[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-usability
Subject:    Re: Idea for KDE 4 - global, smooth zoom feature.
From:       "Aaron J. Seigo" <aseigo () kde ! org>
Date:       2005-05-26 15:18:41
Message-ID: 200505260918.51790.aseigo () kde ! org
[Download RAW message or body]

[Attachment #2 (multipart/signed)]


On Thursday 26 May 2005 04:55, Diego Moya (a.k.a. TuringTest) wrote:
> > if an embedded document has scrollbars, where is it scrolling?
>
> You don't need scrollbars, you have infinite space to represent the
> whole document at once.

an embedded document is usually constrained within a space for layout reasons. 
think of a spreadsheet table in a word processing document. framed layouts 
are fairly commonplace and actually useful. unless you decide to completely 
redefine how compound documents are handle and enforce a "it either fits 
completely or doesn't fit at all" method, you get scrolling.

> See also Overview beside Detail and the 
> Fisheye representation pattern, fisheye menus, etc:

neither of which helps here.

> > 2D metaphores fail at anything more complex than a sheet of paper.
>
> I can't agree with that. You usually don't use infinite zooming with
> paper sheets.

i didn't say you did. that wasn't the point. showing the totality of an item 
isn't always desirable, even if you CAN zoom infinitely. constraints on 
viewable areas are often specifically desired to maintain ratios between 
navigation and content or between areas of content or because of physical 
output considerations.

> See the Expand-in-context representation pattern:

expand in context is a great idea that is not reliant on zooming interfaces 
like raskin's ZUI and which works quite well with our current windowing 
systems.

> An efficient interface for accessing to things when you actually *do*
> remember where they are. Actually, an interface designed to make
> places easier to remember.
>
> In a search interface you can't see things in advance. You start with
> an empty screen and you have to recall the object properties from
> memory.

actually, this isn't true at all. this is a common misconception which i've 
dispelled many, many times on this list. you only need a starting point or to 
be "near enough"

> > all you've shown is that you can fix the problems with spatial by relying
> > on a non-spatial solution. um. heh.
>
> Anything wrong with that? A ZUI is not intended to turn ALL
> information retrieval processes into a "slowly move to your
> destination", you can very well have instant access

> (think again of 
> Firefox incremental text search - that's both spatial and efficient).

how is it spatial, exactly? and the only efficiency it gains is that it 
doesn't cover the results, which is great, but hardly to do with spatial 
representation of information.

> See also the Continuous Highlight patternt for search:
> http://www.cs.helsinki.fi/u/salaakso/patterns/Continuous-Highlight.html

yes, for search interfaces this is quite useful. the relationship to spatial 
representations of objects is nascent.

> But you can also have a non spatial "find files" Continuous-Filter

which doesn't require changing the entire desktop mataphore. "you can fix the 
problems of the infinite plane the same ways you can fix our current 
problems"... then why change the current metaphore?

> The ZUI is a better replacement for the current spatial paradigm (a
> desktop with windows) which we agree is quite broken nowadays - it
> doesn't work well for huge information volumes.

i don't believe its broken for reasons of spatial representation, however. and 
i believe this is easily demonstrated and has been many times over. we have 
the web and libraries to look at for two examples.

> ZUI follows the
> Schneiderman's mantra of "overview first, zoom and filter, then
> details on demand", the desktop doesn't. Remember the

overview of what? you don't need, nor really want, an infinite plane of 
everything for overview first. it's a really great way to show the results of 
specific contexts, but to provide a global view or to try and present _a_ 
representation of things ... no.

> > agreed; yet this is not solved by making things spatial, it's solved by
> > making things self-locating.
>
> Self-locating is the same as "I'm forced to use search to reach it,
> even if I used the object ten minutes ago"?

no. if you used it 10 minutes ago, chances are good that unless you took some 
specific action, it's still nearby. if it isn't nearby, it should be easy to 
recall. "Recently Opened Documents" is a good example of this for 
applications that deal in documents.

> Actually, spatial is an additional attribute which you can use for
> search. "Find all the web pages that I opened yesterday near the KDE
> mailing list".

this is a contextual link, not a spatial one. you can map this in terms of 
spatial coordinates, but that's a mapping only, and one that doesn't scale 
(e.g. when you have 1000 web pages associated with your KDE list) nor one 
that is inherently better than a simple listing once you have defined your 
parameters. spatial is a way to encode search parameters. if you aren't using 
it to encode pathways, don't bother. so: how do you define a spatial search 
for "web pages i opened yesterday nere here" without optimizing for that 
particular search?

i have no problems with using spatial mappings where useful (e.g. calendars), 
i do have a problem with making it the central metaphore since it doesn't map 
consistently well.

> > make an infinite space to zoom around in and you STILL
> > need means to locate things. it's not easier enough given the normal
> > amount of information and how we use it. we need to make organization
> > conceptual and
>
> Normal users (i.e. anyone who is not a programmer) have serious
> problems when everything is conceptual. That's a fact in cognitive
> science. You can't throw at them layers and layers of abstraction
> mappings.

listen to two people talk. read our mythologies and oral histories. we are 
very, very good at imprecisely communicating concepts. watch how people 
describe what they are doing when they go looking for something in their 
computer. our linguistic capabilities and the ability to tell stories 
outstrips our spatial capabilities many times over. this is conceptual and 
has nothing to do with abstractions or mappings. 

