Futures market -- The basic idea of a Turing test is to distinguish between: But how to calibrate these small tasks to estimate how long each such task takes a human?
By having humans bid on contracts to complete these tasks, and making sure each low-bidder really does complete the project in the bid time, we get an accurate estimate of how long each type of task takes.
Then we can finally compare various kinds of tasks to put them into sequence by ascending amount of time required, then interpolate along that scale to find the "just right" point that protects against spambots without discouraging live humans too much, and choose a variety of Turing tests near that point.
To induce people to make bids and take tests, we offer free services in return for completed contracts.