Peter Krantz published this article discussing whether Governments really need APIs for open data? Wouldnâ€™t a â€śdownloadâ€ť of a dataset suffice?
My perspective: It depends!
- Is it one or multiple related datasets? Are you looking to expose the relationships between the datasets?
- How large is your dataset & do you have the infrastructure to support downloads of large files?
- If not, do you have skills to support exposing â€śslicesâ€ť of data from your internal systems as externally-accessible downloads?
- Are you exposing any value-add functionality of your internal systems (i.e. geo-spatial query on a dataset that involves additional APIs like directions)
- How frequently is this data changing? Every minute, every day, once a year? Think real-time GPS data or environment monitoring vs list of facilities.
Hereâ€™s where APIs offer advantage:
- Querying large datasets for relevant bits of data (think a 15GB download vs. a 100KB slice of that same data via an API)
- Real-time or frequent update scenarios (GPS bus tracking, current weather where the data is time sensitive & importing it would be inefficient)
- Exposing relationships in the data (the agency is best suited to expose the relationships in the data it provides via APIs vs just meta-data)
- Using its own APIs (the government uses own APIs for visualizing/interacting with the data or presenting it to citizens as information)
With the right API you can still enable download of the data. Built correctly, an API would be able to handle traffic or bogging down internal systems. This is particularly true for APIs where the data flows from internal systems into a cloud-based open data catalogue, with an API that exposes the data + relationships + downloads. The best of both worlds.
That said, not having an API should not be roadblock to publishing data. Downloads are fine for most common scenarios, but cloud-based API are an evolution toward a more dynamic platform for open gov data.
p.s. Thanks Jury for a good find & letting us all know!