Rank: 1 –> Condon’s chapter supposes that “all the claims in favor of machine scoring” do “what human raters do,” and with those assumptions, Condon provides a handful of losses that inevitably accompany AES.
Argument: Citing  the loss of control over the construct “writing” to an outside agency, the loss of local assessment and the expert rater,  devaluation of lore and consensus about writing standards and expectations, and  the loss in local agency (students’ and teachers’ level of confidence in the text), Condon shows what is pedagogically, institutionally, and economically lost with AES.
- local assessments are better than national assessment
- improved instruction accompanies a culture of assessment at the institutional level
- teachers are (the most) reliable raters
- validity is more important than reliability
- portfolios necessarily promote conversation about learning
Points of interest:
- some institutions pair humans and machines to assess (215)
- schools can order customized tests (more local tests) from testing companies (215)
- argues that machines cannot read, they look “at physical features of a text that, separately, are associated with a certain level of performance (211-212)