I've encountered many people who are dogmatically against anything which can be considered "optimization" in the general English-language sense of the word, and they very often quote verbatim the (partial) quote "premature optimization is the root of all evil" as a justification for their stance, implying that they interpret whatever I'm talking about to be "premature optimization". However, these views are sometimes so ridiculously entrenched that they dismiss pretty much any kind of algorithmic or data-structure deviations from the purest "naive" implementation... or at least any deviations from the way they've done things before. How can one approach people like this in a way to make them "open their ears" again after they shut down from hearing about "performance" or "optimization"? How do I discuss a design/implementation topic which has an impact on performance without having people instantly think: "This guy wants to spend two weeks on ten lines of code?"
Now, the stance of whether "all optimization is premature and therefore evil" or not has already been covered here as well as in other corners of the Web, and it has already been discussed how to recognize when optimization is premature and therefore evil, but unfortunately there are still people in the real world who are not quite as open to challenges to their faith in Anti-Optimization.
A few times, I've tried supplying the complete quote from Donald Knuth in order to explain that "premature optimization is bad" ↛ "all optimization is bad":
We should forget about small efficiencies, say about 97% of the time: premature optimization is the root of all evil. Yet we should not pass up our opportunities in that critical 3%.
However, when supplying the entire quote, these people sometimes actually become more convinced that what I'm doing is Premature Optimization™ and dig in and refuse to listen. It's almost as if the word "optimization" scares them: On a couple of occasions, I was able to propose actual performance-improving code changes without them being vetoed by simply avoiding the use of the word "optimiz(e|ation)" (and "performance" as well -- that word is scary too) and instead using some expression like "alternative architecture" or "improved implementation". For this reason, it really seems like this truly is dogmatism and not them in fact evaluating what I say critically and then dismissing it as not necessary and/or too costly.