File systems.

22 Oct 2004 - 7:03pm
9 years ago
22 replies
498 reads
Tom Hobbs
2004

Hi,

Does anyone have any information (or know where I could some) —
research data etc. — how well the average user (consumer) understands
the hierarchal folder based filing systems implemented into OSs at
present?

My suspicion (through observing people in various types of research) is
that while hierarchal files and folders are familiar and therefore easy
they relate to it, they're really 'understood'. I think this is
supported by opening the average computer and seeing the sporadic
taxonomy the user has created for themselves, the state of their
desktop and/or witness them trying to find something. Or working in a
group of people with a shared file server...

Or is this just because they require a lot of overhead and constant
management to make them work effectively?

The question has arisen in developing an application that could adopt a
'virtual' filing system and what is leveraged from existing (computer)
mental models to create that balance between familiarity and confusion.

-thanks

Tom

Comments

22 Oct 2004 - 9:29pm
Listera
2004

Tom Hobbs:

> The question has arisen in developing an application that could adopt a
> 'virtual' filing system and what is leveraged from existing (computer)
> mental models to create that balance between familiarity and confusion.

In recent history, there have been attempts at "virtual FS" from the "data
soup" of Newton OS to evolving products like GMail, Google Desktop Search
and OS X Spotlight. The jury is still out on this. In iTunes/iPhoto, for
example, you can elect to have your entire music/photo collection managed
opaquely by the respective app and the average user never has to deal with
file storage, from acquisition to 'dynamic folders'. The issue is *when* the
user needs to access specific files manually, then it becomes problematic to
deal with a structure the user had no say in concocting.

Ziya
Nullius in Verba

23 Oct 2004 - 5:20am
Petteri Hiisilä
2004

> My suspicion (through observing people in various types of research) is
> that while hierarchal files and folders are familiar and therefore easy
> they relate to it, they're really 'understood'. I think this is
> supported by opening the average computer and seeing the sporadic
> taxonomy the user has created for themselves, the state of their desktop
> and/or witness them trying to find something. Or working in a group of
> people with a shared file server...

Most people understand it sooner or later. People can learn abstract,
sometimes useless stuff.

The hunter/gatherer optimized human brain can handle recursion, but it's
not very good at it. Files inside folders inside folders inside a folder
inside a folder.

Big bad #1: The good old file system works and can be learnt, but
forming a mental model of it isn't trivial. File system is a dancing
bear. If you want to hire a real dancer, don't hire a bear.

> Or is this just because they require a lot of overhead and constant
> management to make them work effectively?

This is the big bad #2. File system is not good at organizing stuff.
Tagging&filtering is a lot closer to human brain's way of categorizing
and finding things.

> The question has arisen in developing an application that could adopt a
> 'virtual' filing system and what is leveraged from existing (computer)
> mental models to create that balance between familiarity and confusion.

If

1) The files contain some relevant metadata for the user,
2) You can develop an intuitive (perhaps automagical) tagging system,
3) You can develop an effective filtering system,

Go for the virtual solution. Try to use the incremental search whenever
possible.

The filtering system itself can still be recursive, if the recursion
serves a purpose:

Music > Artists > Chopin (Perahia) > Études, opus 10 & 25
Music > Albums > Études, opus 10 & 25
Music > Genres > Classical > Chopin (Perahia) > Études, opus 10 & 25

Take a look at iPod's "virtual filing system". It just works. Handling
the metadata is still a needle under a nail, and I hope that Apple will
work on that.

In the meantime I'll just use Tag&Rename:
http://www.softpointer.com/tr.htm

Best,
Petteri

--
Petteri Hiisilä
Palveluarkkitehti / Interaction Designer /
Alma Media Interactive Oy / NWS /
+358505050123 / petteri.hiisila at almamedia.fi

"The unbroken spirit
Obscured and disquiet
Finds clearness this trial demands"
- Dream Theater

26 Oct 2004 - 3:11pm
Cristian Cheran
2004

> Most people understand it sooner or later.
> People can learn abstract, sometimes useless stuff.

