[Nextgrid]

Absolutely not. A cryptic error message still provides value to the user even if they don't understand it. Not only can they search and ask about it (maybe someone else does know), but they can also figure out a workaround by observing the system and under which conditions the error arises. You don't necessarily need to know what "Error 0x12345" means, but if you notice that it only happens under certain conditions, you can now work around it by not running the system under said conditions.

A bullshit error message on the other hand not only makes searching useless but also prevents the user from exploring towards a solution since they can't even tell different errors apart if they have the same generic message.

[dTal]

They aren't mutually exclusive. The cutesy message also provides useful information (this is our fault, unintended behavior etc). Why not have both?

[mannykannot]

The thing is, cutesiness does not convey anything either useful or specific, and, as this discussion uncontrovertibly shows, may be taken in unintended ways, including as being patronizingly dismissive. This is hardly surprising, as the context in which the error occurs cannot be known to the message's author. One's reaction is immediate and visceral, even if, on reflection, it probably was not intended that way (though I have known one or two developers who always blamed anyone but themselves, and particularly users, for errors.)

[munk-a]

I feel like this is unfortunately a semantic difference only - "cutesy" means different things to different people. Cutesy can just mean "informative but written in a friendly manner" to some people while as you and I understand it to mean "Jovial but unhelpful" - I think the user you're replying to is interpreting it in the first manner. A good example of a "good cutesy" error message might be Chrome's No Internet screen: "It looks like you don't have any internet, here are some ways you might be able to resolve it... but also here's a little dinosaur jumping game to pass the time".