UI for 1.0
Gervase Markham
that to come but, in the mean time, here are st...@mozilla.org's
thoughts for the UI changes we should be making in the run-up to 1.0
(from here on in.)
UI is an area which produces a large amount of strongly-held feeling.
There are often arguments which can drag on, so it's important to
establish who exactly has the last word on UI issues. The rule is
currently that the module owner of the code owns the UI. The very last
word on a particular change lies with st...@mozilla.org, but they may
only be appealed to if you can make a good case that the module owner is
vetoing the change for inappropriate reasons. If your change gets
vetoed, shrug your shoulders and move on.
1) Let's fix what sucks. This may seem obvious, but there's a bug titled
"the ugliest dialog in the product" -
http://bugzilla.mozilla.org/show_bug.cgi?id=94437 . It was filed 40 days
ago and has seen no activity. All who have seen that dialog agree that
fixing it is uncontroversial. But it hasn't been worked on by anyone.
2) Let's pick the low-hanging fruit rather than stripping the entire
tree. If a bit of UI sucks, it is often better to tweak the XUL to make
it suck a bit less, rather than spend far more time to improve it more
but take longer and run into more controversy.
For example, if a particular dialog sucks, you could go down one of two
routes:
a) Produce a spec of a totally different ideal UI for the dialog, with
widgets for a few features than don't exist yet and which require
back-end work. Have a couple of months of argument and refinement in the
newsgroups, then try and find someone to implement it. Have more
argument about the implementation.
b) Say: "It'll suck a lot less if we change the layout in this way and
rename these two text boxes", and watch as people fall over themselves
to fix it in an afternoon. See:
http://bugzilla.mozilla.org/show_bug.cgi?id=99860 for a classic example
of this.
3) Let's take it window by window. A patch containing ten small fixes is
probably more likely to get reviewer and super-reviewer time than ten
one-fix patches. File meta-bugs, collect all the issues for one window
and fix them in one go.
We can make Mozilla's UI suck a lot less for 1.0 by cleaning up the
obviously nasty stuff rather than making larger, more sweeping changes.
In this respect, it's the same as the compiled code - where we are
trying to avoid large, destabilising checkins. The time for redesigns is
past - let's polish, but let's do it in a coordinated way so we aren't
producing millions of one-line patches.
Cormac F
by window.
However, I hope that post-1.0, we take a long hard look at the browser
UI, the menu system in particular. For the novice user, current
mozilla menus are overly complex - e.g. 'view' has too many options,
'search' should go under 'edit', 'go' is redundant, 'tasks' has too
many submenus, etc. etc. In comparison to IE, the learning curve could
prove too steep.
Perhaps it isn't in the spirit of open-source and, but arguments about
changes, large or small, to the menu system might be prevented if a
usability expert were brought in by the mozilla group to submit a
complete specification for the browser suite?
Just a thought.
Cormac F
Gervase Markham <ge...@mozilla.org> wrote in message news:<3BB0DFB1...@mozilla.org>...
Al Rider
sort function.
Sorting when in Manage Bookmarks is necessary; but, not sufficient.
When the user has dozens, even hundreds, of bookmarks, they will not sort them manually. And,
even if done manually, everytime a new bookmark is added, it requires the user to manuallly
place it.
Al.................
Neil
>For the novice user, current mozilla menus are overly complex - e.g. 'view' has too many options, 'search' should go under 'edit', 'go' is redundant, 'tasks' has too many submenus, etc. etc. In comparison to IE, the learning curve could prove too steep.
>
How is 'go' redundant?
>Perhaps it isn't in the spirit of open-source and, but arguments about changes, large or small, to the menu system might be prevented if a usability expert were brought in by the mozilla group to submit a complete specification for the browser suite?
>
I hope advanced users like me won't want large changes to the menu
system. But it is certainly within the capacity of Mozilla to provide an
switchable alternate menu system. We'll just add another menuitem so
that the advanced menus can be turned back on :-)
Gervase Markham
> UI, the menu system in particular. For the novice user, current
> mozilla menus are overly complex - e.g. 'view' has too many options,
> 'search' should go under 'edit', 'go' is redundant, 'tasks' has too
> many submenus, etc. etc.
You're a man after my own heart. :-)
Gerv
Cormac F
Well, maybe it isn't. People have argued for and against this in these
groups for a while. My impression is that few use it, especially those
migrating from IE. 'Go' doesn't seem to contain any useful features
that aren't already available from the browser UI or through the
user's OS. Is it worth keeping a whole menu item for the small number
that use it? Again, thats just my impression - perhaps more people
make use of 'go' than I think. Anyway, I'm waffling now.
>You're a man after my own heart. :-)
Nice to hear a good hiberno/anglo phrase coming from across the
atlantic.
Can you tell I've had 5 hours sleep and three mugs of coffee? I'm
talking complete shite again.
jesus X
I've agreed with a lot of these... But, then I think of it in different terms...
Examples:
All of us who think Search should go under Edit think so because that's the
normal convention. But is it really an editing function when applied to web
pages? No, but then again neither are any of the other functions in the Edit
menu, because you can't edit a page in the browser itself. So, rather than
moving the Search menu, the Edit menu needs renamed, and put search back under
it. But, since Edit is the closest thing we've got, and it's the standard
convention, let's leave it for now. So, yes, Search SHOULD go under Edit.
View is poorly named. Display is a better name for it. Although, I think the
number of options under it are fine to me, they all seem appropriate.
Go, should go under the horrible named Tasks menu, which should be called Window
as is the convention.
And lastly, Help should be the last menu entry. Every single other app follows
this rule of thumb...
--
jesus X [ Booze-fueled paragon of pointless cruelty and wanton sadism. ]
email [ jesusx @ who.net ]
web [ http://burntelectrons.com ] [ Updated April 29, 2001 ]
tag [ The Universe: It's everywhere you want to be. ]
warning [ War doesn't determine who is right, war determines who is left. ]
Ben Ruppel
removed without messing with stuff between help and the rest of the
items. Tasks can be renamed Window, but only if Privacy & Security and
Tools find a new home. Quite frankly, I think Task > Tools is a bit
unintuitive for people looking to import an addressbook or look at history.
jesus X
>
> I think Help isn't last for debug purposes. Debug and QA can later be
> removed without messing with stuff between help and the rest of the
> items.
I'm assuming that they're there for the duration of Mozilla, since "it's not an
end user product!" :)
> Tasks can be renamed Window, but only if Privacy & Security and
> Tools find a new home. Quite frankly, I think Task > Tools is a bit
> unintuitive for people looking to import an addressbook or look at history.
Agreed. How about File > Tools?
Peter Trudelle
> UI is an area which produces a large amount of strongly-held feeling.
> There are often arguments which can drag on, so it's important to
> establish who exactly has the last word on UI issues.
Most of this feeling is purely personal opinion, with no basis in
commonly held UI principals, let alone empirical data from usability
tests, interpreted by experts into recommendations that reliably improve
the UI. I think it would be more useful, in the long run, to establish
who is responsible for the UI, including the visual design and
interaction design. Of primary importance is the normal process for
determining the UI, who has the last word in resolving intractable
issues should ideally be secondary, and rarely needed.
> The rule is currently that the module owner of the code owns the UI.
This is unfortunate, since module ownership is based on coding ability,
which has little or nothing to do with skill in UI design. Also, the
mapping between the modules and the UI is not clear. For instance,
under this model, who owns the menus? Since the word menu doesn't even
occur in mozilla's module owner page, must we assume that individual
module owners own individual menu items, and can add them anywhere they
want? There are similar problems for toolbars, dialogs, windows, etc.
We need an ownership model that matches the UI, not the code. I think
it would be good for mozilla.org to endorse such a model, with expert UI
owners who act as the equivalent of the module owners and
super-reviewers for UI changes. I think the natural divisions for this
ownership would be along applications, components, windows, and perhaps
even their major parts- sidebar, toolbars, and menus.
> The very last word on a particular change lies with st...@mozilla.org
> , but they may only be appealed to if you can make a good case that
> the module owner is vetoing the change for inappropriate reasons. If
> your change gets vetoed, shrug your shoulders and move on.
I'm sure this has to be the case, since mozilla.org owns the code, but
how will these decisions be made? I'm not aware that st...@mozilla.org
itself currently has any more UI expertise than the module owners
(though I'd be glad to be shown otherwise. ;-) In the past, we have
infrequently assembled an informal group of UI owners consisting of
several generally accepted lead UI developers (Hyatt, Ben, Blake, Jag,
et. al.) UE leads (MPT, German, et. al.) plus mozilla.org (that'd be you
Gerv!) and other stakeholders to meet as the PixelJockeys, to handle
intractable UI issues by consensus. I'm sure it has never enjoyed wide
enough participation from mozillans, but decisions these folks make
tend to stick. I think it would be good for mozilla.org to endorse such
a group to settle contentious issues that resist normal process.
> 1) Let's fix what sucks. This may seem obvious, but there's a bug
> titled "the ugliest dialog in the product" -
> http://bugzilla.mozilla.org/show_bug.cgi?id=94437 . It was filed 40
> days ago and has seen no activity. All who have seen that dialog agree
> that fixing it is uncontroversial. But it hasn't been worked on by
> anyone.
Thanks! Your mention seems to have triggered some good work on that
bug. Who could argue with fixing what sucks? There is much to be done...
> 2) Let's pick the low-hanging fruit rather than stripping the entire
> tree. If a bit of UI sucks, it is often better to tweak the XUL to
> make it suck a bit less, rather than spend far more time to improve it
> more but take longer and run into more controversy.
