Parameterized query fragments are embodied as physical tokens.

July 22, 2006 10:13 AM

Direct physical manipulation of data. PDF & you should see the Video. via information aesthetics.
posted by signal (17 comments total)

It's interesting in a quaint kind of way. The physical tokens seem cumbersome and potentially limiting though.

I guess for somebody who can't conceptualize the data they're working with, moving it into physical space could be helpful. I'd like to see what they ultimately do with the interface. For now and for myself, I'll stick with Excel Pivot Tables.
posted by willnot at 10:27 AM on July 22, 2006

Direct physical selection of data, or manipulation of data appearance, says the pedant. Manipulation of data would get you thrown out of the lab. Would I need to carry those knobs around with me?
posted by Mapes at 10:54 AM on July 22, 2006

I'll stick with Excel Pivot Tables.

But isn't that in itself 'physical' manipulation? You're just dragging and dropping representations of knobs instead of actual knobs.
posted by underdog at 10:58 AM on July 22, 2006

I just don't see how a system like this helps. You'd already have to be able to conceptualize the data in a fairly non-physical way (the x and y axis and scatter plots) in order to use it (for instance, you have to map the idea of rotation [turning the token] to scale).

I'm not sure how data can be handled in a tangible way, but this isn't it.
posted by Deathalicious at 11:18 AM on July 22, 2006

mapes, I had the same reaction.
at first I thought they were demonstrating a magnet's targeted effect on RAM/storage.

still, I'm a sucker for projected interactive interfaces.

data miners of the future will have a utility belt like batman
posted by Busithoth at 11:20 AM on July 22, 2006

This seems like a step backwards to me.

Like using a rack of sound equipment to run a Reason workstation.
posted by empath at 11:42 AM on July 22, 2006

You really don't want to see how they do an SQL injection attack against one of these systems.
posted by gimonca at 11:50 AM on July 22, 2006

I just don't see how a system like this helps.

I'm optimistic about the future (the present might suck), so I think it's great.

The Media Lab at MIT does half tech/half art type projects much like this. This isn't supposed to streamline statistics work, it's more an exploratory interface that works on an art level more than a technical one. I could see how this could actually help in a kiosk-type display at a museum or something (turn the dial to change the global temp and watch what happens in our diorama!).

I'm glad there are grad students out there trying new wacky stuff like this. Who knows what the fifth iteration of this project might look like or how much it could help.
posted by mathowie at 11:56 AM on July 22, 2006

Like using a rack of sound equipment to run a Reason workstation.

One of the inherent limitations of keyboard/mouse/video workstations is limited points of contact. You get one gradiated point - the mouse - and maybe 3 or 4 momentary switch points of contact - the keyboard. This is why electronic/computer musicians use interface boxes, midi shell keyboards and other interface boxes with lots of fiddly knobs on them. It's a lot easier to turn three or four or more knobs at once if they're physical rather than software.

Even modern touch screens haven't addressed this issue - except for a very small amount of specialized hardware or theoretical interfaces. I'd love to have a working multiple mouse or multiple contact touch interface for my OS, graphics, gaming and/or musical computer work.

That said this interface project has inherently limited usefulness due to the physical widgetization of the interface. While it's flexible and modular and it can be adapted to many tasks and many kinds of widgets could be dropped in that bay, I'd rather see something done in software and graphical UI or some kind of hybrid system. And as it's an experimental device, this experimental interface probably still doesn't address the "multiple points of contact" problem. I wouldn't doubt that you can only manipulate one, two or three values at once.

I look forward to the day I can easily implement some kind of broad-spectrum haptic interface where I can get my hands all over the OS, interface and data like so much wet clay and not have to throttle my bandwidth down to whatever OS/UI I'm dealing with just to keep from overloading it with input.
posted by loquacious at 12:16 PM on July 22, 2006

Continuing with the comparisons to music tech, Cycling 74's Lemur is another approach to physical manipulation of data. Out of my price range for sure, but it's still an awesome feeling controlling 10 parameters with 5 fingers on an XY axis.
posted by Monstrous Moonshine at 12:18 PM on July 22, 2006

Didn't we used to call "physical manipulation of data" where "tokens are manipulated, interpreted, and graphically augmented on a series of sliding racks" an abacus?
posted by mr_crash_davis at 12:21 PM on July 22, 2006

I studied under Dr. Brygg Ullmer while I was at Louisiana State University. His big deal was tangible visualization, and he runs a heck of a lab there. The cat is crazy-scary smart, and was a delight to study under.

Offtopic:
I was writing a paper for him, last summer, and, upon the first draft, he looked at it, and said: "it's too long and by no means succinet".

I freaked, saying: "there's too much here to be said, I can't even begin to express myself in a mere 15 pages".

Ullmer opened up his briefcase, pulled out Watson & Crick's DNA paper, dropped it down in front of me, and said: "this was done in 3".

It was my single most humbling experience ever.

And, yes, this is my first post. I've been lurking for 5 years, and MeFi finally pushed my buttons enough to make me post.

Congrats Dr. Ullmer. Congrats MeFi.
posted by The Giant Squid at 12:56 PM on July 22, 2006 [3 favorites]

Hot damn my spelling in the previous post is terrible. All apologies.
posted by The Giant Squid at 12:58 PM on July 22, 2006

Monstrous Moonshine: Yeah, the Lemur is one of a very small handful of touchscreens that does the multiple contact thing. I want one. The very thought of it swamps my panties.
posted by loquacious at 1:01 PM on July 22, 2006

Take the notion of physical tokens on display here, and generalize the tokens themselves, and things get more interesting. Having to search around for the Acreage token seems like a step back, yes, but being able to, say, lay a finger on an arbitrary wheel-control and say "Acreage" into the microphone and thus assign that identity to that knob? Ooh baby.

It's a lot easier to turn three or four or more knobs at once if they're physical rather than software.

Exactly. What loquacious said. This seems like a demonstration of the utility of a generalized analog control system—there's a lot of promise here, even if the specific demo isn't all that utile.
posted by cortex at 1:52 PM on July 22, 2006

You really don't want to see how they do an SQL injection attack against one of these systems.

I've got your SQL injector right here, buddy.
posted by klausness at 1:59 PM on July 22, 2006

I enjoy physical manipulation of data, though I suspect the main utility is in real-time manipulation for analysis or art, or learning purposes.

Check out the videos of something tangenitally related - the monome system, "a reconfigurable grid of sixty-four backlit buttons". I've been tempted to FPP it for a while, but this seems as good a place as any. A very versatile means of making and changing music, sounds and patterns with a bunch of buttons. I want one.
posted by MetaMonkey at 4:20 PM on July 22, 2006