New models for building digital trust

Changing digital trust mindsets: Distribute data and avoid honeypots

Jennifer Lee: Sandy, what about organizations that have gotten it wrong? In your experience, what are some of the digital pitfalls that companies have embedded into their business model that hinder trust? 

Sandy: It’s been very disappointing what people have done in the pandemic. An example of this is widespread use of data lakes. Once you have a data lake, more data is exposed. We’re likely to see this occurring more regularly. For example, there are likely going to be vaccine passports just like there used to be TB [tuberculosis] passports. People are going to try and make national registries and statewide registries to support these passports, and that’s a major risk from a centralized data perspective. 

Instead, an alternative approach similar to credit card systems works better. The way it works in the financial system is, I present my credit card to a merchant. That goes via the credit card company to the bank where I have my money. They validate it or not, and then “yes” or “no” comes back to the merchant. The merchant doesn’t know how much I have in my bank, doesn’t even know which bank it is. The credit card company knows which bank, but they don’t know how much the purchase is, typically, and they don’t know anything else about it, too, because all that other information is kept encrypted. 

We could do that with vaccination records where you don’t have these large honeypots of data that can be exploited. If you don’t have a single data lake, if your data’s distributed in the hands of the people who know the data best, because they had to collect it—for example, clinics or your hospital—then you don’t get massive database leaks, because if data is stolen, it’s separate from other data sources.

Jennifer: That’s really helpful. It’s about building frameworks around the organization, not just the data teams, to protect and build trust with our consumers. In your view, once a breach occurs, how does an organization go about rebuilding trust? 

Sandy: How do you rebuild trust? By accountability. If you can continually track where the data goes and detect things early, that’s transparency, right? And then you follow it up with accountability. But also notice that if the data is distributed, you don’t get the massive failures. You may get lots of smaller failures, but they’re smaller, and they’re attributable.

Jeff: I was just going to ask, on the flip side of the honeypot, some may say, we’ll just build the right cybersecurity, the best cybersecurity. How do you ensure in a distributed model that each of those small pockets has state-of-the-art security around it? 

Sandy: Take, for example, the military. They figured out as early as 1500 that putting everything behind a single firewall was a bad idea.⁴ We want to have defense-in-depth, so that if you lose the occasional battle—and you will—you don’t lose everything. And, in addition, you get more accountability because with these distributed data pockets, you know who made a mistake. The result is that you will get much better security, because there’s more visibility into what happens. You need to have that reputational feedback to each of the elements of the organization. 

You also have to understand that you will lose. Bad guys will come up with newer exploits. People will make mistakes. All sorts of things will happen, but you have to minimize your losses and plug them immediately. The most frustrating thing about a lot of data breaches is detecting the problem six months later when they become a big surprise. 

Another interesting consequence is that the people in a local distributed system who collect the data are able to identify issues sooner. Let’s take the COVID-19 example. The health clinic that collects testing data has a much better sense of what the data is, of what it should look like, than somebody in the state capital or the national capital, and should be able to spot anomalies sooner, because they know their population. They know the sort of normal rhythms. And it’s at a finer grain. So you have the potential for, essentially, more eyes on the problem and more familiar eyes on the problem.

What do distributed data systems mean for tech business models?

Michael: Sandy, as you’ve been on this journey, how have you seen the reaction and response from some of those big data houses today, technology companies, for example? What’s their view around the kind of model you’re proposing? And what risks does that expose them to? 

Sandy: It’s a real change of business for most organizations. As an example, we recently helped a large investment management firm that serves retirement accounts whose assets are held by cooperating banks. We helped them manage these accounts without sharing personal data, and, as a consequence, their system is much, much safer. In a similar way, imagine that you as a consumer had data holders that acted as a fiduciary for you. In this case, when a social media platform asks you for your data, they deal with your data co-op rather than dealing with you directly.

A social media platform, investment firm, or an online retailer doesn’t get to see all the data in this model. They just want to know about the critical bits of information that let them execute their value chain. They also have less liability because they don’t own the data, even though they can access insights from the data.

The data partner could be a community-based organization, a “data trust” or “data cooperative.” It doesn’t kill the tech business at all; it’s just a different way of doing business. One of the things it does do, though, is that it breaks a lot of the data monopoly power. If you had lots and lots of data held by lots and lots of hands, and just share insights among them, then businesses can spin up sharing agreements much more quickly than trying to acquire consumer data directly.

Increasing distributed systems builds digital trust at business and government levels

Jennifer: Sandy, I want to shift our discussion to preserving or building digital trust over the next one to three years. What technologies will emerge as organizations seek to improve trust with stakeholders? 

Sandy: Beyond the emergence of data trusts to help people manage their data, the innovation that I think is the most compelling is the emergence of these distributed ledger systems to handle all digital transactions: medical, money, you name it. Singapore is developing a system called UBIN.⁵ The intent is to have a uniform, continually auditable system for logistics chains and payments. This impacts, essentially, most of the big banks and telcos in the Indo-Pacific area. China has a parallel system for its Belt and Road Initiative. It’s big and has already been piloted on a scale of 150 million people. 

Tax authorities are beginning to notice these ledger systems for trade. You can’t today capture all the cross-border taxes across countries, because they’re handled on a whole variety of proprietary systems. What would happen if all that trade were done on a single encrypted system, but now you could say all cross-border trade of type X pays Y percentage? The tax authorities really like it, because it’s not taxing their own citizens, at least not obviously. There’s something like a trillion dollars a year in lost taxes that could potentially be recovered. 

Our research team at MIT helped Switzerland set up the Swiss Trust Chain system with Swiss Post (Switzerland’s national postal system) and Swisscom (the country’s telecom carrier). It supports all digital transactions. The uniformity, continual audit, and ubiquity of these systems make them transformative in terms of cost, security, and, as I said, auditability. 

These ledger systems help the government with taxes and control. They help companies with cost, and knowing their logistics chains. And the fact that instead of having a patchwork of separate systems, you have separate permissions on the same system ensuring security, privacy, and completeness. That Singapore and China are deploying these systems in the Indo-Pacific area means we’re anticipating that three years from now, 10–20% of the world’s trade could be on systems like this. 

Jeff: We covered a lot of waterfront here. Any final thoughts or advice that companies should be thinking about? 

Sandy: This idea of a company having a satellite of fiduciary organizations (“data trusts”) that manage the data rather than trying to own data directly. This changes the mindset from having the data to having the data insights. We then have local data organizations everywhere. They hold the data for citizens, they have the risk, and they get audited. What that creates is a commercial relationship where they share insights with companies, we share offers back for their members, and that’s the way we do business. 

So, all of a sudden, organizations are insulated from brand risk of data loss or mismanagement, and from the problems of data breaches. The digital trust problem migrates, then, to the local organizations, not the company. And that mindset to not holding data directly but having these local, trusted proxies providing insights is a big shift. 

Jeff: Thank you, Sandy. This has been great, and we look forward to continuing to discuss these topics with you in the future.