URL Rot in Software Engineering Scholarly Works

Context: Citations play an important role in scholarly publishing. They increasingly reference Web-based resources via Uniform Resource Locators (URLs). However, URLs are prone to decay over time, threatening the value and integrity of scientific works. Objective: We investigate the persistence of URLs in software engineering publications, analyzing their stability over time, by hosting purpose, protocol, venue, document placement, and other attributes. Method: We extract nearly 144 thousand URLs from over 50 thousand papers published in major software engineering venues indexed in DBLP, spanning the period 1971–2024. Using automated tools and manual verification, we assess URL accessibility, classify hosts, and conduct regression analyses on several potential correlates of URL stability. Results: Overall, 71% of URLs remain active, with an estimated half-life of 11 years—substantially higher than reported in prior studies. URLs in version control systems (89% active) and archival platforms (94% active) exhibit significantly higher persistence than other Web links (71% active). HTTPS URLs (82% active) are more stable than HTTP ones (60% active), possibly due to their relative recency. Variations in URL stability exist across venues, document sections, and domains; however, regression analysis shows no strong correlation between stability and factors such as venue prestige. Conclusion: URL stability in software engineering literature is better than that reported in earlier studies. However, substantial URL decay persists. Increased reliance on archival services can further mitigate this threat. Periodic large-scale studies remain necessary to monitor evolving patterns of URL persistence in scholarly publishing.