There is the notion that in web design everything has to be perfect and work flawlessly every time for every user on every browser and device. But this isn't entirely true. Sometimes, it can be healthy to leave certain (non-critical) functionality broken, if we feel that we would otherwise do things wrong or if we only partially agree with the existing standards. This is a way to actually raise awareness for the issues as they occur and not propagate them early in the entire online ecosystem. If we truly cared for where web design is headed, we would just stop and refuse to do the things wrong. Of what use is it that we stay true to our users when we can't stay true to ourselves? How can we help them if we run against ourselves? Isn't the largest pain our own pain, the one we do to ourselves, by not following our own principles? Can we still look at ourselves in the mirror without feeling guilty for the code we wrote? (honestly)
No. There's no later. The now can't wait. Postponing would only make us more miserable later. Then we'll need to remove the nonsense we produced, not without the potential side effects.
It's better to leave things broken when according to our definition they no longer work and we can't do anything meaningful for that. If we choose to do so, we need to explain that we aren't trying to let anyone down, but to save our inner world, just to be able to keep the least standard that will allow us to serve anyone at all. If our intuition says that something is wrong or harmful in the long term, we need to express disagreement without becoming a real problem. Through an open dialogue with the people whose work we use daily to do our work, we can both find previously missed perspectives. Leaving things (temporarily) broken allows for both sides to improve their approaches in a way that will benefit everyone (possibly at scale). Maybe we shouldn't continue until we come to a solution that is future-friendly. Resisting the urge to follow expert-evangelists could help us in that. If more people left things broken as a sign of disagreement, we could possibly progress faster in the direction we want to go and not in the one we all silently agree with anyway. Maybe a "broken" platform can be created in which designers/developers willingly express what bothers them on the path to adopting an accepted solution and why they won't continue this way (internal reasons of denial). This can help to concentrate useful insight and decrease the possibility for shipping half-baked solutions to everyone. How long should we try to improve things? Until.