Data Spectrum — language evolution

Gavin Starks
2 min readJul 1, 2022


The data spectrum has become part of our language for data infrastructure to describe the market architecture for data.

Reflecting on the language of the data spectrum, and in conversation with Leigh, here’s an evolution of thinking.

Open = any-to-any

Shared = many-to-many

Closed = some-to-some (or none/not shared at all)

The Data Spectrum [cf. history of its development] we defined at the Open Data Institute focussed heavily on trying to help people understand Open Data — data that anyone can use for any purpose for free. It laid foundations for why it was more relevant to think about data in terms of ‘how I can use it’ (which translates as how it can be licenced) rather than its size or origins.

At Icebreaker One, where we are going deep on sharing more sensitive commercial data at web-scale, we have defined Shared Data as data that has a pre-emptive license (that can be defined across sectors and/or applications).

In discussion, we’ve now hit upon another way of describing the context of Open, Shared and Closed relating to the market architecture for data.

This framing of the market architecture is based on the number of connections that can/need to be made with the data.

If you have your own data and don’t want to share it, or just need to share internally (without a licence) or with just a few organisations (through license or contract) then you don’t need a common market-based licensing framework: bilateral contracts are fine (and sharing inside an organisation is covered by employment contracts). This is Closed Data.

As soon as you want to scale to a many-to-many environment it is more efficient, secure and scalable to have pre-emptive licenses (e.g. in Open Banking there are millions of users and hundreds of organisations in the same licensing framework). This can be manifested in a Trust Framework for Shared Data.

If you want to scale further, to an any-to-any environment, then Open Data is the best approach. Again, pre-emptive Open Data licenses can be applied including Creative Commons and Open Government licenses.

This framing, I think, is very helpful to frame how to assess which approach to take when addressing different market needs. The boundaries will remain fuzzy but represent potentially useful tipping points.

For example,

  1. if you want to share with anyone: Open
  2. if you want to share with anyone ‘apart from X’: Shared
  3. if you want to share with 100’s or more: Shared
  4. if you want to share with a few or 10’s: Closed

This thinking is central to our thinking at Icebreaker One, so please do comment here and/or get in touch if you would like to collaborate.