| |||||||||||||
Kamp: A Generation Lost in the BazaarKamp: A Generation Lost in the BazaarPosted Aug 20, 2012 15:23 UTC (Mon) by landley (guest, #6789)In reply to: Kamp: A Generation Lost in the Bazaar by jmorris42 Parent article: Kamp: A Generation Lost in the Bazaar
Bazaar development can't do user interfaces. Any time "shut up and show me the code" isn't the correct response to the problem at hand, the development model melts down into three distinct failure modes:
1) Endless discussion with no resolution,
2) Fork to death with everybody implementing their own vision and then no way to merge them back together,
3) Delegate the problem to nobody by either A) separating implementation from interface and focusing on the engine, B) make it so configurable the fact it has no sane defaults is now somehow your fault.
Bazaar development either needs empirical tests so everybody can agree on what "good" looks like, or a Benevolent Dictator For Life who makes the calls on aesthetic issues.
Trying to do a GUI collaboratively is like asking wikipedia to write a novel. Too many cooks spoil the soup by making it taste bad, not by causing it to lack nutrition. We've only _selectively_ overcome brooks' law.
Rob
(Log in to post comments)
Kamp: A Generation Lost in the Bazaar Posted Aug 20, 2012 15:58 UTC (Mon) by Company (guest, #57006) [Link]
This has nothing to do with graphical user interfaces. It applies equally well to command line utilities, APIs and probably lots of other things.
It's just that GUIs is where everyone can see it. And probably because you got used to the fact that find uses single-dash arguments, perl takes -e while python takes -c and memcpy() uses dest before src while g_value_copy() uses src before dest.
Kamp: A Generation Lost in the Bazaar Posted Aug 20, 2012 16:41 UTC (Mon) by drag (subscriber, #31333) [Link]
> Bazaar development either needs empirical tests so everybody can agree on what "good" looks like, or a Benevolent Dictator For Life who makes the calls on aesthetic issues.
That's how you end up with people calling Gnome developers fascists.
Kamp: A Generation Lost in the Bazaar Posted Aug 20, 2012 17:18 UTC (Mon) by landley (guest, #6789) [Link]
Gnome didn't make just _one_ mistake. But in this context, it couldn't figure out if it was steered by committee or by fiat, so it seesawed back and forth a few times, which didn't go over well.
The standard way to move from "committee" to "fiat" is by forking, an example would be Galeon (and later Firefox) forking off of Mozilla. Or you can start over and be simple but reuse some parts, ala XFCE (done by Olivier Fourdan).
But Gnome started life as a reaction to KDE (a protest over licensing issues), which meant early on KDE got all of the pragmatists and Gnome got all the idealists. There's way more wrong with Gnome than just this one organizational issue. :)
Rob
Kamp: A Generation Lost in the Bazaar Posted Aug 20, 2012 17:28 UTC (Mon) by danieldk (guest, #27876) [Link]
> meant early on KDE got all of the pragmatists and Gnome got all the idealists
Yeah, well, except that big elephant in the room, that likes to wear a colored hat.
Kamp: A Generation Lost in the Bazaar Posted Aug 20, 2012 17:29 UTC (Mon) by drag (subscriber, #31333) [Link]
> Gnome didn't make just _one_ mistake.
Nobody in this thread mentioned any mistakes. At least not yet.
I am merely pointing out that making arbitrary decisions about what is good for your own software project isn't too popular among some Linux circles.
> The standard way to move from "committee" to "fiat" is by forking, an example would be Galeon (and later Firefox) forking off of Mozilla.
Neither of these are really examples of forks. Galeon started off as it's own web browser that used the Mozilla engine and Firefox is Mozilla's own project which they split off the non-browsing bits into separate applications and put a better UI on top of it.
Galeon essentially died, replaced by Epiphany (at least in spirit..) which now uses webkit and has been largely gone unappreciated because Chrome kicks so much ass.
Kamp: A Generation Lost in the Bazaar Posted Aug 20, 2012 21:54 UTC (Mon) by cmccabe (guest, #60281) [Link]
I don't understand why "shut up and show me the code" isn't the right response to a UI discussion. At the end of the day, user interfaces aren't some magical thing that only experts can fathom. They're just like any other engineering problem.
Unfortunately, there's a strong "popularity contest" aspect to user interfaces. We use QWERTY because QWERTY is familiar. QWERTY is familiar because we use QWERTY. etc. Similarly, a lot of arbitrary choices were made back in the last 20 years (close box on the right of the window, etc). Well-meaning engineers who create "unconventional" user interfaces find that most people are allergic to them, simply because they're unfamilar. (Case study: GNOME3.)
Another aspect is that UIs are targeted at different types of users. emacs and vi are great user interfaces-- for programmers. Not necessarily for non-programmers. Failure to design something with the needs of its intended audience in mind is another leading cause of UX failure.
None of this stuff means that UI design is some sooper secret thing that non-designers can't possibly comprehend. Just know your audience, and stick to the familiar whenever possible, and you'll be most of the way there.
Kamp: A Generation Lost in the Bazaar Posted Aug 20, 2012 23:31 UTC (Mon) by hummassa (subscriber, #307) [Link]
UI design (and its supersets, interaction design and experience design) are not sooper seecret, but to do a Good user interface you have to have a Good toolset and knowledge -- just like to make a good program you have to be a good programmer.
Any programmer can programmer an UI. But if it was done without regard to the best practices of UI design (reading the UI guidelines is mandatory, but by no means sufficient), it will be an ugly and/or clunky UI.
