GM fams I believe the new @axisrobotics Challenger Program is addressing a common issue with crowdsourced robot training Not all contributions need to be for every task. A simple pick and place task is quite a different challenge from a difficult articulated task such as drawers, cabinets or appliances. You don't get a 'failed run' if someone who is inexperienced attempts the more difficult one. You may also end up with poor quality training data. Now that Axis is moving those more challenging tasks to a different pool. Access is based on a history of contributions, verification rate, pre and post training activity, task diversity, and consistency of performance relative to other people who are doing the same tasks. The harder tasks are then assigned to qualified trainers, who will receive points that are 2-3 times as high per task. I believe that configuration is good. The more difficult tasks are assigned to those who have demonstrated their ability to create reliable trajectories, and new contributors are given a more forgiving method to learn. The more data, the better it is for robot training. It is even better if the right person is creating the right data.