OP, here. I don't completely disagree with you. As I said, there are always reasons to take them out, and almost always, those reasons are excellent.
A well-executed Easter egg says something good about the team and the project, though. A team that has the time, resources and capability to build something neat or cool or (pleasantly) surprising, is a team that is unstressed and competent.
To be clear, I'm not discussing something like adding a clown giggle to flight control systems. I'm talking about when there are two equivalent paths, neither objectively worse than the other, I prefer when the team follows the path that adds whimsy and fun to the project over when they decide boring is better.
A team that will at least consider adding it will inevitably be more efficient than a team that absolutely will not consider it because they fear extra work.
But over the very LONG TERM projects I see with Easter Eggs tend to have had developers who cared more about the project and were having fun, which leads to an overall higher code quality. (Your own experience may be different).
A perfect example from the article: the performance and memory impact on Windows startup due to loading the ASCII cats’ monospace font.
The best reason to not add Easter Eggs is simply: they add unnecessary (albeit tiny) risk to a project. Technical risk and schedule risk. They are not part of the software's requirements, and they are likely going to be undocumented. Every line of code we add increases risk. Part of our jobs as developers is ensuring every line we add contributes more value than risk. Easter Eggs don't do this. Sorry to be Mr. Serious Raincloud, but please leave Easter Eggs for personal, non business-critical software.