REST APIs are not the gatekeepers to transparency: data with guaranteed access and any predictable format works fine, for many data sets.
Pretend they release the salary survey results in one big, gigantic text file broken down by county and updated yearly. If you wanted to figure out what the average salary of teachers was nationwide, rather than waiting for the government to implement average_salary_by_occupational_class you could just download the entire freaking thing and calculate it. And since you can download the entire freaking thing, nothing prevents you from exposing any API you care to give it. Make it RESTful, make it Big Freaking Java Enterprise XML, put it on an iPhone app responding only to telnet commands written in lolcat, whatever you want.
I've got a project on the back back burner which would benefit great from doing exactly that. (Downloading and exposing an API, not writing in lolcat.) If data.gov gives commercially interesting things I'd publish APIs to my slicing of it just for the PR benefit, since APIs are essentially just a no-marginal-cost industrial biproduct of my own need for the data.
(The engineering resources to develop APIs are no marginal cost. Accessing the API is a marginal cost, but I anticipate it being so low that it isn't worth me worrying about. If it gets to be a problem I'll rate limit it and OSS the scraper code.)
Pretend they release the salary survey results in one big, gigantic text file broken down by county and updated yearly. If you wanted to figure out what the average salary of teachers was nationwide, rather than waiting for the government to implement average_salary_by_occupational_class you could just download the entire freaking thing and calculate it. And since you can download the entire freaking thing, nothing prevents you from exposing any API you care to give it. Make it RESTful, make it Big Freaking Java Enterprise XML, put it on an iPhone app responding only to telnet commands written in lolcat, whatever you want.
I've got a project on the back back burner which would benefit great from doing exactly that. (Downloading and exposing an API, not writing in lolcat.) If data.gov gives commercially interesting things I'd publish APIs to my slicing of it just for the PR benefit, since APIs are essentially just a no-marginal-cost industrial biproduct of my own need for the data.
(The engineering resources to develop APIs are no marginal cost. Accessing the API is a marginal cost, but I anticipate it being so low that it isn't worth me worrying about. If it gets to be a problem I'll rate limit it and OSS the scraper code.)