It's worth mentioning that this is like the whole problem of the
current PCs: eventually people can use it, but they aren't really
comfortable with it, and they often encounter issues coming from the
inherent obscurity of the established GUI (which, btw, relies a lot on
file system, too).

Until we get

1) Better-than-Google search
2) Metadata-based organization
3) Standard taxonomies for common tasks (tough nut to crack)

...we'll have to endure this UI kludge called "hierarchical
file-system" in our designs. And since Microsoft just dropped these
"noble achievements" from the coming Longhorn (WinFS) we'll just have
to emulate them inside our applications if possible.

Cristian Cheran
----------------------
Interaction Designer
Cheran Software
www.cheranware.com

26 Oct 2004 - 5:12pm
Listera
2004

Cristian Cheran:

> 2) Metadata-based organization

Expecting users to create and maintain metadata is a losing proposition.

> 3) Standard taxonomies for common tasks (tough nut to crack)

Not gonna happen.

> And since Microsoft just dropped these "noble achievements" from the coming
> Longhorn (WinFS) we'll just have to emulate them inside our applications if
> possible.

Or switch to another OS that'll ship it in a few months.:-)

Incidentally, the "search" aspects of this will arrive before Longhorn, the
FS related aspects won't.

Ziya
Nullius in Verba

27 Oct 2004 - 7:00am
Kevin Cheng
2004

>> 2) Metadata-based organization
>Expecting users to create and maintain metadata is a losing
proposition.

And what of expecting communities to create and maintain metadata?
Moodlogic (www.moodlogic.net) has been running for years and their
music database, while not complete, is quite rich and up to date.
Similarly, newer communities like Flickr and del.icio.us are doing the
same.

Kevin Cheng (KC)
OK/Cancel: Interface Your Fears
kc at ok-cancel.com
www.ok-cancel.com

27 Oct 2004 - 9:01am
Petteri Hiisilä
2004

>>> 2) Metadata-based organization
>>
>> Expecting users to create and maintain metadata is a losing
> > proposition.

True. It should be automatic. This is the weak chain.

Digital photos should have their own little id3 tags. Just like mp3
files do. Pictures are just about the only media files that don't carry
any metadata at all. For the rest it could be more or less automatic, if
the designers really want it to be. CDDB is a good example.

- Petteri

27 Oct 2004 - 9:35am
Tanya Rabourn
2004

On Oct 27, 2004, at 10:01 AM, Petteri Hiisilä wrote:

>>> Expecting users to create and maintain metadata is a losing
>> > proposition.
>
> True. It should be automatic. This is the weak chain.

Some very useful metadata is automatic of course, e.g. date, size, etc.

> Digital photos should have their own little id3 tags. Just like mp3
> files do.

There are two standards I believe, IPTC and the camera info in EXIF.
http://www.macdevcenter.com/pub/a/mac/2002/11/14/photo_metadata.html
http://www.iptc.org/pages/prel_20041011.php
Just need better apps and cameras to take advantage of it and edit it.

-Tanya

27 Oct 2004 - 9:37am
bill pawlak
2004

> >>> 2) Metadata-based organization
> >>
> >> Expecting users to create and maintain metadata is a losing
> > > proposition.
>
> True. It should be automatic. This is the weak chain.
>
> Digital photos should have their own little id3 tags. Just like mp3
> files do. Pictures are just about the only media files that don't
> carry any metadata at all. For the rest it could be
> more or less automatic, if
> the designers really want it to be. CDDB is a good example.

Many newer digital cameras *do* automatically attach metadata, but it's
typically shutter, aparture, time, etc. The problem I have with the
whole problem of metadata for photos is that they're much more personal
than music. Meaning, I've taken over 4000 pictures with my digital
camera and I'm really the only one who can assign all that metadata if
I want to use it to find things again.

With CDDB, you have people the world over entering in title, artist,
and album info. It works b/c each song has those attributes. However,
there's not many people in the entire world who could look at one of my
pictures and assign metadata such as: "Scott, Ann, and their friend,
Kate at Aunt Jo's House on Deb's 23rd birthday." That's why shutter
speed, aperature, etc. *can* be auto-recorded. Every picture has those
things in common. Too bad it's not as meaningful to the general
population as keywords that _describe the contents of the picture_
would be.