We used to call this 'polishing the turd' in the old Communicator
codebase. The desire and need to get away from that is what prompted
the rewrite. Sucking less is better, and continuous iterative
improvement certainly has its place in a mature product, but doing
something great is almost always well worth the added time and
controversy. However, I concede your point in the context of mozilla 1.0.
> A patch containing ten small fixes is probably more likely to get
> reviewer and super-reviewer time than ten one-fix patches. File
> meta-bugs, collect all the issues for one window and fix them in one go.
This may be expedient from a code standpoint, and can be good when
grouping changes that form a coherent whole, but it can also be
misused, like popular laws that are submitted for vote with lots of
riders. Unless reviewers are careful, and exercise a 'line-item veto'
when needed, we could get numerous questionable changes riding along
with each necessary change. This is especially important in the release
endgame, when each change's risk/reward ratio is magnified in
importance, requiring each to be carefully weighed before deciding
whether it goes in. I have almost always insisted on having at most one
issue per bug, so that it can be resolved or reopened independently, and
everyone is clear what is or is not fixed. Laundry-list bugs are rarely
satisfactorily resolved.
> We can make Mozilla's UI suck a lot less for 1.0 by cleaning up the
> obviously nasty stuff rather than making larger, more sweeping
> changes. In this respect, it's the same as the compiled code - where
> we are trying to avoid large, destabilising checkins. The time for
> redesigns is past - let's polish, but let's do it in a coordinated way
> so we aren't producing millions of one-line patches.
Right, but hopefully we'll have time for more than just the nasty stuff.
Let's also be sure to fix the minor annoyances that hit us dozens of
times per day, such as focus problems, dead keys, lost selections,
missed repaints, etc. While not as severe as other defects when taken
in isolation; when considered in total their frequency can make the
difference between a smooth online experience or one where your app is
constantly getting in your way and tripping you up.
Peter
Gervase Markham
>> There are often arguments which can drag on, so it's important to
>> establish who exactly has the last word on UI issues.
>
> Most of this feeling is purely personal opinion, with no basis in
> commonly held UI principals, let alone empirical data from usability
> tests, interpreted by experts into recommendations that reliably improve
> the UI.
Peter, that sounds very much to me like code for "Netscape UE engineers
should be the only people to have a say in the UI because they are the
only people with resources to carry out usability testing." And
"interpreted by experts" sounds suspiciously like "telling you that it
means what would be most convenient for Netscape." Call me cynical, but
the lack of Netscape's publishing any usability data has been a sore
point in the community for some time. I seem to remember that, at the
very first Pixeljockies six weeks ago, selmer (as chair) agreed with
German that the raw usability data from the last round of tests would be
published, along with any intepretation that German cared to give it.
But it doesn't seem to have materialised.
Which gives the tester more useful UI feedback and experience?
- Running a usability test with a handful participants for an afternoon
- Working in an Internet Cafe 40 hours a week and seeing people
attempting to use browsers all day long
- Listening to what some of the 150,000 people who downloaded our last
milestone say about it
There are a range of ways to become knowledgeable about UI, and (as with
any other mozilla.org thing) this is a meritocracy - people who
demonstrate competence by making sensible points, and writing
defendable, well-thought-out specs, and participating in the community,
will rise to the top.
> I think it would be more useful, in the long run, to establish
> who is responsible for the UI,
You are absolutely right. No-one is claiming the current situation is
anywhere near perfect. However, having a overall UI module owner
requires several things:
- A person who has the time to do it, and is willing to take input from
all sections of the community and make decisions which are not based
solely on the interests of one participant (even if they work for that
participant)
- Everyone else agreeing to abide by the decisions he or she makes
without running off in a huff or claiming that it makes their
work/job/life impossible
- Everyone to realise that they might not always get what they want
This list is probably incomplete; mozilla.org is currently working on a
document which more clearly defines the role of a module owner.
>> The rule is currently that the module owner of the code owns the UI.
>
> This is unfortunate, since module ownership is based on coding ability,
> which has little or nothing to do with skill in UI design. Also, the
This is true. It is unfortunate. See above. :-)
> need an ownership model that matches the UI, not the code. I think it
> would be good for mozilla.org to endorse such a model, with expert UI
> owners who act as the equivalent of the module owners and
> super-reviewers for UI changes.
I'm sure mozilla.org would love this model, and is actively looking for
owners who fit the criteria above.
> (though I'd be glad to be shown otherwise. ;-) In the past, we have
> infrequently assembled an informal group of UI owners consisting of
> several generally accepted lead UI developers (Hyatt, Ben, Blake, Jag,
> et. al.) UE leads (MPT, German, et. al.) plus mozilla.org (that'd be you
> Gerv!) and other stakeholders to meet as the PixelJockeys, to handle
> intractable UI issues by consensus. I'm sure it has never enjoyed wide
> enough participation from mozillans, but decisions these folks make
> tend to stick. I think it would be good for mozilla.org to endorse such
> a group to settle contentious issues that resist normal process.
mozilla.org has a forum for handling UI issues. It's called
netscape.public.mozilla.ui.
Pixeljockies is an inappropriate forum for a wide list of reasons, some
of which are:
- the email alias is Netscape-internal; no-one else can subscribe to it
- the meetings have never been publicised in a mozilla.org forum
together with an invitation to participate
- there is an overwhelming majority of representation of Netscape at the
expense of other interests
>> A patch containing ten small fixes is probably more likely to get
>> reviewer and super-reviewer time than ten one-fix patches. File
>> meta-bugs, collect all the issues for one window and fix them in one go.
>
> This may be expedient from a code standpoint, and can be good when
> grouping changes that form a coherent whole, but it can also be
> misused, like popular laws that are submitted for vote with lots of
> riders.
I think you misunderstand the smallness of "small". :-) This is not an
argument for more than one issue per bug, it's an argument for metabugs,
and letting reviewers spend 20 minutes on review of one patch, rather
than ten lots of five minutes on ten three-liners all in the same dialog.
> Right, but hopefully we'll have time for more than just the nasty stuff.
> Let's also be sure to fix the minor annoyances that hit us dozens of
> times per day, such as focus problems, dead keys, lost selections,
> missed repaints, etc.
This is what I meant by "small" :-)
Gerv
Peter Trudelle
> Peter, that sounds very much to me like code for "Netscape UE
> engineers should be the only people to have a say in the UI because
> they are the only people with resources to carry out usability testing."
Uh, who are you quoting? I didn't say anything remotely resembling
that. Let's not polarize this by putting words in each other's mouths.
Anyone can carry out guerilla usability studies, and spread the
results. Anyone who has valid data should be given more credence than
anyone who has only opinions.
> And "interpreted by experts" sounds suspiciously like "telling you
> that it means what would be most convenient for Netscape." Call me
> cynical, but the lack of Netscape's publishing any usability data has
> been a sore point in the community for some time. I seem to remember
> that, at the very first Pixeljockies six weeks ago, selmer (as chair)
> agreed with German that the raw usability data from the last round of
> tests would be published, along with any intepretation that German
> cared to give it. But it doesn't seem to have materialised.
Well, I know you're not cynical, but you do seem to be trying to twist
what I said. I didn't even mention Netscape, yet you seem to be trying
to portray this as 'us versus them'. Are we not all on the same side?
And FYI, the fist PJ meetings were ~18 months ago, with several
mozillians invited, and some attending. Since well before that, I have
personally been trying to get NS UE to publish their raw tapes, as well
as to invite mozilla.org participants to the tests themselves. Their
inability to do so is largely due to lack of resources, and I don't
believe it makes their results or recommendations suspect. I attend
these usability tests as often as possible, which is why I value the
results and want more participants to experience them. BTW, where is
mozilla.org's published usability data, or any other participant's for
that matter?
> Which gives the tester more useful UI feedback and experience?
I don't know, but all of the ones you list, and more, are useful and
should be considered.
> - Running a usability test with a handful participants for an afternoon
Have you ever attended such tests? There is much value in
professionally run tests, conducted under labratory conditions, with
results evalutated by trained professionals.
> - Working in an Internet Cafe 40 hours a week and seeing people
> attempting to use browsers all day long
I value this kind of data too, and use it informally myself. It is
obviously where MPT gets much of his insight into our UI problems. But
lets realize that it is more subjective and anecdotal, and the users may
be a self-selected class.
> - Listening to what some of the 150,000 people who downloaded our last
> milestone say about it
Are we building this only for people who are downloading milestones? Are
these people representative of users in general? I think not. They
represent a very vocal, technically-savvy, yet tiny minority of the
users targetted by mozilla.org, let alone Netscape. They are very
useful in determining the overall quality of the software, but I think
building the UI to their tastes could drastically limit its acceptance
by the masses.
> There are a range of ways to become knowledgeable about UI, and (as
> with any other mozilla.org thing) this is a meritocracy - people who
> demonstrate competence by making sensible points, and writing
> defendable, well-thought-out specs, and participating in the
> community, will rise to the top.
What is the top for them? You've already said that the UI is controlled
by module owners. That is code ownership, and the UI experts are mostly
not production programmers who will ever become module owners.
> However, having a overall UI module owner requires several things:
>
> - A person who has the time to do it, and is willing to take input
> from all sections of the community and make decisions which are not
> based solely on the interests of one participant (even if they work
> for that participant)
> - Everyone else agreeing to abide by the decisions he or she makes
> without running off in a huff or claiming that it makes their
> work/job/life impossible
> - Everyone to realise that they might not always get what they want
A good description of any UI owner. We need someone in the 'buck stops
here' role, but I'm not sure that any one person having overall UI
ownership would be enough. As with the code, we need area owners and an
overall owner, but also a group with broad expertise to act as
super-reviewers. Of course, heavily implicit in all of this is that
they'd be qualified to make these decisions.
> mozilla.org has a forum for handling UI issues. It's called
> netscape.public.mozilla.ui.
Come on, this is a forum for discussing them, not for making binding
decisions.
> Pixeljockies is an inappropriate forum for a wide list of reasons,
> some of which are:
> - the email alias is Netscape-internal; no-one else can subscribe to it
Note that I encouraged you to endorse something like PJ, not necessarily
PJ itself. PJ is not intended to be wide open to all, that is what
n.p.m.ui is for. It is for discussion and resolution of intractable
issues by the UI leads themselves. I have always adjusted this list as
requested by the participants.
> - the meetings have never been publicised in a mozilla.org forum
> together with an invitation to participate
Not true. While the first meeting, in February 2000 was primarily
Netscape UI owners, we quickly realized the folly of that, and have
invited recognized mozilla UI leads ever since. You can read the
threads from that time in this newsgroup, including my PixelDust notes
from the next meeting on March 8. We resolved several issues in these
meetings, including some ownership questions, after which my role as
moderator was taken over by Paul Hangas. I'm not familiar with what
happened between then and the current invocation, but I still see a need
for such a group, and much value in having wider participation and
endorsement.
> - there is an overwhelming majority of representation of Netscape at
> the expense of other interests
You could also say that about the module owners, or even
st...@mozilla.org itself. It is a natural result of one company
providing more resources for this project than any other single
organization. Does mozilla.org now think that is a bad thing? In any
case, nobody can force mozilla.org or Netscape contributors to attend
these meetings. Overall, many are invited, but few show up. I'd love
to fix that, we need involvement in order to be effective. Who should
be involved? Why aren't they?
Peter
Peter Trudelle
>Ben Ruppel wrote:
>
>>I think Help isn't last for debug purposes. Debug and QA can later be
>>removed without messing with stuff between help and the rest of the
>>items.
>>
>
>I'm assuming that they're there for the duration of Mozilla, since "it's not an
>end user product!" :)
>
Its not? Am I missing a wink here? Isn't mozilla the default browser
on Red Hat Linux, and hopefully other distros? Doesn't that represent
the largest target userbase? Or do you not consider them to be end users?
Peter
Gervase Markham
> Gervase Markham wrote:
>
>> Peter, that sounds very much to me like code for "Netscape UE
>> engineers should be the only people to have a say in the UI because
>> they are the only people with resources to carry out usability testing."
>
> Uh, who are you quoting? I didn't say anything remotely resembling
> that.
Netscape is the only company with the resources to do usability testing
of Mozilla, and by your definitions, only professional UI designers
(such as those employed by Netsape) are experts. You message implied
that "empirical data from usability tests, interpreted by experts" was
an important factor in deciding what the UI should be.
> Let's not polarize this by putting words in each other's mouths.
> Anyone can carry out guerilla usability studies, and spread the
> results. Anyone who has valid data should be given more credence than
> anyone who has only opinions.
In which case, given that no-one has published any data, it's all
opinion. Does this mean the nod should go to whichever side is most vocal?
My view is that anyone who can back up their ideas with reasoned
argument should be given the credence. If you produce a spec and, when
questioned, can justify why it is the way it is and how it fits into the
overall plan, this is much more credible than just chiming in
occasionally with your $0.02.
>> And "interpreted by experts" sounds suspiciously like "telling you
>> that it means what would be most convenient for Netscape." Call me
>> cynical, but the lack of Netscape's publishing any usability data has
>> been a sore point in the community for some time. I seem to remember
>> that, at the very first Pixeljockies six weeks ago, selmer (as chair)
>> agreed with German that the raw usability data from the last round of
>> tests would be published, along with any intepretation that German
>> cared to give it. But it doesn't seem to have materialised.
>
> Well, I know you're not cynical, but you do seem to be trying to twist
> what I said. I didn't even mention Netscape, yet you seem to be trying
> to portray this as 'us versus them'. Are we not all on the same side?
I'm not portraying it as "us vs. them". However, I think it's becoming
more and more true that Netscape and Mozilla have different goals when
it comes to the UI; and the current situation seems to be that any
changes which Netscape's UE team does not agree with magically get stuck
somewhere before making it in; conversely, changes which are not wanted
by the community (and even some parts of Netscape - "them" vs. "them",
if you like) are forced in. I'm sure you know which two particular bugs
I'm thinking of. This is not a community process.
> And FYI, the fist PJ meetings were ~18 months ago, with several
> mozillians invited, and some attending. Since well before that, I have
> personally been trying to get NS UE to publish their raw tapes, as well
> as to invite mozilla.org participants to the tests themselves. Their
> inability to do so is largely due to lack of resources,
Hand the documents over to me, and I'll see they get published. If you
feel it's important for people to be able to watch the tapes as well,
we'll see what we can arrange about that, too. We wouldn't want to let
people's business stand in the way :-) I have never seen a public
invitation for any mozilla.org participant to attend a Netscape
usability test.
> results and want more participants to experience them. BTW, where is
> mozilla.org's published usability data, or any other participant's for
> that matter?
Netscape engineers have expressed opinions in bugs based on usability
data they refuse to reveal. I don't know of any other contributor who
has done this. If any did, I would request the same sort of disclosure.
> I value this kind of data too, and use it informally myself. It is
> obviously where MPT gets much of his insight into our UI problems. But
> lets realize that it is more subjective and anecdotal, and the users may
> be a self-selected class.
"Internet cafe users" is about as broad a cross-section as you can get -
perhaps biased towards the less able end, where people don't have
computers at home or require a helping hand from an assistant. And he
sees hundreds of them using products every day in a natural setting,
whereas the usability tests I am aware of happen every six months or so,
with a dozen people, for a single day.
My point is not that one is better than the other; it is merely that
what I continually hear from German, you and others about how formal
usability testing is the only way to make decisions, and expert
interpretation is needed of the results, is bogus.
>> - Listening to what some of the 150,000 people who downloaded our last
>> milestone say about it
>
> Are we building this only for people who are downloading milestones? Are
This is getting into the "what is Mozilla's target user" debate; but
discounting these people's inputs because they are technologically savvy
implies they none of the 150,000 have the ability to put themselves in
other people's shoes, or have tried Mozilla out on their less
technically advanced friends/relations.
> building the UI to their tastes could drastically limit its acceptance
> by the masses.
It's also possible that building the UI to the tastes of the masses,
whatever they are, could cause our extremely useful testing resource to
go away. This would be a great loss to the Mozilla project.
>> There are a range of ways to become knowledgeable about UI, and (as
>> with any other mozilla.org thing) this is a meritocracy - people who
>> demonstrate competence by making sensible points, and writing
>> defendable, well-thought-out specs, and participating in the
>> community, will rise to the top.
>
> What is the top for them? You've already said that the UI is controlled
> by module owners. That is code ownership, and the UI experts are mostly
> not production programmers who will ever become module owners.
The top is having your opinion respected and listened to by the people
making the decisions, and perhaps (when we have UI module owners) being
one of those.
>> mozilla.org has a forum for handling UI issues. It's called
>> netscape.public.mozilla.ui.
>
> Come on, this is a forum for discussing them, not for making binding
> decisions.
Why do binding decisions have to be made in a face-to-face meeting?
Decisions about every other part of Mozilla are made on IRC, in
Bugzilla, and in the newsgroups all the time. Why is UI different?
>> Pixeljockies is an inappropriate forum for a wide list of reasons,
>> some of which are:
>> - the email alias is Netscape-internal; no-one else can subscribe to it
>
> Note that I encouraged you to endorse something like PJ, not necessarily
> PJ itself. PJ is not intended to be wide open to all, that is what
> n.p.m.ui is for. It is for discussion and resolution of intractable
> issues by the UI leads themselves.
Any decisions about Mozilla's UI should be made by a group of people
consisting solely of those who fit the criteria of being
community-oriented which I outlined in my original post.
As you and I both know, up to this point certain Netscape UE engineers
have asked not to be made UI module owners because they feel they can't
do the job for Netscape and also be responsive to the community. Which
is fine, but if they are attending Pixeljockies, then it's not a meeting
about Mozilla's UI, it's about the UI of Netscape's next release. And it
has therefore no authority to make Mozilla UI decisions.
Face-to-face meetings are inherently inconvenient in a global
organisation - and no record remains of what was said.
Newsgroup/email/Bugzilla discussions are much more suitable.
> I have always adjusted this list as requested by the participants.
I do seem to remember that I enquired whether it were possible to add
mpt, or another external participant, to the list and was told that it
wasn't possible.
>> - the meetings have never been publicised in a mozilla.org forum
>> together with an invitation to participate
>
> Not true.
Let me rephrase that. "The current set of meetings have...".
Gerv
Jesse Ruderman
a simple bookmarks
> sort function.
>
> Sorting when in Manage Bookmarks is necessary; but, not sufficient.
>
> When the user has dozens, even hundreds, of bookmarks, they will not sort
them manually. And,
> even if done manually, everytime a new bookmark is added, it requires the
user to manuallly
> place it.
