Study of 650,000 links reveals 76.7% of 'small web' content has vanished

The digital record of the early 2010s is disappearing faster than previously quantified. A new study by the team behind 0.mk, a Macedonian URL shortener, reveals that the vast majority of links created during the first era of the social web have succumbed to 'link rot.'

In August 2026, the 0.mk team crawled a restored database backup containing 657,607 links created between 2009 and 2014. Of the 655,178 crawlable historical links, 76.7% no longer returned a loading page. When the data was analyzed at the unique-URL level—comprising 492,620 distinct addresses—the survival rate was even lower, with only 21.3% of the pages successfully loading.

According to the 0.mk Blog, the failures were split between infrastructure collapse and content removal. Network-level failures, including DNS issues, timeouts, and TLS errors, accounted for 51.24% of the dead links. The remaining 25.44% of failures were attributed to HTTP errors, such as 4xx and 5xx codes, which typically indicate that a page has been moved or deleted.

The Death of the Small Web

The dataset serves as a unique digital time capsule, capturing both the general internet and the specific national web of Macedonia from 2009 to 2014. The findings highlight a stark divide in digital longevity: centralized giants like Google, Wikipedia, and YouTube showed significantly higher survival rates than the decentralized 'small web.' Personal blogs, independent forums, and local news sites were far more likely to vanish entirely.

This trend underscores a fundamental shift in the internet's architecture. While content hosted on massive platforms persists, the independent nodes that once defined the web's diversity are being erased. The result is a digital history that is increasingly curated and controlled by a handful of corporate entities, while the grassroots record of the early web disappears.

A Resurrection via AI

The study was made possible by the 2026 relaunch of 0.mk. The service had originally shut down in 2014, crippled by spam and the rising costs of hosting. The team noted that the project became sustainable again only through the use of AI tools, which now handle the bulk of development and moderation tasks that previously required prohibitive human overhead.

As the web continues to centralize, the 0.mk study provides a concrete warning about the fragility of digital archives. While AI may help restore the tools of the past, the actual content they pointed to is largely gone, leaving a void where the independent web once stood.