[lmm]

If it were possible to always deliver unpleasant truths kindly then of course we would prefer it. But in practice we have limited time and limited ability, so in reality the choice will often be between delivering an unpleasant truth bluntly and not delivering it at all.

Whenever you say something that a reasonable recipient would find upsetting, it means you've failed, in the same way that when you have to add an explanatory note to a translation, or a comment to your code, that's a failure. But it's still often the least bad way forward.

[frankish]

As someone that feels they fit the "Super-reasonable" criteria, I personally struggle with this aspect: it takes me a long and stressful amount of time to be able to think of how to respond to people in a "Leveler" manner. In my opinion, it's impossible to do this because there is no objective criteria for how to satisfy it!

One might argue to stop at "good enough," but whose the judge on"good enough"? Again, no objective criteria!

I am strongly awaiting LLMs to at least automate "good enough" for me in communication, so that my perfectionism doesn't have to be directly involved.