That is a great question, for which I have no great answer, however in case you were at any time in your life around a windows XP computer you should remember the little dog that used to be there to ask you what kind of files you were looking for. That is way better than clippy!

Iaax Page

True. I'd forgotten the dog, and he was more tolerable than Clippy, if only because he didn't leap into the middle of your work to ask if you needed help. A friend also told me that you could chose other forms for the Office Assistant -- a cat, among other things. She said that the cat lived on her desktop for years; apparently it would just kind of curl up in the corner of the screen and sleep while you did work. Then, when you wanted to find help on something, you could click on it. I'm not a cat person, but this also seems a lot better, behavior-wise, than Clippy.

It seems like cultural forces are a relevant consideration here. I recall that Berm Lee (a Korean ex-Cooperista) was a clippy power user (though I think he used a different character), and reported that Clippy was big in Korea.

Right, and I think my question is: Is there a right way to implement characters so that they'd work in places/cultures other than the Far East? Personally, I think that there's got to be a way, but I don't really have any good examples of characters that animate and personify interactions outside of games in the Western world. Jeeves (of Ask Jeeves -- remember him?) was silly, vaguely insulting and classist, and he didn't really help you find stuff. Fail, obvs. Yahoo! avatars are interesting, but (so far, anyway), they're not doing much other than representing a user in virtual space. Anyway, with the blend of Google voice search, LittleBigPlanet, lolcats, and the more visceral sorts of interactions that attend gestural UIs, I feel like there's something here. Not sure what it is, but I know it's cute.

You could indeed choose alternatives to Clippy - my favourites were the cute robot and the Einstein-like professor (who appeared to be based on a plasticine model!). But behaviour wise, they all responded the same way. You had to tell the cat not to jump up to "help" unless asked, just as you had to ask Clippy.

For an excellent analysis of Clippy's failings as a user interface agent, and some evidence that agents definitely do have a place in interface design, I recommend Luke Swartz's paper "Why People Hate the Paperclip: Labels, Appearance, Behavior, and Social Responses to User Interface Agents" (you can find it at http://jerz.setonhill.edu/weblog)/permalink/3662/

Whoops, sorry, I made a typo in my previous URL; that should be:

http://jerz.setonhill.edu/weblog/permalink/3662/

You know, it seems that the key problem with Clippy was he was uninvited. He always popped up, but you didn't know why or when to expect him. Just when you thought he was gone for good, he would pop back up and interrupt you. His purpose was to engage you - which goes against the whole purpose of using software.

Perhaps a more appropriate way to use animated characters (outside of games) is to use them to anthropomorphize the interaction of the product. Take for example a Coke machine that needs to tell the user it doesn't have change for a $5 bill. A character could be used to give the machine personality. It's a simple use and perhaps not revolutionary, but what else do you really need a character for? Their purpose is only so we humans can relate to what they represent in a more 'human' way.

Microsoft could have used Clippy to soften it's error messages and make them more friendly by simply letting Clippy appear to be the one delivering the message. Not in some sort of long drawn out animation - by simply showing him on the screen with the message would most likely be enough.

Most products are not designed to engage the user or should I say start the conversation. Tutorials may do this, games do this, but most software is for getting work done. I wonder how many uses in Japan use characters in very simple ways. I recall reading how popular sending animated characters along with text messages on phones was several years ago... are they still being used to do such simple things?

@Ben: It took me a while, but I finally got around to reading Luke Swartz's paper. THANK YOU. Wow.

The entire paper was a good read, but I especially enjoyed the section called "Mental Model of the Paperclip" and its discussion of the ways in which the research subjects interpreted the "feedback" that Clippy was providing.

"One advanced-beginner informant noted that it 'tells me I’ve done something wrong ... It's supposed to stop you so you don’t continue on to make a mistake.' The other confused advanced beginner said, similarly, 'It tells me when I need help.'"

Supposed to stop you? Tells you when you need help? (I would have been nice if Clippy knew better than you when you need help, but he never did). Those quotes are as close to pure gold as I've ever seen in research verbatims. Wow. Thanks, again.

Because of the glare, one character is not legible, but this is the gist of it. Since Japanese is very context sensitive, a little more context could make the translation more accurate.

次はどのクイズ??挑戦する？
Next, which quiz ?? do you want to challenge
展示番号をたたくと
Tap the display number and
展示の場所が見られるよ。
the display location can be seen.

I was never a fan of clippy either, but like your friend I did have the cat option enabled.

I can't remember the cat ever actually helping me, but I did like the random things it used to do. Kind of like how I enjoy watching the fox in the gmail teahouse theme go about his business during the day :)

@Fred: Thanks for the translation. As I recall, the DS app was intended to be used in a museum.

@Mathew: I think that the randomness of the behavior can be a major component of the charm. Getting this randomness right is hard, of course, and there are probably a million ways to do it wrong. But I think that this is at the core of what I'm thinking about with characters. (I'll check out the teahouse theme, too. Haven't seen it yet).