> > get rid of all these bad metaphores to the physical world. because only
> > once we make organization conceptual can we make items self-locatable and
> > self-organizing.
>
> No, you replace bad metaphors with better ones. The brain is hardwired
> to use physical information,

limited amounts of physical information. LIMITED. this is what just annoys the 
hell out of me when discussing things with "spacionaughts" is that they 
conveniently ignore the fact that these evolutionary traits are optimized for 
dealing with a handful of items at once, and that the entire point of 
computers in the first place was (and is) to get us beyond that point. our 
physical information processing capacity is amazingly limited and highly 
environmentally reactive (versus useful for proactive organization; think of 
motion detection =)

> just current interfaces don't take 
> advantage of that and force you to think of every detail of the
> desired task in advance.

our brain is also hardwired to tell stories, and current interfaces don't take 
advantage of that. 

> > > In a ZUI you see everything at once
> >
> > this is exactly my point: we have too much to make "seeing everything at
> > once" useful. moreover, "everything" is no longer files, it's also the
> > relationships between files and the temporal nature of things (e.g.
> > presence). the world has moved on since 1984.
>
> Then bring several spatial representations to see those relationships,
> instead of forcing them to be only in the user's mind. Temporal is
> solved with "Stuff I've Seen" and Picasa's "a big vertical folder with
> photos" (time2space representation), hyperspace is solved with graphs
> and/or facets. All of this is compatible with a ZUI: "Overview first,
> zoom and filter, then details-on-demand".

> > > - except if it is very small, but
> > > then you can see -and visually remember- the surrounding context. And
> >
> > no, actually, i can't remember it, visually or not.
>
> Trust me, you can - at least for important things, the ones on which
> you have a habit formed. You know where the rooms in your house are
> located, the bus station respect to your house, the important
> buildings in your city. For finding the forgotten details, you have
> search, indexes and maps.

yes, i can remember quite a few spatial items. those items are dwarfed by 
several orders of magnitude by the discreet units of information in my 
computers. this is a trend that is only quickening, btw.

> You don't need to remember, you can see them from here - everything is
> exposed.

i'm not sure how to get this through to you: i have too many things to make 
this practical. the vast majority of computer users have too many things to 
make this practical.

> If they're too small to actually see them, use search and it 
> will draw a big arrow pointing to the keys (and even a "zap" button to
> move there instantly). Isn't that conceptually better that just
> magically appearing them from nowhere?

no. google works just fine, thank you. and i don't really care where something 
came from when google returns a result. it's irrelevant.

> > and to say that the answer to the hierarchy is to smear it all out into
> > one directory is, honestly, insane. go ahead, put all your files in one
> > directory and zoom in an out. yeah, that's useful. =(
>
> You can build hierarchies in a plane. Put things inside nested
> rectangles. Have you seen Treemaps? For some things they're better
> than the expandable tree widget. For the others, you still *can* use
> trees (graphs).

of course i've seen treemaps. and outside of very specific tasks (primarily 
ones involving the spatial aspects of files: e.g. size on disk) they aren't 
particularly useful for file management. beyond size and relative position, 
their information density is very low.

moreover, the problem with hierarchies is not representation.

> > > Actually, can you think of *anything* that you can do on a desktop and
> > > you can't do on an infinite desktop? (other than hide things under
> > > other things and lose them from view).
> >
> > that's not the relevant point here. the relevant question is how usable
> > it is.
>
> Agreed. Why do you assume it's worse? It's because the previous
> spatial interfaces that you have seen? ZUI is not commonplace, you
> don't know yet what can be done with it.

i've been following as best i can the humane interface project's results and 
progress. while calling me ignorant may be a fun and easy reponse to muster, 
and while it is quite true that "ZUI"s have not been played with as much as 
possible, i'm not sure what that has to do with the fairly obvious challenge 
points there are to making the primary desktop metaphore a ZUI.

> In his book it reports a success with a real system: a timetabling
> application for several hospitals. Nurses were proficient with the
> system in just several minutes.

first, calendaring is perfectly suited to spatial organization for what should 
be pretty obvious reasons. it's also not representative of most desktop 
tasks. a zooming interface is certainly good for certain things (like a city 
map, an organizational chart or a calendar), but discussions of making it the 
central desktop metaphore ignore that we don't use our computers exclusively 
or even primarily for these types of activities.

> Also the Canon Cat was designed and optimized for managing all kinds
> of textual information. Yes it wasn't a big commercial success, but
> the world has moved on since 1984, hasn't it?

i'm not sure what your point is here, other than to drag another Raskin 
failure into the conversation.

> Actually isn't Emacs exactly what are you advocating for with your
> "get rid of spatial representation and bring more dimensions to the
> mix!"?. 

no, it's not. not in the least.

here: i'd love to see what you can do with a ZUI as the primary desktop 
interface. i've stated why i think it will fall on its face. i don't expect 
to convince you of that, though. please prove me wrong. 

in fact, you know what i'd LOVE to see happen? i'd love to see people 
interested in spatial representations stop trying to wedge it into places it 
doesn't fit (e.g. the desktop as a whole) and start working with the places 
that it would be a PERFECT fit for within the desktop, e.g. search 
interfaces, calendaring, etc...

the biggest mistake so many spatialites make is to head for the file manager 
when that's the least compelling argument they could hope to make.

-- 
Aaron J. Seigo
GPG Fingerprint: 8B8B 2209 0C6F 7C47 B1EA  EE75 D6B7 2EB1 A7F1 DB43

[Attachment #5 (application/pgp-signature)]

_______________________________________________
kde-usability mailing list
kde-usability@kde.org
https://mail.kde.org/mailman/listinfo/kde-usability


[prev in list] [next in list] [prev in thread] [next in thread]