We have applied Simon's system for classifying computing education publications to all three years of papers from ICER. We describe the process of assessing the inter-rater reliability of the system and fine-tuning it along the way. Our analysis of the ICER papers confirms that ICER is a research-intensive conference. It also indicates that the research is quite narrowly focused, with the majority of the papers set in the context of programming courses. In addition we find that ICER has a high proportion of papers involving more than one institution, and high proportions of papers on the themes of ability/aptitude and theories and models of teaching and learning.