Contribution: This paper quantifies the phenomenon of more versus better research output in computing research education and elaborates on how the organizational variable known as cognitive distance plays a fundamental role in mediating such more versus better research output relation. Background: To improve the current educational system, investigation and quantification is needed of the "silos." Cognitive distance--a measure of the differences in background, culture, and expertise between collaborators--may be a factor influencing the lack of quality and variety in research outputs. Addressing this is a key enabler for fruitful collaboration. Research Question: Does collaboration with similarly expert researchers yield better research? Methodology: A quantitative survey provides baseline data for cognitive distance while publication data allowed creation of a co-authorship network between 123 researchers in a European computing research department. The network was analyzed through quantitative and qualitative research methods. Findings: Increased expertise overlaps across sub-fields of computing is a strong predictor for further collaboration (quantity), but research impact (quality) decreases with larger overlaps. This reveals an educational silo effect in doctoral computing education and, consequently, a flaw in the connected research output. The lack of a single, agreed way to evaluate research impact across sub-fields further hinders cross-departmental collaboration among doctoral students. Conclusion: Three recommendations emerge for policy makers and educational leaders: 1) departments should be cross-functional and focused on societal interests; 2) communities of practice should be created at the level of doctoral education and upward; and 3) departments should hold matchmaking and speed-meeting events regularly within and across institutions.