As part of a property price modelling project, I wanted to build an 'open Walkscore' using Pandana, as suggested in their original paper (https://onlinepubs.trb.org/onlinepubs/conferences/2012/4thITM/Papers-A/0117-000062.pdf).
This led to thinking about:
• what could be in a walkability index, answer: everything - which also means you can use the same design for other modes,
• how to incorporate people's desire to visit multiple places: incorporates infinite destinations, but with diminishing returns to increasing destination numbers, and
2/3
• how to reconcile the catchiness of a '0-100' score while imposing less judgement on what 'full walkability' looks like: complicated, but the diminishing returns thing helps.
Surprising absolutely no one, Australian cities don't perform very well outside of the very centre, apart from good walking access to parks and some amenities like local cafés.