Very few opensource packages actually used log4j directly. Libraries generally use some sort of shim like commons-logging or slf4j so they can log into whatever your application chooses as a long sink.

Merely declaring a transitive dependency didn't mean that the code was executing nor did it mean that user input was actually being logged through those libraries.

Log4j v2 (the vulnerable version) just wasn't that common at the time. Most everyone had moved on to Logback, and Log4j v1 (not vulnerable) was still more popular than v2.