Please vote for http://bugzilla.mozilla.org/show_bug.cgi?id=77411.
jesus X
> jesus X wrote:
> >Ben Ruppel wrote:
> > I'm assuming that they're there for the duration of Mozilla, since "it's not
> > an end user product!" :)
> Its not?
No. Do you really need to even ask this question after all this time?
> Am I missing a wink here?
A wink, no. A clue, apparently.
> Isn't mozilla the default browser on Red Hat Linux, and hopefully other
> distros?
It very well may be. This is a choice of the distributor, not Mozilla. I'm sure
RedHat made it so, because of Chris Blizzard's fantastic involvement, and his
being on their payroll as their Mozilla programmer.
> Doesn't that represent the largest target userbase?
Of Mozilla? No. Mozilla itself is being targeted at people who want to roll
their own distro, much like the Linux kernel isn't targeted at John Q. Public
but at Linux distros.
> Or do you not consider them to be end users?
Of course RedHat isn't the end user. That's silly. That's like saying that car
dealerships are the end users of cars. There are lot of people who use Mozilla
directly from mozilla.org but that's not the target in mind. I don't consider
myself an end user of Mozilla so much as a contributor who takes binaries, beats
on them, shakes out bugs, and tries to create helpful reports, confirm reports,
etc.
jesus X
> Most of this feeling is purely personal opinion, with no basis in
> commonly held UI principals, let alone empirical data from usability
> tests, interpreted by experts into recommendations that reliably improve
> the UI. I think it would be more useful, in the long run, to establish
> who is responsible for the UI, including the visual design and
> interaction design. Of primary importance is the normal process for
> determining the UI, who has the last word in resolving intractable
> issues should ideally be secondary, and rarely needed.
At this point, the UI is functional, and nominally usable. Continuing on the
subject of the target market of Mozilla, distributors, not end users, this type
of UI is all that's needed. The difficulty of changing the UI is not difficult
at all to change, so any distributor that chooses to do research, and have it
interpreted by "experts" (who very often times have no more idea of usability
than anyone else, and also have a nasty tendency to make stupid decisions, more
to follow) as much as their budget can tolerate.
Further, UI experts are worth their weight in lead. UI experts are just as hit
and miss as everyone else. The most "famous" UI expert research led to the lead
balloon "Bob" which cost many millions of MS's research dollars. UI experts are
worthless when you have the direct input on hundreds of thousands of people who
actually use the product and test it out daily.
Peter Trudelle
>
> Netscape is the only company with the resources to do usability
> testing of Mozilla, and by your definitions, only professional UI
> designers (such as those employed by Netsape) are experts. You message
> implied that "empirical data from usability tests, interpreted by
> experts" was an important factor in deciding what the UI should be.
It is, but I never gave any such definition, or said that Netscape was
the only company that could do this, those are your words.
> In which case, given that no-one has published any data, it's all
> opinion. Does this mean the nod should go to whichever side is most
> vocal?
The data exists. The fact that it is not (yet) published does not
negate its value. You seem to have no problem taking MPT's
interpretation of observations at his web cafe at face value without
having the original data published, why do you have such a problem with
recommendations put forth by Netscape? I find this particularly
puzzling since you are in fact yourself a Netscape employee, and have
complete access to all of this data.
> I'm not portraying it as "us vs. them". However, I think it's becoming
> more and more true that Netscape and Mozilla have different goals when
> it comes to the UI; and the current situation seems to be that any
> changes which Netscape's UE team does not agree with magically get
> stuck somewhere before making it in; conversely, changes which are not
> wanted by the community (and even some parts of Netscape - "them" vs.
> "them", if you like) are forced in. I'm sure you know which two
> particular bugs I'm thinking of.
No, I don't know which bugs you mean. Netscape as a company does have
some goals for the UI that have no place in mozilla (*cough*
Netscape.com *cough*), but I am convinced that the target end users are
substantially the same. If they are not, then we are wasting our time
trying to share such a similar UI. I cite as example the fact that
Netscape has been used for years by a large number of Linux users. Now
that Mozilla is replacing Linux in RedHat and presumably elsewhere, it
will be addressing the exact same userbase.
> Hand the documents over to me, and I'll see they get published. If you
> feel it's important for people to be able to watch the tapes as well,
> we'll see what we can arrange about that, too. We wouldn't want to let
> people's business stand in the way :-) I have never seen a public
> invitation for any mozilla.org participant to attend a Netscape
> usability test.
They are not mine to hand over. You have the same access to them that I
do though, so I'm hoping that your offer to distribute them, or at least
a non-proprietary subset, will be welcomed by the UE team.
> "Internet cafe users" is about as broad a cross-section as you can get
> - perhaps biased towards the less able end, where people don't have
> computers at home or require a helping hand from an assistant. And he
> sees hundreds of them using products every day in a natural setting,
> whereas the usability tests I am aware of happen every six months or
> so, with a dozen people, for a single day.
You're welcome to your opinion, but I stand by my statement that such
studies are "more subjective and anecdotal, and the users may be a
self-selected class". I would encourage you to attend a professional
usability study before criticizing them.
> My point is not that one is better than the other; it is merely that
> what I continually hear from German, you and others about how formal
> usability testing is the only way to make decisions, and expert
> interpretation is needed of the results, is bogus.
What is bogus is your continued attempts to put words in my mouth.
Please stop. I have never said, nor do I believe that German or anyone
else has said that "formal usability testing is the only way to make
decisions". This is doubly absurd because it isn't even a way to make
decisions, it is merely one way to provide data to the decision process.
> This is getting into the "what is Mozilla's target user" debate; but
> discounting these people's inputs because they are technologically
> savvy implies they none of the 150,000 have the ability to put
> themselves in other people's shoes, or have tried Mozilla out on their
> less technically advanced friends/relations.
I didn't discount their input, I just don't take it for granted that
they represent the vast majority of users. In my opinion, they do not.
Overwhelmingly, the views expressed by these and other participants are
personal preferences. We should all try to discount our own preference
when making decisions about the UI.
> It's also possible that building the UI to the tastes of the masses,
> whatever they are, could cause our extremely useful testing resource
> to go away. This would be a great loss to the Mozilla project.
Yes, that is a very real danger, which is why we need to be clear on who
we are building this for. Is it just for the mozilla.org participants,
or do we want this codebase to be widely used? At a time when about 93%
of the people on earth have never used a computer, do we really want to
base UI decisions on the personal preferences of the .0025% who
currently download mozilla?
> Why do binding decisions have to be made in a face-to-face meeting?
> Decisions about every other part of Mozilla are made on IRC, in
> Bugzilla, and in the newsgroups all the time. Why is UI different?
I never said they had to be made in face-to-face meetings, why do you
ask? The difference with UI is that it needs a coherent design overall.
Thousands of individual decisions cannot add up to a good UI. In a
project this large, it seems impossible to have a single designer, which
would be desirable, so there has to be consensus among the leaders.
> As you and I both know, up to this point certain Netscape UE engineers
> have asked not to be made UI module owners because they feel they
> can't do the job for Netscape and also be responsive to the community.
No, I don't know that. In what public forum did this happen?
> Which is fine, but if they are attending Pixeljockies, then it's not a
> meeting about Mozilla's UI, it's about the UI of Netscape's next
> release. And it has therefore no authority to make Mozilla UI decisions.
Who said it was? I pointed it out as an example of something _LIKE_
what I thought was needed. There is no need to attack it, or criticize
it for what it is not. However, if module owners get together in any
forum and agree on a particular UI, then by your own rules they would
have the authority.
> Face-to-face meetings are inherently inconvenient in a global
> organisation - and no record remains of what was said.
> Newsgroup/email/Bugzilla discussions are much more suitable.
Meetings have their place, as do these other forums. There is nothing
like a meeting for getting people to all agree on something at the same
time.
> I do seem to remember that I enquired whether it were possible to add
> mpt, or another external participant, to the list and was told that it
> wasn't possible.
Your memory is a little unclear, as my messages to you of August 27th
and 28th clearly show, I added everyone you requested to the alias,
although it was not possible to add them all to the Netscape-internal
scheduling software. However, notices of the meetings were sent out to
the list in advance, and minutes were sent out after. If mozilla.org
has any sanctioned scheduling software, I'll use it. I have asked in
the past, in order to better schedule sheriff shifts, but was told it
was still being investigated.
Peter
Gervase Markham
>> testing of Mozilla, and by your definitions, only professional UI
>> designers (such as those employed by Netsape) are experts. You message
>> implied that "empirical data from usability tests, interpreted by
>> experts" was an important factor in deciding what the UI should be.
>
> It is, but I never gave any such definition, or said that Netscape was
> the only company that could do this, those are your words.
You and I both know that, in the near future, Netscape is the only company
who will be doing formal usability testing on Mozilla. And, given the
closed-ness of the data, Netscape's experts are the only people who ever
get to interpret it. Therefore, saying "empirical data from usability tests,
interpreted by experts" is the most important thing is equivalent to saying
that Netscape should decide what the UI is.
>> In which case, given that no-one has published any data, it's all
>> opinion. Does this mean the nod should go to whichever side is most
>> vocal?
>
> The data exists. The fact that it is not (yet) published does not
> negate its value.
This is like saying of some software: "the source code exists; the fact
that it is not published does not negate the value of the software."
This is a question of freedom in an open source project. My point is
that if opinion founded on hidden data should not be given weight; each
person should be able to make their own assessment of the same data.
