[danielkyne]

Great question! Right now it's open ended to allow participants to leave whenever they want (not overly high tech!)

We've got loads of plans for how we can improve this over time though, using NLP & word-embedding to provide a broad participation across all the core themes and powering the order of opinoins by decision-trees, k-means clustering and matrix factoristation so that we're showing the right opinions to the right people to get a predictable view of the entire population of users.