I'm always amazed at places like corbis.com (or many stock photo
sites). Do a search for "friends" and each picture has sometimes 40+
other keywords (metadata) associated with it. They must have whole
teams of people assigning & validating that data. They're welcome to
come over to my house and help me get all that sorted out, too.

That being said, however, I wonder if an image-specific metadata
framework couldn't be constructed that could at least get people
started. For example:

People
- Friends
- Family
- Aunt Jo
- Coworkers
- Strangers
- Celebrities
- Politicians

Places
- North America
- Canada
- USA
- Colorado
- Boulder

The geographical breakdown would be easier to construct, since it's
(mostly) static and universal... the Canadian provinces will always be
in Canada (except maybe Quebec, for many unbelivable reasons. But I
digress... ;)

But there's of course still going to be problems, esp. with the
"People" example above. What if I happen to live next to Governor
Ahnold and he comes to a BBQ? Friend? Celebrity? Politician? Yes.
All of the above, but that's what the metadata is for.

bill

__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com

27 Oct 2004 - 9:51am
Coryndon Luxmoore
2004

> Many newer digital cameras *do* automatically attach metadata, but it's
> typically shutter, aparture, time, etc. The problem I have with the
> whole problem of metadata for photos is that they're much more personal
> than music. Meaning, I've taken over 4000 pictures with my digital
> camera and I'm really the only one who can assign all that metadata if
> I want to use it to find things again.

If you were to make the camera location aware with GPS or some other location-based service you could potentially automap the address/location and maybe even correlate it with your personal address book or social network account.

It would be cool to see your pictures arranged by physical location with a date range to adjust the time period. --CML

27 Oct 2004 - 10:10am
bill pawlak
2004

--- cluxmoore <cluxmoore at dakasa.com> wrote:

> If you were to make the camera location aware with GPS or some other
> location-based service you could potentially automap the
> address/location and maybe even correlate it with your personal
> address book or social network account.

True. But it still doesn't capture the _context_ of the photo. a
lat/long or a physical address is helpful in many instances, but still
misses the mark, in general For example, all of the following are
meaningful, but in different contexts.

1) 41.9541N 87.6605W
2) 1060 West Addison, Chicago, IL 60613
3) Wrigley Field

> It would be cool to see your pictures arranged by physical location
> with a date range to adjust the time period. --CML

not *my* pictures...not the recent ones, anyway... mostly just a bunch
of pictures of my kids in various shades of pink pajamas. :)

bill

__________________________________
Do you Yahoo!?
Yahoo! Mail Address AutoComplete - You start. We finish.
http://promotions.yahoo.com/new_mail

27 Oct 2004 - 10:27am
Coryndon Luxmoore
2004

We could add in that creapy face recognition software as well... Could possibly automatically tie it to events in a calendar app as well...like say your kid's birthday or bed time ;)

If you can break down the walls between the data there are lots of intelligent relationships you can draw the problem is that the file systems and file types and applicatios are not really geared towards eliminating these walls. The sooner we can break these barriers down the better.

--CML

27 Oct 2004 - 10:58am
Petteri Hiisilä
2004

> True. But it still doesn't capture the _context_ of the photo. a
> lat/long or a physical address is helpful in many instances, but still
> misses the mark, in general For example, all of the following are
> meaningful, but in different contexts.
>
> 1) 41.9541N 87.6605W
> 2) 1060 West Addison, Chicago, IL 60613
> 3) Wrigley Field

If you have that data, capturing the context is simple. The "place" tag
must be personal anyway.

Every pick near this location (all 1563 of them) is from our summer cottage.

Exact location should be as exact as possible.
Personal location should be as personal as possible.
The connection between these two is a one-time operation. I guess it has
to be made by the user.

I hope that those GPS chips get cheaper. That's a brilliant idea, Cluxmoore.

Not to mention that mobile phones already know their location. And they
have cameras.

What we need is a holiday camera that can also receive those few phone
calls that you might receive during your weekend/trip. And surely it
should be able to send postcards, since it hopefully contains your
address book.

- Petteri

27 Oct 2004 - 11:00am
Petteri Hiisilä
2004

> Every pick near this location (all 1563 of them) is from our summer
> cottage.