> You seem to have no problem taking MPT's
> interpretation of observations at his web cafe at face value without
> having the original data published, why do you have such a problem with
> recommendations put forth by Netscape?
mpt does publish "original data"; for example, when we decided to unbind
Backspace from GO_BACK, he explained exactly what the problem was it was
causing for users in his experience. In general, he justifies his
recommendations in extensive (and sometimes tedious ;-) detail.
> I find this particularly
> puzzling since you are in fact yourself a Netscape employee, and have
> complete access to all of this data.
Merely because I am (currently) on the inside of the firewall does not
solve the problem; I am not the only person who has an interest in that
data. I can't exactly go into a bug and say "Actually, this
recommendation based on closed UI data is, in fact, right, but I can't
tell you why either."
>> I'm not portraying it as "us vs. them". However, I think it's becoming
>> more and more true that Netscape and Mozilla have different goals when
>> it comes to the UI; and the current situation seems to be that any
>> changes which Netscape's UE team does not agree with magically get
>> stuck somewhere before making it in; conversely, changes which are not
>> wanted by the community (and even some parts of Netscape - "them" vs.
>> "them", if you like) are forced in. I'm sure you know which two
>> particular bugs I'm thinking of.
>
> No, I don't know which bugs you mean.
In the first part, I'm talking about my proposed changes to the menu
bar, which I believe were supported by mpt, blake and ben. In the
second, I'm talking about View | Languages and Web Content, which was
opposed by roughly the same crew.
> Netscape as a company does have
> some goals for the UI that have no place in mozilla (*cough*
> Netscape.com *cough*), but I am convinced that the target end users are
> substantially the same.
Have a look at German's menu spec over at:
http://makeashorterlink.com/?K6CD5201
It seems clear to me that the reason that Netscape are pushing for a
top-level search menu is because they want to fill it out with entries
such as "Search Business" and "Search Reference Desk", and not because
the (denuded) form of the menu is actually good for Mozilla. Moreover,
to bolster the case and make the menu seem more credible, Find in This
Page and Find Again have been co-opted from the Edit menu, where they
are in a clear majority of other applications. (This means that we can't
solve the problem easily by making Search a commercial overlay.)
> If they are not, then we are wasting our time
> trying to share such a similar UI. I cite as example the fact that
> Netscape has been used for years by a large number of Linux users. Now
> that Mozilla is replacing Linux in RedHat and presumably elsewhere, it
> will be addressing the exact same userbase.
I would have thought that Linux users are (at least, not yet) at all
similar to Netscape's target user base, which seems (at least to me) to
be mostly less tech-savvy people on Windows. Are you saying Mozilla's
target userbase is the average Red Hat user?
>> Hand the documents over to me, and I'll see they get published. If you
>> feel it's important for people to be able to watch the tapes as well,
>> we'll see what we can arrange about that, too. We wouldn't want to let
>> people's business stand in the way :-) I have never seen a public
>> invitation for any mozilla.org participant to attend a Netscape
>> usability test.
>
> They are not mine to hand over. You have the same access to them that I
> do though, so I'm hoping that your offer to distribute them, or at least
> a non-proprietary subset, will be welcomed by the UE team.
I have made this offer before, and we agreed at Pixeljockies that it
should happen. What else can I do to make it possible? I can't just
publish stuff, as you know. Or can I? ;-)
>> "Internet cafe users" is about as broad a cross-section as you can get
>> - perhaps biased towards the less able end, where people don't have
>> computers at home or require a helping hand from an assistant. And he
>> sees hundreds of them using products every day in a natural setting,
>> whereas the usability tests I am aware of happen every six months or
>> so, with a dozen people, for a single day.
>
> You're welcome to your opinion, but I stand by my statement that such
> studies are "more subjective and anecdotal, and the users may be a
> self-selected class". I would encourage you to attend a professional
> usability study before criticizing them.
I think that the large number of users provides much better statistical
sampling. You are also likely to pick up smaller things that a shorter
test might not.
Anyway, my point was not that usability studies should be conducted in
Internet cafes; my point was that there are a wide variety of ways to
become knowledgeable about UI.
>> My point is not that one is better than the other; it is merely that
>> what I continually hear from German, you and others about how formal
>> usability testing is the only way to make decisions, and expert
>> interpretation is needed of the results, is bogus.
>
> What is bogus is your continued attempts to put words in my mouth.
> Please stop.
You are right, I shouldn't be putting words into your mouth. I apologise
that the "you" crept into that sentence. I was quoting German, as well
as I remember, at the first (or was it second) Pixeljockies meeting.
>> It's also possible that building the UI to the tastes of the masses,
>> whatever they are, could cause our extremely useful testing resource
>> to go away. This would be a great loss to the Mozilla project.
>
> Yes, that is a very real danger, which is why we need to be clear on who
> we are building this for. Is it just for the mozilla.org participants,
> or do we want this codebase to be widely used? At a time when about 93%
> of the people on earth have never used a computer, do we really want to
> base UI decisions on the personal preferences of the .0025% who
> currently download mozilla?
Again, you assume that all UI suggestions from this group are based on
personal preference.
>> As you and I both know, up to this point certain Netscape UE engineers
>> have asked not to be made UI module owners because they feel they
>> can't do the job for Netscape and also be responsive to the community.
>
> No, I don't know that. In what public forum did this happen?
I have been told this on several occasions by staff members of
mozilla.org. If it's not true, great :-)
>> Which is fine, but if they are attending Pixeljockies, then it's not a
>> meeting about Mozilla's UI, it's about the UI of Netscape's next
>> release. And it has therefore no authority to make Mozilla UI decisions.
>
> Who said it was? I pointed it out as an example of something _LIKE_
> what I thought was needed. There is no need to attack it, or criticize
> it for what it is not. However, if module owners get together in any
> forum and agree on a particular UI, then by your own rules they would
> have the authority.
Indeed; it remains that the Pixeljockies meetings that I have attended
have all been attended by only Netscape employees; this means that a few
vital people were missing for any far-reaching Mozilla UI decision to be
made.
>> Face-to-face meetings are inherently inconvenient in a global
>> organisation - and no record remains of what was said.
>> Newsgroup/email/Bugzilla discussions are much more suitable.
>
> Meetings have their place, as do these other forums. There is nothing
> like a meeting for getting people to all agree on something at the same
> time.
But, in this case, the lack of formal record means that things agreed on
are never acted on! Two examples from the first meeting:
1) the build system was going to change to permit small forks in the UI
by allowing, for example, Netscape to pull a different dynamic overlay
definition file (contents.rdf) to Mozilla
2) the raw usability data from the last usability study was going to be
published, along with German's interpretaton of it
>> I do seem to remember that I enquired whether it were possible to add
>> mpt, or another external participant, to the list and was told that it
>> wasn't possible.
>
> Your memory is a little unclear, as my messages to you of August 27th
> and 28th clearly show,
So it is. He can be (and now is) on the mailing list, but he can't be
automatically invited to the meetings.
Gerv
Peter Trudelle
You and I both know that, in the near future, Netscape is the only company
who will be doing formal usability testing on Mozilla. And, given the
closed-ness of the data, Netscape's experts are the only people who ever
get to interpret it. Therefore, saying "empirical data from usability tests,
interpreted by experts" is the most important thing is equivalent to saying
that Netscape should decide what the UI is.
This is like saying of some software: "the source code exists; the fact that it is not published does not negate the value of the software."
This is a question of freedom in an open source project. My point is that if opinion founded on hidden data should not be given weight; each person should be able to make their own assessment of the same data.
mpt does publish "original data"; for example, when we decided to unbind Backspace from GO_BACK, he explained exactly what the problem was it was causing for users in his experience. In general, he justifies his recommendations in extensive (and sometimes tedious ;-) detail.
In the first part, I'm talking about my proposed changes to the menu bar, which I believe were supported by mpt, blake and ben.
Have a look at German's menu spec over at:
http://makeashorterlink.com/?K6CD5201
It seems clear to me that the reason that Netscape are pushing for a top-level search menu is because they want to fill it out with entries such as "Search Business" and "Search Reference Desk", and not because the (denuded) form of the menu is actually good for Mozilla.
I would have thought that Linux users are (at least, not yet) at all similar to Netscape's target user base, which seems (at least to me) to be mostly less tech-savvy people on Windows. Are you saying Mozilla's target userbase is the average Red Hat user?
...if the Mozilla Organization doesn't produce a base UI which aims to be suitable for the largest possible number of users (rather than just the experienced ones), no-one will. And that can only be bad for the project as a whole.
I think that the large number of users provides much better statistical sampling. You are also likely to pick up smaller things that a shorter test might not.
Anyway, my point was not that usability studies should be conducted in Internet cafes; my point was that there are a wide variety of ways to become knowledgeable about UI.
But, in this case, the lack of formal record means that things agreed on are never acted on! Two examples from the first meeting:
1) the build system was going to change to permit small forks in the UI by allowing, for example, Netscape to pull a different dynamic overlay definition file (contents.rdf) to Mozilla
2) the raw usability data from the last usability study was going to be published, along with German's interpretaton of it
So it is. He can be (and now is) on the mailing list, but he can't be automatically invited to the meetings.
Peter
Gervase Markham
I'm sorry if any of my comments have sounded snappy. I am rather jaded
and frustrated on several UI issues (this is a reason, but not an
excuse.) When I calm down, a lot of what you say sounds more reasonable :-)
If I'm not careful, I am in danger of going round in increasingly irritated
circles. As something useful has come out of this discussion (see below) I
suggest we stop now :-)
>> But, in this case, the lack of formal record means that things agreed
>> on are never acted on! Two examples from the first meeting:
>> 1) the build system was going to change to permit small forks in the
>> UI by allowing, for example, Netscape to pull a different dynamic
>> overlay definition file (contents.rdf) to Mozilla
>> 2) the raw usability data from the last usability study was going to
>> be published, along with German's interpretaton of it
>
> Okay, now we're getting somewhere. The meeting notes and action items
> are distributed to the PJ list,
The meeting at which these two items were decided resulted in a
2-paragraph circulated summary, mentioning no action items, and neither
of the above points. And I know this is not your fault :-)
> so I think 'never' is premature. Why don't you put these items on the
> agenda for the next meeting, so we can follow-up?
An excellent idea. Thank you :-)
Gerv
Peter Trudelle
> I'm sorry if any of my comments have sounded snappy.
Aw, I know your heart is in the right place. I'm sorry for getting
emotional about this too. I guess we both just have deeply-held
feelings about the UI.
> I am rather jaded and frustrated on several UI issues (this is a
> reason, but not an excuse.)
Same here.
> If I'm not careful, I am in danger of going round in increasingly
> irritated
> circles. As something useful has come out of this discussion (see
> below) I
> suggest we stop now :-)
I agree. I suggest that, in addition to just stopping this thread, we
continue trying to move forward, and get a mutually satisfactory process
in place for resolving some of these longstanding issues. I know there
has in the past been some friction between various parties where the UI
is concerned, and that the current process still isn't working, but as
an old participant in a new role, I'm hoping we can make a fresh start.
thanks,
Peter
Tim Hunt
Do you follow what goes on in the performance newsgroup? Some group of
people (not quite sure who they are) hold weekly performance meetings
with the agenda and minutes published on the newsgroup. It makes it
clear that someone is keeping an overview of the various performance
issues, and chivvying people along and making things happen. Would a
similar idea work for UI?
Tim.
Matthew Thomas
>...
> At this point, the UI is functional, and nominally usable. Continuing
> on the subject of the target market of Mozilla, distributors, not end
> users, this type of UI is all that's needed.
No it's not. The objective of the Mozilla Organization (as far as I
know) is to get the Mozilla code used by as many people as possible. And
the more learnable and usable the human interface is for the base
Mozilla code, the more organizations will be willing to distribute it.
As long as Mozilla module owners persist in being content with interface
which is merely `functional and nominally usable', the Mozilla
Organization will not be carrying out its role as effectively as it
could be.
> The difficulty of
> changing the UI is not difficult at all to change, so any distributor
> that chooses to do research, and have it interpreted by "experts" (who
> very often times have no more idea of usability than anyone else, and
> also have a nasty tendency to make stupid decisions, more to follow)
> as much as their budget can tolerate.
That would be nice if it was true, but it's not. At the moment, there is
no Mozilla distributor (that I know of) willing to contribute the
resources and expertise necessary to produce a high-quality usable
interface for Mozilla. Sun and IBM have the expertise, but aren't
contributing the necessary resources. Netscape has the necessary
resources (or nearly so), but not the expertise. Those involved in the
Mozillafied version of the AOL client probably have both the resources
and the expertise, but they are hampered by a requirement to make as
much money as possible at the expense of usability (such that they might
well have their clock cleaned by MSN Explorer in the next few years).
> Further, UI experts are worth their weight in lead. UI experts are
> just as hit and miss as everyone else. The most "famous" UI expert
> research led to the lead balloon "Bob" which cost many millions of
> MS's research dollars.
That's not fair. Usability research was not to blame for Bob. He was
largely the fault of Bill Gates; if Gates hadn't been married to the
*marketing* person who was one of the main proponents of Bob and the
Office Assistant, we might well have never seen either of them.
<http://post-gazette.com/businessnews/19990523bob6.asp>
> UI experts are worthless when you have the
> direct input on hundreds of thousands of people who actually use the
> product and test it out daily.
>...
Direct input from hundreds of thousands of people can be useful for
designing user interface -- but only in the same way that a reflection
in a distorting mirror at a fun park can be useful for telling what
someone looks like. <http://www.useit.com/alertbox/20010805.html>
I saw an example of this when I got one of my flatmates to try out
Mozilla for the first time a couple of months ago, with me sitting
behind him resisting the urge to tell him how to do things. Previously,
I'd told him about the problem Mozilla still had where clicking in
Navigator's address field wouldn't select all the text. He'd told me he
actually preferred the click-to-place-caret behavior, so that he could
edit URLs without having to click twice. But when he actually came to
use Mozilla, he only used the address field for entering completely new
URLs, and he took about five seconds to select and delete all the text
in the field -- so long that he got mildly frustrated doing it.
--
Matthew `mpt' Thomas, Mozilla UI Design component default assignee thing
<http://mozilla.org/>
Matthew Thomas
>
> Peter Trudelle wrote:
> >
> > jesus X wrote:
> > > I'm assuming that they're there for the duration of Mozilla, since
> > > "it's not an end user product!" :)
> >
> > Its not?
>
> No. Do you really need to even ask this question after all this time?
>
> > Am I missing a wink here?
>
> A wink, no. A clue, apparently.
That was uncalled for. Peter's been away for a while, and hasn't yet
succumbed to the Mozilla Organization groupthink. Cut him a little slack.
> > Isn't mozilla the default browser on Red Hat Linux, and hopefully
> > other distros?
>
> It very well may be. This is a choice of the distributor, not Mozilla.
What is that supposed to mean? Should the Mozilla Organization
be discouraging Linux distributions from using Mozilla?
> I'm sure RedHat made it so, because of Chris Blizzard's fantastic
> involvement, and his being on their payroll as their Mozilla
> programmer.
I think you have it the wrong way round. Chris Blizzard's involvement
was surely a result, not a cause, of Red Hat's decision. I expect other
Linux distributions will set Mozilla as their default Web browser too,
and it will be nothing to do with whether or not any of their employees
happen to work on the code. It will be everything to do with Mozilla
being the most functional and compatible browser available on the platform.
> > Doesn't that represent the largest target userbase?
>
> Of Mozilla? No. Mozilla itself is being targeted at people who want to
> roll their own distro,
Which will inevitably be trivial in comparison with the number of Linux
users who use Mozilla and don't roll their own Mozilla distro. So the
answer to Peter's question is yes.
> much like the Linux kernel isn't targeted at
> John Q. Public but at Linux distros.
Given that Linux distros (and Linus himself) are constantly striving to
make the Linux kernel more useful for John Q. Public, that distinction
is pretty meaningless.
> > Or do you not consider them to be end users?
>
> Of course RedHat isn't the end user. That's silly.
Of course it's silly, it's not what Peter said. He said the Red Hat
*userbase*.
> That's like saying
> that car dealerships are the end users of cars. There are lot of
> people who use Mozilla directly from mozilla.org but that's not the
> target in mind.
And Mozilla's success will continue to be inversely proportional to the
prevalence of that attitude.
> I don't consider myself an end user of Mozilla so much
> as a contributor who takes binaries, beats on them, shakes out bugs,
> and tries to create helpful reports, confirm reports, etc.
>...
I don't consider myself an end user of Mozilla, but only because Mozilla
takes so long to do anything that I often choose to start up another
browser instead.
Matthew Thomas
>...
> > In which case, given that no-one has published any data, it's all
> > opinion. Does this mean the nod should go to whichever side is most
> > vocal?
>
> The data exists. The fact that it is not (yet) published does not
> negate its value. You seem to have no problem taking MPT's
> interpretation of observations at his web cafe at face value without
> having the original data published,
What original data? Would you like to buy me a camcorder so I can strap
it to my belt? :-)
> why do you have such a problem
> with recommendations put forth by Netscape? I find this particularly
> puzzling since you are in fact yourself a Netscape employee, and have
> complete access to all of this data.
I think the problem is that (like everyone else in n.p.m.ui) I don't
claim authority from usability studies, but instead explain why I think
a particular design is more usable. But some Netscape people don't
explain why they think a particular design is more usable; instead they
claim authority from usability studies, *and* they can't or won't
publish those studies.
>...
> > As you and I both know, up to this point certain Netscape UE
> > engineers have asked not to be made UI module owners because they
> > feel they can't do the job for Netscape and also be responsive to
> > the community.
>
> No, I don't know that. In what public forum did this happen?
It didn't happen in a public forum, but it was reasonable to expect that
you'd know about it since (IIRC) you're the manager of the person involved.
> > I do seem to remember that I enquired whether it were possible to
> > add mpt, or another external participant, to the list and was told
> > that it wasn't possible.
>
> Your memory is a little unclear, as my messages to you of August 27th
> and 28th clearly show, I added everyone you requested to the alias,
> although it was not possible to add them all to the Netscape-internal
> scheduling software. However, notices of the meetings were sent out
> to the list in advance, and minutes were sent out after.
>...
Just to clarify, it appears that I am on the Pixeljockeys mailing list,
since I'm getting messages from it where I'm not in the To: or CC: line.
I find it pretty useless, however, since practically all the messages on
it are of the form `The next meeting will be {yesterday} at {some time
when mpt was at work}', or the form `The last meeting was poorly
attended, so take a look at the {meeting minutes|discussion
document|Windows .EXE file} posted at <http://foobar/url/that/doesnt/work>'.
jesus X
> That was uncalled for. Peter's been away for a while, and hasn't yet
> succumbed to the Mozilla Organization groupthink. Cut him a little slack.
May be, I apologize for it then, but being out of the loop is no excuse, since
Mozilla has NEVER BEEN an end user product.
