Let's start with the bottom and work out way to the top. The neighborhood known as Brittingham Apartments was number 15 in the nation with a score of 99.6 - nearly perfect. The first clue that this is not really a neighborhood should be the fact that the word 'Apartments' is half the name. Similarly there are 'neighborhood associations on Madison's official Neighborhood Association map (above) for condos. No offense to the residents of Sherman Terrace, but a home owner's association does not a neighborhood make. The second clue is that only 86 people live here. In fact, both of the Madison neighborhoods on Richard Florida's top 25 list are the only places with less than 100 residents. When you finally start to get an idea of what Brittingham Apartments looks like on the ground - seeing the n'hood boundaries over streets and what those streets look like (below) - you can tell using automated Bikescore for neighborhoods is missing something, or at least not representing everything. While the one street that is actually in the neighborhood does look like a nice bike ride, but with less than 500 feet of road it would be quite a short one. The dead end street of Brittingham Apartments features a park but let's riders out onto S. Park, a fairly busy and high speed road that has a bike lane but it not for novices.
The other Madison neighborhood listed among the Nation's best received a perfect score. It's called Parkside Resident and it's located next to Brittingham Apartments. In fact, if I pointed to the two spots on a map, you would probably guess they were in the same neighborhood. Parkside Resident is so small that depending on where the boundary is drawn (if the western edge is moved a few feet) it would have no streets at all. When it's drawn generously it will include a portion of S. Park - a 4 lane divided road with a bike land and 30mph (below).
I am certainly in favor of using the neighborhood as a unit to evaluate amenity, engage the public, and empower communities. Neighborhood Associations are one of the most common civic activities that people regularly participate in. They are a way for people to exercise democratic power over what happens near their homes, and I think they are useful for stimulating actions that improve communities. When one neighborhood sees another doing great things, they may wonder why that's not happening in their neck of the woods and organize to change that. It could be biking-related, crime-driven, about parks or schools, or even something as simple as planting flowers in a traffic circle. One reason they work so well is they connect local interest to action by capitalizing on identity. The example of these super-bikeable Madison neighborhoods make it obvious that the automated bike score analyses have some short comings. While some may have been avoided by taking some important precautions in preparing the data - they could have limited the analysis to n'hoods with more than x # people, or of a certain minimum area - there are other parts to being a neighborhood and to being a great neighborhood that are more nuanced and less countable.
Other than being mildly funny to think about going for a bike ride in these places, one thing the analysis fails to do is link the unit of evaluation to one of action. This is something I would argue appropriately-defined neighborhoods would be well suited to do. If the residents of Parkside Resident found themselves lacking in another metric, I doubt there would be much utility in organizing the people who live in that one building to change something for themselves or at the city-level (particularly something like biking which relies on a network of infrastructure and policies to be useful, safe, and successful). They likely only have representation with the City as part of a larger group of neighborhoods. And I wonder if people on the City Council would even know where or what Parkside Resident is, or if the people who live there even consider themselves part of the Parkside Resident 'neighborhood.' This is one aspect of neighborhoods that makes them not only difficult to quantify but also to cultivate, they exist in the culture and the minds of the citizenry perhaps more than they do on the books. Basically, I think neighborhoods are communities that are initially spatially defined, and that over time come to embody the character of their residents - at least the best do, and that's difficult to measure.