Hacker News new | past | comments | ask | show | jobs | submit login

You haven't actually said anything bad about it except "It's GNOME".

Could you explain what's bad about GNOME?




GTK (and EFL, which Elementary actually uses and was absolutely indistinguishable to me in casual use, but I know GNOME and derivatives better) feels bad and makes me sad to use it. Like...that's it. And I realize that's not a satisfying answer, and that it's super subjective, but that's pretty much it. Navigating around feels off, the way it nods to OS X but adds very obviously Windows stuff on top of it is kind of hard to deal with, it doesn't feel cohesive. It also looks pretty rough (too much unrelieved flat-panel gray) and text rendering is nowhere near as nice as OS X.

My Linux Mint desktop runs Cinnamon, but it's a lot less of a pain there because the only apps I run are Atom, Chrome, and Terminator (which is not a great terminal emulator but whatever, it's fine).


[Woops - apologies for wall of text. The HN reply box is very small, I didn't realize I typed this much.]

FWIW, I think what you're describing may fall under https://en.wikipedia.org/wiki/Uncanny_valley. It's a pet theory I've had for a while.

I, too, have thought that GTK has a sense of "un-finished-ness" ever since I saw it for the first time. My first impressions of GTK (circa 2006) were that it looked quaint, but didn't seem reasonable for serious use.

People had a go at XP's Luna theme for being "Fisher-Price" too - the kind of thing they might give a[n ostensibly less-intelligent] family member (or similar) to use, but not something they'd use themselves.

This draws an interesting distinction between the equally important halves of https://en.wikipedia.org/wiki/Look_and_feel: you can legitimately judge an OS by its look and get a few people to agree with you, but if the feel is right, the OS will win anyway.

I've always thought the GTK guys have very conspicuously lacked anyone who understands what the feel part of UX is, or that they need someone who understands that. Either that, or the threshold we're discussing is only fixable by doing work in every application. Objectively I do figure that the GNOME project has at least one decent UI designer on board, so it's more likely that it's probably something that requires collective single-minded agreement on a standard, something that requires focused investment from app developers, or both.

I've tried Enlightenment too, once several years ago, and once a few months ago. I don't actually remember what my setup of a few years ago looked like, but I remember there being a cute water effect animation for the desktop, which was really cool (especially the fact that it ran seemingly perfectly smoothly on a _several_-year-old 700MHz Pentium III laptop!). More recently I tried it out again (specifically focusing on Terminology), on a ridiculously old machine (800MHz Duron) as well as a modern i3 box. The Duron more or less choked full stop, but testing on the i3 was like... I've thrown all this extra hardware at this thing, and for what? I'm not really getting that much here.

Enlightenment is nice, but it also feels incomplete. I think E and GNOME probably both suffer from limited user testing and validation. The thing with for example Windows is that you have both official testing groups (and the presumable video data from sessions) along with an entire company of designers and developers to make noise about features. The result is something that feels incredibly intuitive and comfortable to use at the little-things-so-small-your-brain-doesn't-consciously-notice-them level, which, tying back to Uncanny Valley, is I think the area that makes the absolute most difference.

Worded differently, I very occasionally try and clip out loops of music I listen to. (I'm still looking for a nondestructive, fast music editor for Linux that will let me crossfade stuff. I find it hard to believe this doesn't exist. I use Rezound at the moment, I'm yet to see if Radium explodes on my ancient ThinkPad.) But with the looping thing, sometimes I'll extract a clip out of a song, and after Repeat #471 (where I've been repeatedly playing just the loop point), I reckon it sounds pretty good... so I zoom out and play the entire ~30 second loop... and suddenly the loop point sounds incredibly out of place. I think a similar thing can happen in any sort of abstract, subjective design - designers and developers can spend so much time on their work that they temporarily lose ground truth (possibly using a mechanism that shares some traits with https://en.wikipedia.org/wiki/Semantic_satiation, something I descovered recently). That's why a giant testing/validation/feedback team who are paid to shout at you and tell you why your thing is terrible is NEEDED. Open source doesn't have enough of this - it's either toxic vitriol that needs IP-blacklisting or 10 PRINT "that is the most awesome thing ever" 20 GOTO 10. Well, in more cases than is useful, at least; I'm sure sensible discussion does take place, but if it worked this wall of text and associated commentary wouldn't exist...

Finally, I think that, in trying to make something absolutely amazing, these projects are "punching above their weight", if you will; trying to put out a product beyond their collective capacity, either deliberately or indirectly. In every aspect of life, any activity can be engaged, but not every activity can be pulled off with confidence and competence. Trying something without confidence frequently doesn't work; trying something with confidence but without competence often turns to disaster and just pans out wrong.


This is a really, really good post, and I feel like you're hitting on the right thing.

You should write a blog post, not a HN comment.


Not much nowadays, but they still sometimes just change things "just because". They still have a bug with changing the number of volume steps, took they 5 years to accept any patch and it's still inadequate.

Luckily that's not frequent though.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: