Lars-Erik Østerud wrote: > Why couldn't plain and easy stuff be left as options, instead of > forcing users to write and install add-ons to make things simpler?
Because one of Firefox's principles is to NOT add a zillion options in attempt to please everybody.
I was initially concerned about the new url autocomplete-menu format, too. It took a little time to adjust to, but I think it's a clear win now.
>> Why couldn't plain and easy stuff be left as options, instead of >> forcing users to write and install add-ons to make things simpler?
> Because one of Firefox's principles is to NOT add a zillion options in > attempt to please everybody.
> I was initially concerned about the new url autocomplete-menu format, > too. It took a little time to adjust to, but I think it's a clear win now.
> Justin
Just as an arbitrary data point, I'm not. Right now, it's still in the way of me getting to the locations I want. I type in, forcing me to ignore the suggestions, and then at some point, I take a deep breath, and start investigating what's displayed.
> Just as an arbitrary data point, I'm not. Right now, it's still in the > way of me getting to the locations I want. I type in, forcing me to > ignore the suggestions, and then at some point, I take a deep breath, > and start investigating what's displayed.
Is that the matching algorithm or the display?
I find your arbitrary data point interesting (and saddening) but not particularly useful in helping me understand what aspect of the design is getting in your way. I crave more detailed introspection!
> It works with tonight's Minefield, but not with yesterday's Fx3.0b2 release, > even though it's indicated as working with Firefox: 3.0b2pre 3.0b3pre
Good catch.
With 3.0b2, you can still use the browser.urlbar.richResults pref.
Instead of marking the matches with BOLD + underline (looks a bit ugly I think), how about marking it with a different background color - just like with "find" ? -- Lars-Erik - http://www.osterud.name - ICQ 7297605 My Firefox tweaks: http://osterud.name/firefox.html
Mike Beltzner wrote: > On 19-Dec-07, at 6:19 PM, Axel Hecht wrote:
>> Just as an arbitrary data point, I'm not. Right now, it's still in the >> way of me getting to the locations I want. I type in, forcing me to >> ignore the suggestions, and then at some point, I take a deep breath, >> and start investigating what's displayed.
> Is that the matching algorithm or the display?
> I find your arbitrary data point interesting (and saddening) but not > particularly useful in helping me understand what aspect of the design > is getting in your way. I crave more detailed introspection!
I think there are two aspects. Probably most prominently, I'm surfing via URLs, and those get de-emphasized. So I need to look harder to actually find what I know. The second piece that I find disturbing is that there seems to be a concurrence between me actually consiously scanning the suggestions and some timeout that gives me more hits. That is, as soon as I start reading, the list changes.
I'm having a "too much data" problem, too, I guess. That might come from not seeding it enough, or the search not being good enough yet. Just played with it, http://flickr.com/photos/axelhecht/2124897766/, for "bug", 2 out of 10 links have useful data in the display. It took me a bit to figure out that the litmus links are in there because those queries have a "withbugs=all" query param in them.
Oh, and are we going to release note "use a different profile for porn"? ;-)
> Mike Beltzner wrote: >> On 19-Dec-07, at 6:19 PM, Axel Hecht wrote:
>>> Just as an arbitrary data point, I'm not. Right now, it's still in >>> the >>> way of me getting to the locations I want. I type in, forcing me to >>> ignore the suggestions, and then at some point, I take a deep >>> breath, >>> and start investigating what's displayed.
>> Is that the matching algorithm or the display?
>> I find your arbitrary data point interesting (and saddening) but not >> particularly useful in helping me understand what aspect of the >> design >> is getting in your way. I crave more detailed introspection!
> I think there are two aspects. Probably most prominently, I'm surfing > via URLs, and those get de-emphasized. So I need to look harder to > actually find what I know. The second piece that I find disturbing is > that there seems to be a concurrence between me actually consiously > scanning the suggestions and some timeout that gives me more hits. > That > is, as soon as I start reading, the list changes.
OK, I've heard this before, and am working on some ideas to see if we can make it easier to see the URLs. The design goal here is actually to make it easier to pick URLs out from this list (since one can see more of them than before, and they are painted green, as they are in search results) so it's certainly not good that we're interfering with that.
> I'm having a "too much data" problem, too, I guess. That might come > from > not seeding it enough, or the search not being good enough yet. Just > played with it, http://flickr.com/photos/axelhecht/2124897766/, for > "bug", 2 out of 10 links have useful data in the display. It took me a > bit to figure out that the litmus links are in there because those > queries have a "withbugs=all" query param in them.
As far as I can tell, that would happen the same way with the less rich results, the only difference being that you might not be able to see the matching part of the URL because part of the horizontal space would be given to a snippet of the page title.
Not matching against query params is an interesting optimization, though; I think we considered it, but realized that sometimes it would be useful.
> Oh, and are we going to release note "use a different profile for > porn"? ;-)
We never have before, I don't see why we'd start now. :)
On 2007-12-20, Mike Beltzner <beltz...@mozilla.com> wrote:
> On 19-Dec-07, at 6:19 PM, Axel Hecht wrote:
>> Just as an arbitrary data point, I'm not. Right now, it's still in the >> way of me getting to the locations I want. I type in, forcing me to >> ignore the suggestions, and then at some point, I take a deep breath, >> and start investigating what's displayed.
> Is that the matching algorithm or the display?
