India desperately needs a comprehensive policy framework for data governance

For example, in addition to model, age and claims history, automobile insurers can consider how a particular vehicle is driven and prevent tax evasion by sharing income tax data with the Goods and Services Tax network. can. Yes, sometimes another one can actually add up to eleven!

Thus, interconnections are even more important than collecting standalone datasets. Therefore, it is important to separate the issues related to value creation and value extraction.

There is a wide range of policy, legislative and standard data frameworks being proposed by the central government as well as the states. These include the Personal Data Protection Bill, the National Strategies for Cyber ​​Security and Artificial Intelligence (AI), and the Framework for Non-Personal Data and Responsible AI. And, proposed platforms across all domains often require data-sharing mandates with government involvement.

personal vs non-personal

Data that directly or indirectly identifies a particular individual is considered ‘Personal Data’ (PD). Accordingly, every other data should be ‘non-personal data’ (NPD).

However, such mutually exclusive and orthogonal binary classification has limitations. For example, you can be uniquely identified even in a large crowd if you have location tracking enabled on only one mobile phone, even though the location data is considered NPD.

Individuals have been re-identified with over 90% accuracy by combining anonymized data (considered as NPD) with public datasets! In addition, the anonymization itself can be reversible such that a spectrum analyzer can show the ratio of the basic colors of a particular shade.

More ways to split!

Other data classifications include: at rest – in transit; on the shore – in the cloud; encrypted – unencrypted; structured – unstructured; low frequency – high frequency; real time – historical; national – trans-national; physical – physical; public sector – private sector; individual – community; Raw – processed.

The list is really endless. Even researchers often combine empirical data with simulated data to test their hypothesis.

There may be more than one way to skin a cat, but data can be harvested and harvested in more than a ton of ways! Yes, there are ways to extract data in a very short time and in an infinitely fast and focused manner. After all, online search results, shopping recommendations, medical treatments and even loan offers can be over-personalised.

Algorithmic trading in securities can be initiated based on the concurrent linking of seemingly disparate datasets such as weather forecasts, currency exchange rates and crude oil production.

Similarly, fraud can be detected using an array of factors such as location, frequency, quantum and, lo and behold, even how much pressure the user exerts on the mobile screen and the way it is tilted. Is!

The ability to generate ‘synthetic data’ using ‘digital twins’ also opens up immense opportunities for innovation. For example, the best vaccine candidates are being identified and tested for their efficacy against potential mutations of the SARS-CoV-2 virus even before they emerge.

policy framework

Data is the elephant in the digital room. To tackle this, India needs a comprehensive policy framework for data governance, not being blinded by discrete one-dimensional tools based on simple binaries that can crack and overlap.

Ronald Coase, the Nobel laureate of economics, famously said: “If you torture data long enough, it will confess.” Nevertheless, data must be handled jointly with accountability and transparency to ensure its proper use for the larger public good. .

Binaries can be useful in appreciating different aspects of the data, but Schrödinger’s cat might just meow “the-lines-they-a-bluerin” inspired by another Nobel laureate, Bob Dylan!

Deepak Maheshwari is a Senior Fellow at the Center for the Digital Future.