Direct manipulation Model vs. Object Manipulation Model?

26 Sep 2005 - 8:00am
10 years ago
3 replies
1502 reads
Meikel Steiding

Hey guys,

i guess i've got somehow a stupid question for you but I couldn't
really find a solid answer on the net so I would like to bother you

I am a student who is writing his final thesis which involves a huge
amount of Interaction Design. I never had lectures in this topic so
nearly everything is learned by book or net. That is why I consider
my question as a rather stupid one.

You guys all know about the direct manipulation model. The software I
designed used that in the first concept as well. But I figured out
that for my personas this model is definitely inappropriate. So I
changed my whole input interaction to something I call the Object-
Manipulation model. I don't know if that is the right term and I
would like to ask you guys if there is something like an official
term for it or if my choice kinda hits the right point.

The interaction design of my software is based on consuming the least
amount of attention. The user needs to focus on his other task (i.e.
reading text) instead of worrying or thinking about the control of my
software. In the first place I offered the user a free form pencil
where he/she could draw nearly everything with it. But drawing with
the mouse is more than awkward and the user needed to decide about
size, position, color and form of his shapes. Since making choices
requires attention the approach of giving the user a tool with an
infinity amount of possibility is somehow wrong direction. If I am
not totally mistaken this could be considered as part of the direct
manipulation model.
In the new version the user selects an area and choose an appearance
for it. So the computer decides about size, position and form of the
shape. The user selects items anyway during his workflow so instead
of giving him a pen and let him/her alone with the shapes I provide
him/her with some sort of library where he/she can simply assign an
appearance to his/her selection.
So there is an object (selection) and the user can do something with
it (change appearance, delete it, …) Is it valid to name it Object-
Manipulation model or is there a more appropriate term for it?
For example a lot of graphic applications uses this kind of input
model as well. There is a pen where the user can go crazy if he/she
has some solid mouse skills but there is also the possibility to
select something and let the computer do the work while the user can
focus on the outcome.

Any help in a correct model definition is more than appreciated.

Thanks in advance and greetings from the rainy north of germany.



26 Sep 2005 - 10:03am
Dan Saffer

Interesting question. What you are basically asking is if there is
something between direct and indirect manipulation in that model
(created by Ben Schneiderman, I believe). While I completely
understand where you are coming from, I have no idea if the model
would support that at all. It seems very either/or (how HCI). It
might be a more interesting model if it was a continuum.

Your problem fits in an odd space since direct v. indirect assumes
that there is already an object to be manipulated.

In your case, since the computer is "doing the work" and deciding on
the size, etc., to my mind, that's more indirect manipulation. You
aren't doing anything except determining a position. If, like in, say
Visio, you dragged an object out from a side panel and proceeded to
color, scale and size it by tinkering with the object itself, that
would be direct manipulation. My opinion, anyway.


Dan Saffer
Sr. Interaction Designer, Adaptive Path

26 Sep 2005 - 10:58am
Fred van Amstel

> In the new version the user selects an area and choose an appearance
> for it. So the computer decides about size, position and form of the
> shape.

Maybe it´s the same model Microsoft designers are trying to apply to
the new Office Interface. Jensen Harris describes it here:

"When given a range of options at different levels of detail, people
tended to approach formatting in roughly three broad steps:
1 Apply an overall look from a set of pre-existing styles
2 Refine the appearance by choosing from a set of styles for a
specific subset of the object
3 Fine-tune the look using a few advanced tweaks (using dialog box
controls, typically)"

> So there is an object (selection) and the user can do something with
> it (change appearance, delete it, …) Is it valid to name it Object-
> Manipulation model or is there a more appropriate term for it?

I think it´s indirect manipulation in the first time, but if you can
change the *properties* of the object by innate controls, then it´s
direct manipulation in the second time.

.{ Frederick van Amstel }. Brazil ´´ Curitiba
¶ ...''''''''''||
ICQ 60424910 / MSN van_amstel at

28 Sep 2005 - 7:10am
Meikel Steiding

just to finish this topic; Here is the way how I solved the issue for =20=

my final thesis.

Dan Saffer really hit the right point when he mentioned that it is =20
not a new model, it is more like something in between the direct and =20
indirect manipulation approach.

Ben Shneiderman wrote in 1974 that the following items characterize =20
the direct manipulation:

=95 Visual representation of the manipulated objects
=95 Physical actions instead of text entry
=95 Immediately visible impact of the operation

Well in my case the first two items are still very very important and =20=

thus part of the direct manipulation model. The only thing I am =20
worried about is the last point-the rich visual feedback. It is cool, =20=

it is great, and it is even important to have rich visual feedback =20
but as Donald A. Norman states it out: =93Direct manipulation, first =20
person systems have their drawbacks. Although they are often easy to =20
use, fun, and entertaining, it is often difficult to do a really good =20=

job with them. They require the user to do the task directly, and the =20=

user may not be very good at it.=94 (Norman, POET, p.185) This is =20
exactly the issue in my software. The user has way to many options =20
and possibilities and thus gets distracted from his main task. (kids =20
like candy but it is not always wise to give the kids what they =20
like ;-) ) This is why I classified my model as a compromise of the =20
direct manipulation paradigm it is direct manipulation with less =20
=91direct=92 feedback. This puts the attention away from the control of =20=

tools or the interface and let the computer doing the job in a very =20
accurate and repetitive manner. After all that is where 'he' is good at.
So for interaction concept where attention is a crucial item I would =20
recommend to look into the tool(!) design and cut down some of the =20
0,1,infinity behavior.

anyway just wanted to finish the topic.

Thanks Dan and Fred for their input.


Syndicate content Get the feed