Re: [Usability] User problems and practices with modern desktopsystems

[<john erling blad aftenposten no>]

> > A journalist might write 10-20-30 articles a day (or even more), or
> > something like 2 - 10 000 articles a year. It will soon be apparent
that
> > you can't continue to dump articles in a single directory for very
long.
> 
> This reminds my why I am so glad that we have a standard system for
> Recent files it goes a long way to taking the sting out of finding
> documents as in most cases the last documents you used are most likely
to
> be ones you will want to use again.

This is where thing gets a bit fuzzy. It is not the most recent files
you are most likely to try to find. It is the files you don't know about
or those that falsify your claims or creates a contradictionary
statement.

Imagine someone making a political claim, "nuclear power is harmless",
while he 3 years ago was imprisoned for demonstrating for some
environment activist outside a nuclear power plant. Note that this is an
example. You could be writing a howto, not noting that someone already
wrote something about the same stuff.

This problem is fairly easy to tackle by automatic query generators that
inspects the document you are working on and generates a proper query
for a search engine from that. There are several commercial systems that
do something like that with varying degree of success.

> What I think we are missing is a standard "Document History" for the
> document that overflow and leave the Recent files list.  With a
document
> history you can massively reduce the search space to only the
documents
> that were used previously in the current application.
> 
> I make massive use of this kind of functionality, my Mozilla History
file
> is over 15 MB at the moment (and the history file on my laptop is
probably
> even more massive) but it is easily searchable so I can get away with
it.
> My history file in graphics application runs to only a few hundreds
files
> but despite not being searchable is still quite useful.
> 
> This problem can be and should be tackled from both directions,
although
> automatically aggregating and searching information is useful it is
also a
> good idea to encourage users to organise their data little better.

I don't think it is a very good idea to enforce to much manual
classification on the user. He simply won't use it. What I do believe is
possible is to use the information already available to make it easier
to find it later. 

> Recently on the desktop-devel list  Marc O'Morain brought back up the
> notion of standard folders like Docmuments, Pictures, and so on but
> unfortunately it suffered the bikeshed effect and the discussion
didn't go
> very far but I believe he intended to formalise his ideas and push
them
> again soon.  I hope he does.

Directories like Documents, Pictures and so fort don't really add much
information as the information already exist as suffixes, ownership and
so forth.

What Pictures will do is to collect those files in a common directory.
But think about it, what if Pictures wasn't a real directory but some
kind of meta folder. It could be a collection of all pictures on the
system. Of course it would be a huge folder. Now, what about some kind
of balancing algorithm which added the directory instead if there was a
directory with a lot of images? This is something that is more or less
like locate.

But really, isn't this a special case of one single file type? Perhaps
it could be generalized somehow?

Maybe instead of Documents there could be something like "My files", all
of which are owned by the current user? This would limit the number of
files a lot compared to a complete traversal of the complete file
system! 

Perhaps a folder like "At same time" (in lack of a good word) which
contains files opened at the same time as whatever you have open right
now. Most likely you would have a database with previous results from
scan by lsof to make this work.

Perhaps you could even have a folder "Similar files". That could be
interesting. At least one such system relies on plain mapping into a
Bayesian vector space for classification.

An alternate presentation of actual files in such folders could be
something like a normal file system but pruned to only show what's
included, like images or files with a specific ownership. I personally
don't like this as the hierarchies will hide information.

Now, if there was some kind of list where you could easily organize sort
order on several levels according to metadata. For example music, you
could organize your music collection within artist - album - track that
is folder hierarchy of depth 2 with tracks in album folder or album -
artist - track with track in artist folder.

Actually, this could be some kind of folder with meta order. I like
meta, it is so.. uhm .. unspecific..) Then there could be a music folder
with one organization and other Pictures with a completely other
organization. For example date - location - tema. Given the situation
the user could create as many such folders as necessary.

You would also need some kind of method to set explicit inclusion and
exclusion of files to get the one of interest but that is pretty simple.
That is, you wouldn't want to list mp3's in a images folder and you most
likely don't want to list backup files in such folders.

John