Here's an example of the gameplay in one of the games being used to determine if you should be hired for a job.

This is being used *right now*.

What if... hear me out... you gauge risk differently in the work place vs a simulated environment where the stakes are low?

The ultimate goal of a measure like this is to categorize applicants in to categories like high risk, low risk etc. But it's a black box. You don't see why the algorithm places you in one category or another. This is important when it's tied to your potential livelihood.

Another problem with this is that you won't know if it was the clicker game, or your qualifications that denied you a job. Hopefully your application will be evaluated as a whole, but if it's a screener question you might be eliminated before a human sees your app.

As you might guess, the actual mechanics of games like this are kept hidden specifically because companies don't want you to *game* their *game*. But keeping this secret is also terrible for accountability.

What if the duration of when a bubble popped was so random it interfered with the predictive variable? You would essentially be rolling a dice and hoping it comes up with a number your employer liked.

There's also the problem that we don't know how the underlying models and correlations were created. What if the baseline was created by a group of people who were very risk-adverse? Then there would be a strong bias against the average population.

And what about the make-up of that group? What demographic variables were controlled for? Which weren't?