I don't think the London historical offset should be blamed for such errors, nor should New York's nor any other zone's historical, local mean time offsets. If the tzdb is being used in a buggy way, that's on the user - in this case, pytz. Virtually all systems use tzdb in some form but don't blindly take the earliest historical offsets in a common usage pattern.
Fun fact: for many years the ECMAScript spec states explicitly that JavaScript implementations (i.e., a browser's implementation of JS) should use the wrong time zone information! Wrong in the sense of using an offset for _today_ rather than the offset that was applicable at the time of _the Date object_. Maybe they've changed that in recent years, I can't remember, but here's more info: https://codeofmatt.com/javascript-date-type-is-horribly-brok...
Fun fact: for many years the ECMAScript spec states explicitly that JavaScript implementations (i.e., a browser's implementation of JS) should use the wrong time zone information! Wrong in the sense of using an offset for _today_ rather than the offset that was applicable at the time of _the Date object_. Maybe they've changed that in recent years, I can't remember, but here's more info: https://codeofmatt.com/javascript-date-type-is-horribly-brok...