Confidence

[elainegrey]

There was a post to the ACM's news alerts last week about a game developed to help one improve decision making skills. This morning, after three calls from operations last night, two which came after i was asleep, and after uncomfortable night sweats, i wanted a distraction. It seemed a good time to play a game and take quizzes.

First, a warning. If misspellings and gramatical errors are a trigger for you, do not visit this website. EG: "Variaty of interactive games. There are some interesting ideas and games on estimation and probabilitis." Probabilitis sounds like a horrible disease to contract. On the other hand, what are you doing reading my journaling?

So the description was as follows:

http://www.epsrc.ac.uk/newsevents/news/2011/Pages/computergames.aspx

You're late for a train. Will you be able to catch it if you run? Or will that result in the stress of wasted effort? To maximise your chances of reaching the right decision, you'll need to take into account all information available to you (which may change minute by minute). But it also helps if, using this information, you try to make an appraisal of your chances, which will be more accurate if you take into account how you tend to interpret such information, based on previous experience. For example, maybe you know whether you tend to be over-or under-confident in similar situations.

In the same way, the prototype game (available for anyone to try out at the World Of Uncertainty quiz website) teaches people to take their uncertainty into account and learn from experience when faced with simple choices.

"It's the first ever online quiz designed to let people estimate how sure they are of their answers and score more highly if they don't ignore their uncertainty but realistically assess it," says Dr David Newman, who has led the project.

"Whether the choices facing us are simple or complex, a greater awareness of uncertainty and of our own biases can improve the quality of our decision-making. We believe there's real potential for people to acquire that awareness through computer games."

The prototype game is at http://quiz.worldofuncertainty.org/ . It's not quite up to the level of quizzes that allow you to post your results and share with your friends, although theoretically there is a leaderboard. It was engaging to play a quiz game where one can balance the point award based on how certain you are of your answer.

If you are completely uncertain, you will get 10 points for a right answer, 10 for a wrong; complete certainty is 20 points for right, 0 for wrong. The points don't stay fixed at a spread of 20, though. If you are 30% confident, you can earn 15 for a right answer or 9 (or 8) for a wrong.

So, when asked where the biggest library in the ancient world was, i was 100% confident in the answer. When asked what the main ingredient in Cullen skink was, i just took the 10 points for being wrong.

I've known i'm cautious in game play. I'm terrible to play at chess with my family, because i basically play a game of defense. At the end of the game i'm terrible at finishing, chasing the few of the opponents remaining pieces around the board. It's not so much that i want to win as that i don't want to loose.

Seeing my underconfidence measures is intriguing.

Now i need a game that can measure my expectations of myself, where i am overconfident that i will have an amount of energy and motivation, and fail to live up to them.

And again: i play that "game" not to loose, and often just don't set any expectations.

Interesting.

Comments

[arlie]

I just tried it, and I think their estimates are off. To be precise, I'm not as underconfident as they estimate.

The problem is that the quizzes are "multiple guess". In several cases, I hadn't a clue, and guessed randomly. My expected payoff is (1 * right-payoff + 3 * wrong-payoff) / 4. I didn't run these numbers, what with the time limit, I just kind of presumed "no clue" would produce that effect. Maybe not.

The other problem is that there were questions where I had no hope of naming the correct response, without hints - but with hints, I could use tricks of the trivia trade. For example, when asked "which actress..." the answer is almost always the only one whose name I recognize. Or when asked which alcohol is used in what's obviously a British recipe from a certain time period, pick an alcohol common in Britain, and oft mentioned in that time period. (It didn't even have the one I'd have named, given a free choice.)

Anyway, I feel like my confidence is good - I know when I'm guessing - but somehow the game doesn't represent that.

Of course what was really bizarre was acing the demo game, in spite of guessing.