Uncanny Valley
Recently I’ve been thinking about the concept of uncanny valley.
Uncanny valley is what happens when artificial things get so realistic they’re almost perfect. There’s a point where they start becoming less believable the more realistic they become. In the case of artificial humans they become more corpse-like the more real they become. There’s something repulsive about them.
Especially if they’re moving.
Check out this robot goat thing. So real it’s creepy!
This robot child totally creeps me out.
Simpler artificial things are more believable. We’re able to suspend disbelief and apply all the missing emotional and physical characteristics to make them more real than the more realistic simulacra (is that the right word?).
The film industry have been struggling with uncanny valley for years, since CGI has become good enough to render entire characters in the film. If you’ve seen Final Fantasy - The Spirits Within you’ll know what I mean. Toy Story is more believable because it cleverly sidesteps uncanny Valley.
Game designers are having to contend with Uncanny Valley as game consoles are now powerful enough to render “almost human” characters in games.
I’d argue that games were almost as believable when the characters were a small collection of coloured pixels. Nintendo are doing a great job of bypassing Uncanny Valley with their games.
Scott Mcloud’s fantastic Understanding Comics explains that when a cartoon is simple it’s easier for people to connect with, because the lack of features widens the amount of people the cartoon could be representing. It’s easier for the reader to put themselves in the place of a simple character than it is a complex one.
Which brings me back to Tweenbots. Funny little fragile cardboard robots that you can’t help but love.
They were an experiment in seeing if people would help them. The results were amazing - One bloke even turned one round, sending it back the way it came saying “Don’t go that way, it’s dangerous!” even though he thought nobody was around.
I wonder if uncanny valley applies to interaction design.
It’s clear that when a person is interacting with a complex object or system they give it human characteristics. There was a test done somewhere (I forget where, but it’s mentioned in Alan Cooper’s book) where people were asked to complete a set of tasks on a computer then fill in an online form to say how well they thought the system performed. Half of the test participants filled in the form on the computer they used for the tasks and half of them used a new computer to fill in the form. The ones that used the same computer were actually more polite than the ones that used another machine. This would appear to suggest that people thought they might offend the machine!
So if the pursuit of better user experience has led us to make systems that are more human, more realistic and more helpful are we in theory in danger of falling into Uncanny Valley?
This might mean that simpler, more incomplete, humbler digital experiences could provoke a higher level of emotional engagement than the slicker ones.
Food for thought, right?