It can even be successful! What you said about QWERTY is true, because one of the tools in the UI designer's toolset is "to establish the current mental model of an user when dealing with a product or service". And one of the great problems with UI design from the point of view of a programmer is that good UI design requires a lot of user testing and friction -- you know, things the proverbial programmer is not quite fond of.
Kamp: A Generation Lost in the Bazaar Posted Aug 23, 2012 17:14 UTC (Thu) by jedidiah (guest, #20319) [Link]
No. The "best practices of UI design" are pretty worthless. The main focus should always been the end user and the use case. Otherwise you get nonsense like Ribbon and GNOME3. You end up with lots of academics completely divorced from reality spouting "principles" that may or may not have anything relation to anything.
You end up with UI disasters that go into the release of a project without even getting decent end user feedback first.
Even the much over hyped interfaces created by Apple suffer from this kind of foolish ivory tower approach to real problems.
Kamp: A Generation Lost in the Bazaar Posted Aug 23, 2012 18:48 UTC (Thu) by hummassa (subscriber, #307) [Link]
> No. The "best practices of UI design" are pretty worthless.
As I said in other topic, people who say stuff like this do not know anything about UI design. And this is PROVED by your next phrase:
> The main focus should always been the end user and the use case.
THAT is the FIRST best practice of UI design. Design for the user, thinking about the user, prototyping, iterating, testing with the user, etc.
> Otherwise you get nonsense like Ribbon and GNOME3.
Well, actually... user tests show that the Ribbon can actually be good when it's well designed, and lots of people around the world use Office 201x without lots of problems. Some people actually conducted PROPER testing on it.
GNOME3 does not have anything like that AFAICT, so, you are putting two very different things (united only for your dislike of them) together. Anedoctal evidence == No evidence.
> You end up with UI disasters that go into the release of a project without even getting decent end user feedback first.
Repeating what I already said: *real* UI designers don't do that.
> Even the much over hyped interfaces created by Apple suffer from this kind of foolish ivory tower approach to real problems.
At Apple, there was one engineer responsible for the calls on UI design. They had a lot of good eye for details -- and that's why it is where it is today -- but not a lot of best UI design practices like user testing. Specially in the last interactions. And they are lost in the controversial skeuomorphism fad (the jury is still out if it's good, bad or just plain ugly -- no conclusive user testing AFAIK).
Kamp: A Generation Lost in the Bazaar Posted Aug 23, 2012 19:57 UTC (Thu) by renox (subscriber, #23785) [Link]
>Well, actually... user tests show that the Ribbon can actually be good when it's well designed, and lots of people around the world use Office 201x without lots of problems. Some people actually conducted PROPER testing on it.
An user interface which takes lots of vertical space which is at premium instead of being on the sides which are much less used, "PROPER testing"?
Speaking about failure, I remember the Gnome's decision to use "spatial browsing" which was backed supposedly by usability research, they had to revert it in the end.
Kamp: A Generation Lost in the Bazaar Posted Aug 23, 2012 20:38 UTC (Thu) by hummassa (subscriber, #307) [Link]
> An user interface which takes lots of vertical space which is at premium instead of being on the sides which are much less used, "PROPER testing"? Somehow I doubt it!
The MSOffice201x ribbon actually takes less vertical space than LibreOffice's menu plus two (or three, in some cases) lines of toolbars. It is also spatially better distributed, and the size relations between the icons reflect their "use count" in user testings. See Deborah Hix's usability works... IIRC lateral toolsets have a problem because the distance the mouse has to travel is bigger. Anyway, people who do a lot of DTP or other "give me my vertical space" application IME make use of a rotated display, whenever possible!
> Speaking about failure, I remember the Gnome's decision to use "spatial browsing" which was backed supposedly by usability research, they had to revert it in the end.
I have heard those rumors, but I have never seen any real research.
Kamp: A Generation Lost in the Bazaar Posted Aug 24, 2012 6:40 UTC (Fri) by dgm (subscriber, #49227) [Link]
> Anedoctal evidence == No evidence.
Lots of "anecdotes" == overwhelming evidence.
Kamp: A Generation Lost in the Bazaar Posted Aug 24, 2012 20:08 UTC (Fri) by rgmoore (✭ supporter ✭, #75) [Link] Lots of "anecdotes" == overwhelming evidence. No. You only get meaningful, much less overwhelming, evidence if you have a representative sample, and listening to whomever yells the loudest is not a reliable way to get it. Improper sampling is exactly why the plural of anecdote is not data.
Kamp: A Generation Lost in the Bazaar Posted Aug 23, 2012 19:42 UTC (Thu) by rgmoore (✭ supporter ✭, #75) [Link] Otherwise you get nonsense like Ribbon and GNOME3. You're not helping yourself out with those examples. It's admittedly anecdotal, but my personal experience with both is that I initially hated them for being different, but found that they're actually more efficient than the old system once I figured out how to use them. If anything, I'd hold them up as examples of how we should be skeptical of people's instinctive and uninformed reaction to UI changes, and pay more attention to people who have actually studied this stuff.
Kamp: A Generation Lost in the Bazaar Posted Aug 21, 2012 9:04 UTC (Tue) by fb (subscriber, #53265) [Link]
> 3) Delegate the problem to nobody by either A) separating implementation from interface and focusing on the engine, B) make it so configurable the fact it has no sane defaults is now somehow your fault.
Ouch. I've seen "B" so often in my life that it just isn't funny.
[...]
Does anyone else has memories of a dreaded question about 'less' charset configuration that took one full screen shoot of text when installing Debian? Every time I faced that I realized that there was really nobody in charge of reviewing the Debian installation steps.
|
|||||||||||||