When Infrastructure Promises Automation But Delivers Manual Labor: The Hidden Cost of Broken Research Data Systems

Research data infrastructure promises seamless credit attribution: publish a dataset, receive a DOI, and watch citations automatically populate your ORCID profile. Reality tells a different story. This presentation exposes the systematic failures in the interconnected systems that researchers, librarians, and data managers depend on for tracking dataset impact. Over four months, I attempted to automate citation tracking for datasets in the National Neighborhood Data Archive (NaNDA) using APIs from CrossRef, OpenAlex, and Google Scholar. Every automated approach failed due to missing metadata, inconsistent DOI registration, and datasets treated as second-class scholarly objects. The final solution required ten days of manual searching across Google Scholar, Web of Science, Scopus, and publisher websites—precisely the labor these systems claim to eliminate. The infrastructure breaks at predictable points: DataCite records missing author information despite ORCID integration; ORCID profiles failing to import dataset contributions from linked repositories; repositories unable to surface citing articles to dataset creators; citation databases prioritizing journal articles over dataset citations. Each failure shifts invisible labor onto already-stretched data professionals who manually connect researchers to their scholarly impact while maintaining the fiction that systems "just work." This presentation documents these failures through a detailed case study of citation tracking for a federally-funded open data repository. I'll demonstrate where automation promises exceed technical reality, quantify the hidden labor cost in staff time and opportunity cost, and argue that genuinely transforming data services requires honest assessment of infrastructure limitations. Until we acknowledge that current systems require substantial manual intervention masquerading as edge cases, we cannot advocate effectively for the resources needed to sustain quality data services or push vendors toward meaningful improvements. Attendees will leave with practical strategies for citation tracking when automation fails, evidence to support resource requests, and a framework for evaluating infrastructure promises against technical capabilities.