow a clean author–keyword pattern. A larger group carry the vocabulary of revision: draft, rev, final, FINAL, final2, last, approved. Occasionally a filename admits fatigue — thisone_final.pdf — and occasionally it admits geography, with departmental abbreviations and short institutional codes standing in for authorship.

What the filenames do not do is identify people reliably. Surnames recur; initials substitute for given names; a paper with four authors may be filed under the first, the senior, or the one who owned the laptop. Anyone reconstructing authorship from filenames alone would produce a distorted map. That distortion is itself useful evidence: it shows how unevenly the field distributed the administrative work of submission.

Timestamps and the Shape of a Deadline

The file properties are more candid than the filenames. Creation and modification dates cluster in the days preceding the camera-ready deadline, with a marked concentration in the late evening and early hours. A recurring pattern places edits between midnight and four in the morning, followed by a burst of activity the following afternoon. This is not surprising; deadlines shape labour. But the timestamps make the shape visible, and they invite a question the proceedings themselves cannot answer: how much of what we now read as considered prose was written against a clock?

More revealing still are the gaps. A modification date later than the creation date by several weeks suggests a paper revisited after acceptance; a modification date after the conference suggests a file reopened for a correction, a reprint or a departmental record. In a handful of cases the metadata records an edit made years afterwards, presumably when a copy was deposited in an institutional repository. The paper did not change; its file did. For an archive, that distinction matters.

Producer Fields and the Software of a Discipline

The producer and creator fields offer a rough census of the tools that made UK HCI work possible in 2009. Microsoft Word dominates, with version strings that place most documents on the 2007 release. A smaller but coherent group carries pdfTeX identifiers, indicating authors who wrote in LaTeX and generated the final file directly. OpenOffice appears in the margins, as does Acrobat Distiller, the latter often in files that passed through a professional or departmental workflow.

Font embeddings tell a related story. The common pattern — Times New Roman for body text, Arial for headings — points to the conference template doing its job. Computer Modern in the LaTeX files points to authors who adapted rather than adopted it. A recurring fault is the unembedded font, which produces substitution on machines that lack the original and, in a few cases, visible reflow in the archived copy. These are small failures, but they are the kind of small failure that a 2026 reader notices immediately and a 2009 reader may never have seen.

A Question of Paper Size

One of the quietest divisions in the corpus is physical rather than intellectual. A4 pages predominate, as one would expect from a UK-hosted conference with a largely British authorship. But a meaningful minority of files are set to US Letter, usually where a transatlantic collaboration is visible in the affiliation list, and a few documents are internally inconsistent, mixing page sizes within a single PDF. The effect on the printed proceedings was cosmetic. The effect on the record is more interesting: it shows a field whose technical conventions were still being negotiated at the level of the page box.

What the File Cannot Say

Metadata is generous about production and silent about thought. It can tell us that a paper was revised late, that a font was missing, that a file travelled from one machine to another. It cannot tell us why an argument changed, which reviewer prompted the change, or what was said in the corridor afterwards. Nor does it reliably preserve the supplementary material — the questionnaires, the pilot data, the demonstration videos — that many authors once held and few deposited.

The most serious gap, for 2026 readers, is accessibility. The overwhelming majority of the HCI 2009 files are untagged PDFs: reading order is not encoded, headings are not marked, images carry no alternative text, and scanned pages defeat text extraction entirely. A screen-reader user encounters the proceedings as a sequence of uncertain fragments. This is not a criticism of the 2009 authors, who worked to the standards of their time, but it is an obligation for the archive now. The file is the find; the remediation is the work that follows.

The papers remain the primary record, and nothing here displaces them. But a discipline that studies interfaces should be willing to read its own artefacts closely. The filename, the timestamp and the producer string are not clutter around the scholarship. They are the sediment in which it settled, and they tell us, plainly, how the work was actually done.