The Tor EoL enforcement policy reduced the excluded relays' consensus-weight share from 13.25% in the first 2019 exclusion to roughly 7% of advertised bandwidth in the 2023 exclusion, demonstrating that iterative enforcement combined with improved operator communication (email, mailing list, social media) steadily reduces version fragmentation in a volunteer relay network over a 4-year horizon.

Monte Carlo simulations across four EoL exclusion rounds (2021–2023) show that natural relay churn is the dominant driver of Tor client anonymity loss, with EoL exclusions giving AS and IXP adversaries an advantage equivalent to only ~4 days (AS adversary) or ~9 days (relay adversary) of the network running without exclusion. In the 2021 exclusion event, up to ~1% of additional clients were compromised; by 2023 this figure dropped to ~0.30%, attributed to improved operator update discipline and Tor Project communications.

§5.1–5.3, Figure 5–7 evaluation

Across four analyzed exclusion rounds, a minority of rejected relays consistently accounted for over 50% of the security contribution (measured by individual path-compromise protection against AS, IXP, and relay adversaries) provided by all excluded relays combined, enabling selective prioritization of relay update outreach.

§6, Abstract evaluation

Default OS package repositories (Debian, Ubuntu LTS, FreeBSD quarterly) routinely ship Tor versions that lag behind the Tor Project's minimum accepted version, causing relays installed from default repos to be excluded immediately or within months of deployment. Debian LTS typically ships a Tor version that becomes excluded toward the end of the OS lifetime; Ubuntu LTS does not generally ship an LTS Tor version at all.

§3.3.2, Figure 3 deployment

In the April 2024 Tor exclusion event, 440 of 7,765 relays were excluded; only 68 (15%) returned with the same fingerprint within one month, while 372 (85%) never returned. Of the 553 relays running to-be-excluded versions one month before the deadline, only 75 updated in time.

§3.3.4, Figure 4 evaluation