As another arbitrary data point, I have the beta installed on my computer at home, and a friend's first reaction on seeing the new drop down was that the huge text was some sort of accessibility feature for people that couldn't read normal size text. I agree that the larger-than-normal text looks a bit weird and clunky.
As for the matching algorithm - it seems pretty good, but is not so great when you've only typed 1 or 2 letters, so I wonder if flashing up a whole load of stuff is useful enough to counter the rather distracting effect of it.
Going in the other direction from simplicity, I feel there is some merit in Daniel Glazman's thoughts about customisation - http://www.glazman.org/weblog/dotclear/index.php?post/2007/12/20/Fire... Quite often I do actually know whether I want a site that I've visited a lot or a site that I've visited recently or something that I've previously bookmarked. If, when faced with a lot of results, I could somehow tweak the matching as I go (by doing something more than typing more letters to be a found together), that might be good. But I guess that is something for an extension...
> Going in the other direction from simplicity, I feel there is some merit
> in Daniel Glazman's thoughts about customisation - > http://www.glazman.org/weblog/dotclear/index.php?post/2007/12/20/Fire... > Quite often I do actually know whether I want a site that I've visited a > lot or a site that I've visited recently or something that I've previously > bookmarked. If, when faced with a lot of results, I could somehow tweak > the matching as I go (by doing something more than typing more letters to > be a found together), that might be good. But I guess that is something > for an extension...
Edward Lee is working on what he calls "adaptive learning"[1] which accomplishes exactly what you are asking for. As far as I know, it is scheduled to be included in Firefox 3.
Adam Kowalczyk wrote: > Michael Lefevre wrote: > > Going in the other direction from simplicity, I feel there is some merit >> in Daniel Glazman's thoughts about customisation - >> http://www.glazman.org/weblog/dotclear/index.php?post/2007/12/20/Fire... >> Quite often I do actually know whether I want a site that I've visited a >> lot or a site that I've visited recently or something that I've previously >> bookmarked. If, when faced with a lot of results, I could somehow tweak >> the matching as I go (by doing something more than typing more letters to >> be a found together), that might be good. But I guess that is something >> for an extension...
I've heard a lot of people ask for deep customization here, but I think that's best for extension fodder at the moment. We've been working hard to keep the performance impact down to a minimum, and it feels like adding these alternate paths might affect that. Seth might know more.
I will be posting to my blog today or tomorrow with some userChrome.css hacks that people can use to change the colours, though, so that we can get some alternate ideas generated.
I do, otoh, like Daniel's idea for Shift-Delete removing a single item from the history; anyone know if there's a bug on that?
> Edward Lee is working on what he calls "adaptive learning"[1] which > accomplishes exactly what you are asking for. As far as I know, it is > scheduled to be included in Firefox 3.
Indeed, changes to the matching algorithms are in the works, and I think there's a lot more we can do to make sure that the intelligent results are truly intelligent.
> OK, I've heard this before, and am working on some ideas to see if we > can make it easier to see the URLs. The design goal here is actually > to make it easier to pick URLs out from this list (since one can see
Enabling the old style with an option to not display the "title" would do that (a plain drop-down with the icon and the URL - no more).
Make it an option somewhere.
I had allready in my "userChrome.css" for 2.0 made the "title" field narrower (you can even remove it), and I have found a way to remove the "title" and adjust the position of the URL in the "rich list" as well using "userChrome.css" (and negative margins and stuff).
So if you don't include these options, they will still be there, but harder to get (more tech knowledge) for the non-tech users. -- Lars-Erik - http://www.osterud.name - ICQ 7297605 My Firefox tweaks: http://osterud.name/firefox.html
Michael Lefevre wrote: > As another arbitrary data point, I have the beta installed on my computer > at home, and a friend's first reaction on seeing the new drop down was
How will this new two-line complicated drop-down work for those using tools for blind (reading lists) or poor-sight (high contrast, huge font). I would think the old style was much better for those? -- Lars-Erik - http://www.osterud.name - ICQ 7297605 My Firefox tweaks: http://osterud.name/firefox.html
Lars-Erik Østerud wrote: > Make it an option somewhere.
Adding an option is not free; it adds code bloat and creates additional testing paths, and in cases where the option is user facing, it adds information that the user must parse when looking for a certain preference or option that they actually care about.
So far I haven't heard a convincing argument about why this needs to be an option as opposed to an add-on.
> I had allready in my "userChrome.css" for 2.0 made the "title" field > narrower (you can even remove it), and I have found a way to remove > the "title" and adjust the position of the URL in the "rich list" as > well using "userChrome.css" (and negative margins and stuff).
> So if you don't include these options, they will still be there, > but harder to get (more tech knowledge) for the non-tech users.
Non-tech users don't care about URLs as much as they care about page titles.
Mike Beltzner wrote: > Non-tech users don't care about URLs as much as they care about page > titles.
I don't think that's strictly true, FWIW. I would agree that they don't usually understand the entire structure of URLs. However, they are certainly familiar with domains, nowadays when many printed adverts include website addresses and the phrase "visit goats-insurance dot com" attacks us a dozen times a day. I see plenty of people calling websites by their domains. I suppose this happens especially in cases of websites which don't have clear, memorable, or concise names, which makes it easier to just use their domains instead.
Based on these observations, I think that non-tech users type domains in their Location Bars all the time.
