Because, in general, the dev and the code reviewer(s) can’t accurately predict the impact of the changes. It’s a consequence of having a massive and complex code base. Now, best practices say that they should have automation in place that runs before the pull request can be approved, that tests against regression and unintended consequences. But, far too often, these things are deprioritised by management because of the “ship it now, we’ll fix it later” mindset.
To put it more succinctly, infinite profit growth is the reason everything is shit.
https://wego.here.com/ is decent