When a researcher in 2005 prepared a paper for the annual Symposium of the Pattern Recognition Association of South Africa (PRASA), the first thing they did was download the "Information for Authors" kit. This kit, typically a ZIP archive or a set of web pages, contained everything from LaTeX style files to exacting instructions on page margins and font sizes. Unlike today's automated submission portals that validate PDFs on upload, the mid-2000s author kit was a self-contained manual that placed the entire burden of compliance on the author's shoulders.
What Was Inside an Author Kit?
Although the exact contents varied between conferences, a typical Information for Authors kit from the 2004–2006 period contained the following components:
- Template files — LaTeX class files (
.cls,.sty) and Microsoft Word templates (.dot) that enforced the conference's page layout, often two-column with specific margins and font sizes. - Formatting guidelines — A PDF or HTML document detailing everything from the required font (usually Times New Roman 10 pt or 12 pt) to the maximum number of pages (commonly 6–8 pages for full papers, 4 pages for short papers).
- Copyright and consent forms — A blank form to be signed by all authors, transferring copyright to the conference organizers or the sponsoring society.
- Submission instructions — Whether to email the PDF to the program chair, upload via a simple HTML form, or mail a CD-ROM. Many conferences required both a digital and a printed copy.
- Anonymization rules — For double-blind review, the kit explained how to remove author names from the document properties and avoid self-citations that could reveal identity.
- Camera-ready checklist — A final list of checks before submitting the definitive version: embed all fonts, ensure no hyperlinks break, confirm page count.
These kits were often hand-crafted by the publication chair or a volunteer technical committee member. The resulting files were then uploaded to the conference website, usually a static HTML page maintained by a university department. For a concrete example of such a kit, see The PRASA 2004 Author Kit: A Window into Pre-Automation Academic Publishing, which preserves the original formatting instructions used for that symposium.
The Evolution of Author Kits (2002–2007)
In the early 2000s, author kits were often distributed as a single ZIP file containing a plain text README, a LaTeX template, and a Word template. By 2005, the kits had become more elaborate: they included example PDFs, detailed FAQ pages, and sometimes even a sample paper that demonstrated the exact formatting. The shift from paper proceedings to CD-ROM and later to online PDF archives drove much of this change. Conference organizers realized that a poorly formatted paper would stand out in the digital proceedings, so they tightened the specifications.
The PRASA symposia, for instance, evolved their author kits year by year. The PRASA 2002 Symposium Archive: A Snapshot of Pattern Recognition Before Deep Learning shows that the 2002 kit was a simple one-page HTML document with minimal instructions. By 2004, the kit had grown into a 10-page PDF with screenshots of how to set margins in Word. This increase in detail reflects the growing diversity of author backgrounds: no longer were all participants LaTeX users; many were coming from industry or applied fields and needed explicit guidance for Microsoft Word.
The Role of the Author Kit in Ensuring Consistent Proceedings
Before automated PDF validation tools became common, the author kit was the primary mechanism for enforcing uniform formatting. Organizers trusted that authors would follow the instructions, but in practice many papers arrived with incorrect margins, missing fonts, or wrong page sizes. The publication chair often had to manually fix these issues or send papers back for revision. This process was time-consuming and error-prone, which explains why some conferences began to require that authors submit both the source file (LaTeX or Word) and the compiled PDF — so that the organizers could regenerate the PDF if needed.
The author kit also served as a historical record of the conference's production values. By examining the kit, one can infer the technical constraints of the era: the page size was usually A4 or US Letter, color printing was rare and expensive, and figures had to be grayscale or low-resolution due to storage limits on CD-ROMs. These constraints shaped the visual style of mid-2000s pattern recognition papers, which often appear dense and monochrome compared to today's glossy, full-color publications.
Common Pitfalls and How Authors Navigated Them
Interviews with mid-2000s researchers reveal a set of recurring challenges when working with author kits:
- LaTeX compatibility — Many conference-provided class files were written for older distributions (e.g., LaTeX2e with obsolete packages). Authors had to manually patch
.styfiles to compile on their local TeX installations. - Font embedding — Even when the kit instructed authors to embed all fonts, many PDFs produced by Word 2003 or earlier versions did not embed Type 1 fonts correctly, leading to garbled text on the reviewer's machine.
- Page count violations — The strict page limit was a constant source of anxiety. Authors often resorted to reducing font size in figures or squeezing references into smaller type, sometimes violating the guidelines.
- Copyright form logistics — The signed copyright form had to be scanned and emailed or faxed. Lost forms were a frequent headache for conference secretariats.
These challenges were documented in the conference's own communication channels. For example, the Inside the PRASA 2003 Archive: Pattern Recognition in South Africa includes a note from the publication chair reminding authors to use the correct template and warning that papers not conforming would be excluded from the proceedings.
The Decline of the Traditional Author Kit
By 2008, centralized conference management systems like EDAS and EasyChair began to handle submission requirements programmatically. Authors uploaded PDFs directly into a web portal that checked page count, file size, and font embedding automatically. The author kit shrank to a single page of basic instructions, with templates hosted on the submission system itself. The era of the hand-crafted author kit — with its lovingly detailed README and bespoke LaTeX class — was coming to an end.
For the historian of pattern recognition technology, these kits offer a rich source of metadata about the community's publishing practices. They reveal the tools authors used (LaTeX vs. Word), the constraints of the medium (paper vs. CD-ROM), and the values of the field (rigor, consistency, reproducibility). The Table of Contents in HTML: How Mid-2000s Researchers Structured Digital Proceedings further illustrates how the output of these kits — the papers themselves — were assembled into the final proceedings.
One concrete detail that captures the spirit of the era: the PRASA 2004 author kit specified a 12-point Times New Roman font for the body text, single-spaced, in a two-column layout with 0.75-inch margins. Papers were limited to eight pages, including figures and references. Authors were instructed to submit their PDF by email to the program chair, with the subject line exactly "PRASA2004 Paper Submission". Any deviation meant the paper risked being returned unread. That level of exactitude — enforced by a single volunteer with a dial-up connection — defined the author experience of mid-2000s academic publishing.