I think we do need to have empathy in our dealings with people online, and in general it's in our own best interests to do so. Many open source projects' lifeblood are their communities, and other things being equal, you'll get more contributions if you're not a complete jerk.

The flip-side is that high quality maintainable code is the product of top-notch commits, and rejecting commits is sometimes necessary to keep the standard of quality high. A good maintainer shouldn't cave to pressure of accepting a flawed commit just to avoid hurting someone's feelings.

This article in fact had what looks like a prime example of that. The comment mentioning a PR might "break a limit" but "we'll cross that bridge when we get to it" was touted as an example of how to give guidance. I'd argue that code quality slipped right there as a direct result of social pressure to accept a subpar commit.

It's not easy by any measure, but I think it pays to be not only clever and kind, but also consistent and firm when it comes to reviewing people's work.