[tomrod]

> If you call designers artists, remove objective KPIs, and give them free reign, you get one-button mice. Similarly, if you let engineers build something, you end up with git's cli.

Am I in the wrong to think both of these are plenty functional to the point I like them?

A one button mouse functions a lot like a touchbad when you factor in meta keys (at least to my memory -- I've only ever used one in conjunction with schooling years due to low-Apple footprint). And having learned git cli recently, it seems to make sense to me.

But I agree generally -- silo boundaries can lead to poor results due to ignoring second order effects and holistic user experience (your users aren't siloed).

[ethbr0]

Regarding the optimal number of buttons on mice, I'd respectfully observe that the average number of human fingers on a hand is 5.

And as for git cli, I have no problem with it (conceptually or practically), but I also accept that most people aren't us.

It's moderately difficult to build something that's perfect for yourself: it's much harder to build something that's optimal for everyone not-you.

[culebron21]

I used Mercurial CLI in 2010-2016, before finally having to completely switch to git. I almost never had to memorize or look hg commands up. In Git, I routinely need to google that same answer on SO for flags. I do understand that it's the real inner machinery, and back in 2009 I did read how git works, but still cli is hard to remember.

[twic]

In some ways, the Git CLI is easier: whatever you're trying to do, the command you need is always "checkout".