We could also start with an imperfect solution, offer it as a free API (maybe even self-hosted and communicating with other services p2p) and wait for users to select or insert missing addresses, until we eventually converge to a good OSS database. If it's a single service being shared by everyone, you would need to insert your address once and then it would be part of the database forever, and you would get the right result at any other time in the future.
There is also a dirty but hard to attack option:
- Start from the NAG
- Build an opaque AI process which is hard to audit and that is tuned until it produces a result close to PAF but with a few extra errors
- Sell the new database to the government, government open sources that
- Directors get paid their share
- Company get sued out of existance by RoyalMail
- Government pays a few millions in 20 years, if the RoyalMail experts can prove anything in court
I disagree with the report, I think it's feasible with a bit of creativity.
The government also has this: https://www.data.gov.uk/dataset/091feb1c-aea6-45c9-82bf-768a...
We could also start with an imperfect solution, offer it as a free API (maybe even self-hosted and communicating with other services p2p) and wait for users to select or insert missing addresses, until we eventually converge to a good OSS database. If it's a single service being shared by everyone, you would need to insert your address once and then it would be part of the database forever, and you would get the right result at any other time in the future.
There is also a dirty but hard to attack option: - Start from the NAG - Build an opaque AI process which is hard to audit and that is tuned until it produces a result close to PAF but with a few extra errors - Sell the new database to the government, government open sources that - Directors get paid their share - Company get sued out of existance by RoyalMail - Government pays a few millions in 20 years, if the RoyalMail experts can prove anything in court