In this paper, we promote a model for peer-based intelligent tutoring that leverages the past learning experiences of students with a repository of learning objects. Consistent with McCalla’s ecological approach, we determine appropriate peers and appropriate learning objects to direct a new student's learning. In particular, we focus on allowing peers to provide annotations of learning objects. We revisit a procedure developed to select which annotations to present to students in order to improve their learning: one that combines a modeling of the reputation of the annotation (based on its approval or disapproval by previous students), the reputability of the annotator (based on the reputation of all annotations left by the student) and the similarity of the raters with the new student. Our focus is on developing effective validation of the procedure’s benefit, using an approach of simulated student learning. This is achieved by developing algorithms in greater detail and then making particular design decisions for the simulation in order to manage the reputability of the annotators and annotations in a way that enables the best learning objects to be employed for the tutoring. We are able to demonstrate the value of our proposed approach using distinct measures of rater similarity. We conclude with a comparison to related work and a view to future directions for the research. As a result, we present an approach for interpreting data from interactions with previous students in order to influence how to interact with current and future students, to enable effective learning.