> What is that supposed to mean? Should the Mozilla Organization
> be discouraging Linux distributions from using Mozilla?
No. But because THEY choose to do so, that should not be confused with a choice
by mozilla.org to support hundreds of thousands of end users. Fortunately, most
Linux users are savvy enough to not expect shrink-wrap quality (well, what USED
to be shrink wrap quality, not MS's version of it) from a beta product.
> I think you have it the wrong way round. Chris Blizzard's involvement
> was surely a result, not a cause, of Red Hat's decision.
Oh, I agree. I'm just saying I'm sure the fact that they're paying an employee
to work on Mozilla keeps them fortified in the Mozilla camp. And I love them for
it. They recognize that it's in their best interest to have a top quality
browser platform (and extensible app platform) for Linux. After all, OSs don't
bring in users, apps do.
> > > Doesn't that represent the largest target userbase?
> > Of Mozilla? No. Mozilla itself is being targeted at people who want to
> > roll their own distro,
> Which will inevitably be trivial in comparison with the number of Linux
> users who use Mozilla and don't roll their own Mozilla distro. So the
> answer to Peter's question is yes.
The final answer, from what I can see and have seen is that Mozilla is not an
end user product in the sense that it has a high level support, distribution,
and marketing infrastructure. If those are key factors for a user, they should
go with a distro such as NS6.1 or Beonex, etc.
> > much like the Linux kernel isn't targeted at
> > John Q. Public but at Linux distros.
> Given that Linux distros (and Linus himself) are constantly striving to
> make the Linux kernel more useful for John Q. Public, that distinction
> is pretty meaningless.
Again, I'm drawing my lines along the support/distribution/marketing sections.
> > > Or do you not consider them to be end users?
> > Of course RedHat isn't the end user. That's silly.
> Of course it's silly, it's not what Peter said. He said the Red Hat
> *userbase*.
Then they are the end user of Redhat distros, which include Mozilla. I
misinterpreted.
> > I don't consider myself an end user of Mozilla so much
> > as a contributor who takes binaries, beats on them, shakes out bugs,
> > and tries to create helpful reports, confirm reports, etc.
> I don't consider myself an end user of Mozilla, but only because Mozilla
> takes so long to do anything that I often choose to start up another
> browser instead.
And yet you are a Module contact/owner for the UI and part of the UI team? Maybe
it's time you either retired from that post, or decided to grin and bear using
Mozilla to help increase your urge to help speed things up as much as you can.
Maybe I'm expecting too much, but hearing this from a major player is a but
disconcerting.
jesus X
> <http://post-gazette.com/businessnews/19990523bob6.asp>
Ok, Just because the source you used is my hometown newspaper, I'll give you
this one. ;)
Dylan Schiemann
--- jesus X <jes...@who.net> wrote:
> Matthew Thomas wrote:
> > I don't consider myself an end user of Mozilla,
> but only because Mozilla
> > takes so long to do anything that I often choose
> to start up another
> > browser instead.
>
> And yet you are a Module contact/owner for the UI
> and part of the UI team? Maybe
> it's time you either retired from that post, or
> decided to grin and bear using
> Mozilla to help increase your urge to help speed
> things up as much as you can.
> Maybe I'm expecting too much, but hearing this from
> a major player is a but
> disconcerting.
>
I was surprised to read this too. mozilla has been my
default browser on my primary home and work machines
for about a year now. Without using it all the time,
it isn't possible to find and report all of the bugs
and problems with the browser. However, from what I
have seen, Matthew does great work, and the mozilla ui
is far more usable than it was even a few months ago.
I find mozilla's performance is quite reasonable on my
Win2K Athlon 1 GHz 512 RAM system. It was terribly
too slow on a RedHat 7.1 Pentium MMX 233 64 MB RAM
system, but upgrading to 256 MB RAM made its
performance tolerable, and usable for casual browsing.
-Dylan Schiemann
http://www.sitepen.com/
http://www.dylanschiemann.com/
__________________________________________________
Do You Yahoo!?
Listen to your Yahoo! Mail messages from any phone.
http://phone.yahoo.com
Stuart Ballard
>
> > I don't consider myself an end user of Mozilla, but only because Mozilla
> > takes so long to do anything that I often choose to start up another
> > browser instead.
>
> And yet you are a Module contact/owner for the UI and part of the UI team? Maybe
> it's time you either retired from that post, or decided to grin and bear using
> Mozilla to help increase your urge to help speed things up as much as you can.
> Maybe I'm expecting too much, but hearing this from a major player is a but
> disconcerting.
<devil's advocate>
Actually it could be considered a valuable asset in a UI person. If he
used the browser every day, like you and I do, he'd be quickly blinded
to the UI deficiencies. Avoiding using the browser too much allows him
to come to it from an outsider's perspective and see the UI as a
newcomer might see it.
</devil's advocate>
- Stuart (who is an end-user of mozilla -- but only because I like the
flexibility and am fully capable of supporting myself, including making
tweaks to enhance Moz's usability to *me*, something certainly not true
of most end-users).
Gervase Markham
>>takes so long to do anything that I often choose to start up another
>>browser instead.
>
> And yet you are a Module contact/owner for the UI and part of the UI team? Maybe
> it's time you either retired from that post, or decided to grin and bear using
> Mozilla to help increase your urge to help speed things up as much as you can.
How exactly can a UI designer improve the speed of Mozilla? And how can
being continually frustrated by the slow speed improve the amount of
work he does?
> Maybe I'm expecting too much, but hearing this from a major player is a but
> disconcerting.
If you buy mpt a fast machine on which Mozilla performs acceptably, I'm
sure he'd be happy to run it more of the time.
Gerv
jesus X
> How exactly can a UI designer improve the speed of Mozilla? And how can
> being continually frustrated by the slow speed improve the amount of
> work he does?
>
> sure he'd be happy to run it more of the time.
My point was more along the lines of wondering how dedicated a person could be
who doesn't even use the product. Is this not a valid question?
Blake Ross
> netscape.public.mozilla.ui.
This is a newsgroup. st...@mozilla.org should be (and have been)
promoting this as a way to inform people of coming UI changes and let
them take part in shaping them, not as a place where a decision is made.
--Blake
Gervase Markham
> who doesn't even use the product. Is this not a valid question?
I personally measure mpt's dedication by the amount of bugmail he gets
and reads, the amount of bugs he comments intelligently in, and the
number of quality specs he produces.
Gerv
Neil
>I find mozilla's performance is quite reasonable on my Win2K Athlon 1 GHz 512 RAM system. It was terribly too slow on a RedHat 7.1 Pentium MMX 233 64 MB RAM system, but upgrading to 256 MB RAM made its performance tolerable, and usable for casual browsing.
>
And I thought running it on my Pentium 133 32 MB RAM was tolerable... I
would say Pentium II 266 128 MB RAM makes it usable. (This on Win9x
platforms).
Peter Trudelle
>Peter Trudelle wrote:
>
>
>What original data? Would you like to buy me a camcorder so I can strap
>it to my belt? :-)
>
You bet I would, imagine the never ending stream of usabilty data we'd have!
>I think the problem is that (like everyone else in n.p.m.ui) I don't
>claim authority from usability studies, but instead explain why I think
>a particular design is more usable. But some Netscape people don't
>explain why they think a particular design is more usable; instead they
>claim authority from usability studies, *and* they can't or won't
>publish those studies.
>
Maybe sometimes we don't really know why, we're just going on empirical
data? Perhaps if you or others had access to the tapes, you'd be able
to take the time to explain the reasons why some things fail and others
succeed.
>It didn't happen in a public forum, but it was reasonable to expect that
>you'd know about it since (IIRC) you're the manager of the person involved.
>
Well, I still don't know who we're talking about , but now I'm really
curious. I've never managed any UE people here, AFAIK.
Peter
Peter Trudelle
>...Mozilla takes so long to do anything that I often choose to start up another
>browser instead.
>
I hereby dedicate the Netscape Navigator team's performance work in the
MachV project to making Mozilla so ass-kicking fast that MPT starts to
feel this way about the other browsers.
Peter
Matthew Thomas
>
> Gervase Markham wrote:
>...
> > If you buy mpt a fast machine on which Mozilla performs acceptably,
> > I'm sure he'd be happy to run it more of the time.
>
> My point was more along the lines of wondering how dedicated a person
> could be
I generally spend about as much time doing Mozilla stuff as I spend
working at my day-job. Is that sufficient, or would you like me to give
up eating and sleeping as well? :-)
> who doesn't even use the product. Is this not a valid
> question?
>...
No, for two reasons. Firstly, because it is not true that I `don't even
use the product'; I said `I often choose to start up another browser
instead'. About 50 percent of my browsing is done with MSIE, about 45
percent with Mozilla, and about 5 percent with iCab. (Before Dave
Hyatt's style matching rewrite, the percentages were about 75 % for
MSIE, 15 % for Mozilla, and 10 % for iCab.)
And secondly because late last year, there was a period of about a month
during which I was unable to run Mozilla at all (because of new access
restrictions on the computers at university), but I continued merrily on
giving UI advice regardless. Hardly anybody noticed.
Matthew Thomas
>
> Matthew Thomas wrote:
> >
> > That was uncalled for. Peter's been away for a while, and hasn't yet
> > succumbed to the Mozilla Organization groupthink. Cut him a little
> > slack.
>
> May be, I apologize for it then, but being out of the loop is no
> excuse, since Mozilla has NEVER BEEN an end user product.
Red Hat, Mandrake, et al. seem to think it is. They're obviously out of
the loop too. Who's going to tell them the bad news?
