- From Data To Information (and back)
Just now·4 min read
Extending the taxonomy to anti-rival goods — using the latest innovations from computing and cryptography symbiotic goods become network goods — in a guest article series by Aurel Stenzel
Is data a public good? Should it be a private good? Or even a common good? In our first piece of the blog post series From Common Goods to Data Commons, we gave a brief overview of the different types of economic goods and which category data belongs to. In order to categorize economic goods, we have two important characteristics: excludability and rivalry. If we can exclude others from the consumption of a good, we call the good excludable, and non-excludable otherwise. If a good decreases in quantity after it has been consumed, we call the good rival and non-rival otherwise. As described in our introductory post: Ostrom, the Commons and Today’s Data Economy, Elinor Ostrom changed the way we look at those categories and somewhat “doubled” the types of economic goods. Now, we consider four different types:
A typical example of a private good is food. One can exclude others from eating it and as soon as it is eaten, no one else can eat it. An example of a public good is the weather forecast broadcasted on the radio. Everyone (who owns a radio) can turn it on and listen to it independently of the number of other listeners.
IIf data is being consumed (ergo: shared), the data is still there for further usage. Also, similar to a tweet published on Twitter, it is very hard to take back control of data that has been shared. Therefore, data is non-rival and non-excludable and intuitively considered as a public good. However, with the latest innovations from cryptography, data can be turned into a private good (or in any other of the four categories). For example, with Secure Multi-Party Computing (SMPC), multiple parties can execute mathematical operations on encrypted data. Data is not being shared as clear text anymore and can be interpreted by none but the original owner. All parties contribute a private input (which remains private throughout the operation) and receive a public result back.
Let’s turn to a specific and simple example: a group of individuals wants to calculate its average salary. The group decides to use SMPC for the calculation. The calculation proceeds as follows (footnote 1).
No one outside the group can participate — the data became excludable. Also, the group could decide to do the operation only once. As the data is never being shared in clear text, the individual data input cannot be reused and therefore becomes rival. Working on data with SMPC makes data a private good. For more details, see the following SINE foundation library post link.
Data carries very special characteristics, and consequently, the above categories have to be extended even further. Data is neither rival nor non-rival, it is actually anti-rival (other scholars call it super-additive, see footnote 2). While non-rival goods are not reduced in case of consumption (e.g. weather forecast), anti-rival goods even increase. Data unfolds its true value if it can be put into context and is converted to actionable information. By comparing their salaries and putting it into context, the group turned data into valuable information and actually created more data (i.e. average salary).
In its original state, data is considered to be non-excludable and is therefore considered a Symbiotic Good. However, as explained with our example above, with encryption technology, data can be made excludable and therefore becomes a Network Good.
SMPC helps to maintain data privacy even after the data has been shared. Nevertheless, it comes with an important flaw. If the different inputs remain private, who controls that the submitted inputs are correct? As a side note, this flaw actually exists in any data-sharing situation. It is hard to verify the truthfulness and quality of data, therefore leaving room for abuse. One person could submit a very high salary in order to impress the others or to protect her/his own data. We need governance in order to ensure truth-telling from all involved parties. The fact that data is anti-rival can be used to give the right incentives. Combining Ostrom’s insights with data’s special characteristics gives us the possibility to form a new kind of Commons — the Data Commons. Learn more in our next blog post in two weeks.
Make sure to -
#1: Our example has the flaw that e.g. Bob and Charly could team up and calculate Alice’s salary backward from the average salary. However, the more parties join the operation, the safer SMPC becomes.
#2: See for example https://www.oecd.org/publications/enhancing-access-to-and-sharing-of-data-276aaca8-en.htm
- Date of publication:
- Thu, 04/08/2021 - 09:14
Click on the link - it will be copied to clipboard