I’m thinking, particularly having used Pol.is myself, it’s not quite clear to the user how these different decisions are being made about who goes in which groups and things like that. I’m really interested in your questions about how do you ensure democratic oversight of these algorithms, of associative intelligence?