> > What is that supposed to mean? Should the Mozilla Organization
> > be discouraging Linux distributions from using Mozilla?
>
> No. But because THEY choose to do so, that should not be confused with
> a choice by mozilla.org to support hundreds of thousands of end users.
Yes, whether or not Mozilla is an end user product should not be
confused with whether or not the Mozilla Organization provides support
for it. They're completely separate issues, so I'm wondering why you
even brought that up.
> Fortunately, most Linux users are savvy enough to not expect
> shrink-wrap quality (well, what USED to be shrink wrap quality, not
> MS's version of it) from a beta product.
s/savvy/bloody-minded/
s/beta/distributed/
:->
>...
> Oh, I agree. I'm just saying I'm sure the fact that they're paying an
> employee to work on Mozilla keeps them fortified in the Mozilla camp.
> And I love them for it. They recognize that it's in their best
> interest to have a top quality browser platform (and extensible app
> platform) for Linux. After all, OSs don't bring in users, apps do.
Actually, I think Red Hat recognize that it's in their best interest to
have a top-quality *browser* for Linux. Linux has quite enough
`extensible app platforms' already. After all, `platforms' don't bring
in users, apps do.
>...
> The final answer, from what I can see and have seen is that Mozilla is
> not an end user product in the sense that it has a high level support,
> distribution, and marketing infrastructure. If those are key factors
> for a user, they should go with a distro such as NS6.1 or Beonex, etc.
Having high level support, distribution, and marketing infrastructure is
not a property of end user products; it is a property of paid-for
products. These are intersecting sets, but not equal sets. In an
open-source project, users tend to do most of the support and marketing
themselves, so whether other organizations exist to do those things too
is irrelevant to whether the project is an end-user application.
>...
> > I don't consider myself an end user of Mozilla, but only because
> > Mozilla takes so long to do anything that I often choose to start up
> > another browser instead.
>
> And yet you are a Module contact/owner for the UI
No, I'm default assignee thing for the UI Design component in Bugzilla.
I don't own any modules.
> and part of the UI
> team?
If such a UI team exists, I have not had the pleasure of making their acquaintance.
> Maybe it's time you either retired from that post, or decided to
> grin and bear using Mozilla to help increase your urge to help speed
> things up as much as you can.
I have absolutely no knowledge on how to speed up the Mozilla code. I
wouldn't even know where to look. (Nor does anybody know, apparently,
due to a lack of profiling tools for Mac OS.)
And grinning and bearing using Mozilla full-time wouldn't help me
improve its human interface design; on the contrary, it would make me
less productive. Consider the following numbers (fairly unscientific,
but typical of my experience); comparison numbers for other browsers are
provided to show that the fault is with Mozilla, not with my computer.
iCab Netscape MSIE Mozilla
Task 2.51 4.78 5.0 2001100108
----------------------------------------------------------------------
Load <http://slashdot.org/>:
16 seconds 11 seconds 18 seconds 7 seconds
Load <http://doc.weblogs.com/>:
26 seconds 55 seconds 38 seconds 48 seconds
Load <http://scripting.com/>:
53 seconds 79 seconds 46 seconds 41 seconds
Start up (time from launch to first window displaying <about:blank>):
5 seconds 5 seconds 8 seconds 25 seconds
Open 20 windows <http://bugzilla.mozilla.org/attachment.cgi?id=47748>:
test didn't work 11 seconds 9 seconds 91 seconds
Average time to open a window, calculated from the above:
-- 0.5 seconds 0.5 seconds 4.6 seconds
Close the second of two open windows:
~0.4 seconds ~0.3 seconds ~0.4 seconds ~1.2 seconds
As you can see, Mozilla is generally faster than other browsers at
rendering pages, but most of my time using Mozilla is spent doing things
other than rendering pages -- opening and closing windows, dialogs, and
menus, scrolling through pages, and redrawing windows after they've been
covered by another app. And for these tasks, Mozilla is excruciatingly
slow. For example, in each of the page load figures for Mozilla shown
above, fast though they are, about two seconds of the total is spent
closing the URL auto-complete menu.
> Maybe I'm expecting too much, but
> hearing this from a major player is a but disconcerting.
>...
It's disconcerting for me too. I've been waiting patiently for two years
for Mozilla to become fast enough for everyday use -- for it to meet
basic usability requirements such as opening a new window within 1
second so that the user won't get distracted from their task. Big
improvements have been made, but it would appear that we're still months
or years away from that goal.
And while we're not there yet, Mozilla is immensely frustrating to use.
I don't want to wait between 3 and 30 seconds (depending on how many
other apps I have running) for a new browser window to open. I simply
have better things to do.
jesus X
> > and part of the UI team?
> If such a UI team exists, I have not had the pleasure of making their
> acquaintance.
I was informed there is a group of people involved with UI issues. If I'm wrong,
so be it.
Henri Sivonen
dyl...@yahoo.com (Dylan Schiemann) wrote:
> Matthew Thomas wrote:
> > I don't consider myself an end user of Mozilla,
> but only because Mozilla
> > takes so long to do anything that I often choose
> to start up another
> > browser instead.
> I was surprised to read this too.
I wasn't. Mozilla starts up very slowly compared to Mac IE 5 on Mac OS
Classic. Mozilla also has higher requirements than Mac IE 5.
Interestingly, on Mac OS X 10.1 Mozilla starts up faster than Mac IE
5.1. However, it is Mac IE 5.1 (and other apps) starting up slowly--not
Mozilla starting up fast. (With several other apps already running on a
533 MHz G4, IE took 66 seconds to start. Mozilla took "only" 45 seconds.
On the same machine Mac IE 5.0 starts up on the Classic virtual machine
in less than 10 seconds. When using Mac OS 9.x only, Mac IE 5.0 would
start up even faster.)
> Without using it all the time, it isn't possible to find and
> report all of the bugs and problems with the browser.
Back when I had a PowerCenter Pro 210 with 48 MB of RAM, I was quite
able to report Mozilla bugs even though it was so slow that I used Mac
IE 5 most of the time.
--
Henri Sivonen
hen...@clinet.fi
http://www.clinet.fi/~henris/
Asa Dotzler
> jesus X wrote:
>
>> Ben Ruppel wrote:
>>
>>> I think Help isn't last for debug purposes. Debug and QA can later be
>>> removed without messing with stuff between help and the rest of the
>>> items.
>>>
>>
>> I'm assuming that they're there for the duration of Mozilla, since
>> "it's not an
>> end user product!" :)
>>
> Its not? Am I missing a wink here? Isn't mozilla the default browser
> on Red Hat Linux, and hopefully other distros? Doesn't that represent
> the largest target userbase? Or do you not consider them to be end users?
>
> Peter
>
I'm not positive but I'm pretty sure that Chris does up a special build for
RedHat that has these menu items removed. Making changes to the UI and
packaging is what I call "building a distribution". Netscape does this with
Mozilla code too. They put quite a bit more in with IM, net2phone, spellchecker,
web/aol mail integration, etc. but the basic idea is the same.
--Asa
Asa Dotzler
> jesus X wrote:
>
>>Matthew Thomas wrote:
>>
>>>That was uncalled for. Peter's been away for a while, and hasn't yet
>>>succumbed to the Mozilla Organization groupthink. Cut him a little
>>>slack.
>>>
>>May be, I apologize for it then, but being out of the loop is no
>>excuse, since Mozilla has NEVER BEEN an end user product.
>>
>
> Red Hat, Mandrake, et al. seem to think it is. They're obviously out of
> the loop too. Who's going to tell them the bad news?
Red Hat doesn't distribute a mozilla.org binary. They modify the UI and
make special packaging. It is a Red Hat binary, not a mozilla.org
testing binary. This is just a lesser degree of the same thing that
Netscape does with its distribution, Netscape 6.whatever.
--Asa
Peter Trudelle
> I'm not positive but I'm pretty sure that Chris does up a special
> build for
> RedHat that has these menu items removed. Making changes to the UI and
> packaging is what I call "building a distribution". Netscape does this
> with
> Mozilla code too. They put quite a bit more in with IM, net2phone,
> spellchecker,
> web/aol mail integration, etc. but the basic idea is the same.
Okay, so they aren't the exact same bits, they remove the Debug and QA
menus, and perhaps change a few other things. I'll concede that the
Mozilla binaries that contain these items are not intended for end
users, but I'm not sure if you mean to imply that therefore Mozilla is
not targetted at end users. The browser RedHat ships is still
substantially Mozilla. I don't have that version of RH yet, but a quick
look at their website leads me to believe they don't rebrand it or
change it functionally in any significant way.
My main point is that typical end users are getting Mozilla. Some
distributors may add a lot to it, some may just remove the
testing/debugging UI. I still think we should optimize for these users,
since they represent the largest possible growth in the userbase, and
hence the biggest opportunities for mozilla.org developers.
Changes/additions that appeal only to techies and other small groups
should be considered, but should not be allowed to make it difficult for
Mozilla and derivative works to be accepted and used by the masses. A
common denominator browser may not sound sexy to everyone, but if we
could make it simple, fast and stable, it could be popular with end
users, as well as make an excellent base for distributors to build more
specialized products on. Mozilla.org could itself use it as a base for
building a killer high-end browser, tuned for the most demanding mozillians.
Peter
Asa Dotzler
I have no substantial disagreements with this post. I guess I could have
just said "the debug and qa menus aren't going away in mozilla.org
testing binaries" and that would bave been enough :)
--Asa