Oh. Every *place*.

Don't talk to anyone while you're writing... You might place the wrong
words.

- Petteri

27 Oct 2004 - 12:47pm
Peter Bagnall
2003

On 27 Oct 2004, at 15:51, cluxmoore wrote:
> It would be cool to see your pictures arranged by physical location
> with a date range to adjust the time period. --CML

Somewhat off topic but...

I've thought for a while that a great service would be "pictures I wish
I'd taken". The idea is that you use GPS to tag where you were when you
took the shot, and which direction you were facing. Then if your photo
turns out really well you upload it to piwit.com (to pick a stupid name
for the service) along with that GPS data. But if it turns out badly
then you log on to piwit.com and find photos taken by other people of
the same place, and pick one you like, hence rescuing your memories.

Of course, this only covers one genre of photography, but it would be
an interesting project. What I'd find most interesting is to find out
what sorts of places people would be interested in putting on there.
Linking this sort of thing into mapping services would also have huge
potential. Eventually it would have historical interest too!

I think I'm right in saying that some cameras already have GPS, so
maybe this is already possible. Also, it would be possible then to do
something CDDB like too, since you'd have other compelling reasons to
go to the site and give info. Of course good photographers might want
some payment, either in real money or in prestige for their efforts.

Anyhow, I'll never have time to build it! Hope someone else does!

--Pete

----------------------------------------------------------
Always design a thing by considering it in its next larger
context - a chair in a room, a room in a house, a house in an
environment, an environment in a city plan.
- Eliel Saarinen, 1873 - 1950

Peter Bagnall - http://people.surfaceeffect.com/pete/

27 Oct 2004 - 1:28pm
Listera
2004

cluxmoore:

> If you were to make the camera location aware with GPS or some other
> location-based service you could potentially automap the address/location and
> maybe even correlate it with your personal address book or social network
> account.

<http://snipurl.com/a3ll>

Ziya
Nullius in Verba

27 Oct 2004 - 1:49pm
Peter Bagnall
2003

Didn't someone post a map that combined the satellite photos with the
maps here a little while back?

More interesting than satellite photos though are photos from a less
exotic altitude of 5-6 feet. Otherwise you need to fly over your
destination before you recognise it. Photos from person height can
really give you landmarks.

You end up with stuff like this...
http://www.crowle.org/

but with the potential of being able to dynamically construct a tour.

--Pete

On 27 Oct 2004, at 19:28, Listera wrote:

> [Please voluntarily trim replies to include only relevant quoted
> material.]
>
> cluxmoore:
>
>> If you were to make the camera location aware with GPS or some other
>> location-based service you could potentially automap the
>> address/location and
>> maybe even correlate it with your personal address book or social
>> network
>> account.
>
> <http://snipurl.com/a3ll>
>
> Ziya
> Nullius in Verba
>
>
>
> _______________________________________________
> Interaction Design Discussion List
> discuss at ixdg.org
> --
> to change your options (unsubscribe or set digest):
> http://discuss.ixdg.org/
> --
> Questions: lists at ixdg.org
> --
> Announcement Online List (discussion list members get announcements
> already)
> http://subscribe-announce.ixdg.org/
> --
> http://ixdg.org/
>
>
----------------------------------------------------------
I would feel more optimistic about a bright future for man if
he spent less time proving that he can outwit Nature and more
time tasting her sweetness and respecting her seniority.
- Elwyn Brooks White, 1899 - 1985

Peter Bagnall - http://people.surfaceeffect.com/pete/

27 Oct 2004 - 2:24pm
Tom Hobbs
2004

>
>> 2) Metadata-based organization
>
> Expecting users to create and maintain metadata is a losing
> proposition.

True. But their is a huge amount of potential to create this
automatically if devices and systems imbed and share it. I would thing
the objective is to create a system that doesn't require users to do a
lot work up front or as they go. This the problem with current file
systems.

The other thing to do is to create this metadata through 'drag and
drop' associate, i.e. if I put this image with this images it gets some
of the information of the other images.

