On (2), my guess is that the 1-point category covers a broad range of events, some of which are worth significantly more than 2% of starting the median project, and some of which are very unlikely to lead to any real-world impact at all.
On (3), I’d note that it may be inadvisible to evaluate community builders on a 50:1 ratio even assuming that is the funders’ true preference. If the super-high scoring item is uncommon enough, its existence during a period being evaluated may be a fairly weak predictor of whether it would be present in a future period. For instance, I care a whole lot more about airplane crashes than near-misses, but the number of near-misses in the past few years might be a stronger predictor of future crashes than the number of crashes in the past few years.
On (2), my guess is that the 1-point category covers a broad range of events, some of which are worth significantly more than 2% of starting the median project, and some of which are very unlikely to lead to any real-world impact at all.
On (3), I’d note that it may be inadvisible to evaluate community builders on a 50:1 ratio even assuming that is the funders’ true preference. If the super-high scoring item is uncommon enough, its existence during a period being evaluated may be a fairly weak predictor of whether it would be present in a future period. For instance, I care a whole lot more about airplane crashes than near-misses, but the number of near-misses in the past few years might be a stronger predictor of future crashes than the number of crashes in the past few years.