27 Oct 2004 - 2:35pm
Listera
2004

Peter Bagnall:

> Didn't someone post a map that combined the satellite photos with the
> maps here a little while back?

Yes I did, but this is Google. Imagine the integration/synergy they could
pull off with the rest of their offerings. Perhaps the world's largest
metadata store mapped to, well, sat-maps. Lordy!

Ziya
Nullius in Verba

27 Oct 2004 - 2:37pm
Listera
2004

Tom Hobbs:

> But their is a huge amount of potential to create this
> automatically if devices and systems imbed and share it.

Absolutely, but that's something entirely different. That's why GOOG's
market cap surpassed Yahoo's last week. :-)

Ziya
Nullius in Verba

27 Oct 2004 - 5:55pm
Listera
2004

Kevin Cheng:

> And what of expecting communities to create and maintain metadata?

I'm not sure, Kevin.

Years ago before the Internets thingy, I wrote a personal CD collection
manager app like everyone else on the planet. At first I entered all sorts
of metadata. CDs were new at the time, but since I was working with Sony my
collection grew by the hundreds each year (job had its perks :-).

Soon, however, I gave up any sort of metadata maintenance altogether. If a
well-organized person like me with all the good intentions and motivation in
the world won't stick to it nobody else will either. Then the Internets came
along and Google showed the power of the algorithmic approach over the
manual. I remain deeply skeptical of business goals where endusers are
required to maintain metadata.

Flckr is new. Pictures are tagged by endusers, but that's really the only
realistic way to get them seen by others, since people are probably not
going browse at large, horizontally. I'm not sure where that's going to go.
Notice that over the past two years people have advocated the
self-classification of blogs and merging/aggregating those definitions under
various schemes. It went nowhere.

There are obviously lots of *relational* metadata that happens as a matter
of course *within* the community that could be algorithmically harnessed.

I'm skeptical, but who knows.

Ziya
Nullius in Verba

28 Oct 2004 - 12:25am
Listera
2004

> Absolutely, but that's something entirely different. That's why GOOG's market
> cap surpassed Yahoo's last week. :-)

Few people emailed me about what the significance of Google's acquisition of
Keynote that I mentioned was all about. I got a presentation to finish for
tomorrow, so I'll let the always-interesting John Battelle describe it:

<http://battellemedia.com/archives/001004.php>

Suffice it to say, it has enormous data/info presentation and interaction
possibilities, merged with Google's other properties and traffic.

Ziya
Nullius in Verba

28 Oct 2004 - 8:26pm
Rahul Nair
2004

> Message: 41
> Date: Wed, 27 Oct 2004 07:37:44 -0700 (PDT)
> From: bill pawlak <billpawlak at yahoo.com>
> Subject: Re: [ID Discuss] File systems.
> To: discuss-interactiondesigners.com at lists.interactiondesigners.com
> Message-ID: <20041027143744.57348.qmail at web61007.mail.yahoo.com>
> Content-Type: text/plain; charset=us-ascii

> That being said, however, I wonder if an image-specific metadata
> framework couldn't be constructed that could at least get people
> started. For example:
>
> People
> - Friends
> - Family
> - Aunt Jo
> - Coworkers
> - Strangers
> - Celebrities
> - Politicians
>
> Places
> - North America
> - Canada
> - USA
> - Colorado
> - Boulder
>
> The geographical breakdown would be easier to construct, since it's
> (mostly) static and universal... the Canadian provinces will always be
> in Canada (except maybe Quebec, for many unbelivable reasons. But I
> digress... ;)

Check out Mor Naamans work at Stanford on using GPS coordinates to
automatically collect metadata about photographs. The paper is at
http://dbpubs.stanford.edu/pub/2004-43 while the live system with a
few thousand photographs can be found at
http://shark.stanford.edu:4230/cgi-bin/flamenco/mor_full/Flamenco?username=default
You can sort the pictures according to date, time, location, weather,
daylight, etc...

Marc Davis at UC Berkeley also does some interesting work with users
adding metadata to cameraphone images. That paper is at
http://fusion.sims.berkeley.edu/GarageCinema/pubs/pdf/pdf_63900590-3243-4FA0-845E4BF832AA8BCC.pdf

Syndicate content Get the feed