I admit to my biases.
As a former long-time employee of a company that provides finger and face technology for the Federal Bureau of Investigation’s Next Generation Identification (NGI) system, as well as driver’s license and passport technology in the United States and other countries, I am reflexively accustomed to thinking of a proven identity in governmental terms.
Because the government is always here to help.
What this means in practice is that whenever I see a discussion of a proven identity, I reflexively assume that the identity was proven through means of some type of governmental action.
- Perhaps the identity was tied to a driver’s license identity maintained by a state agency (and checked against other states via AAMVA’s “State to State” to ensure that there are no duplicate identities).
- Or perhaps the identity was proven via the use of a database maintained by a government agency, such as the aforementioned NGI or perhaps a database such as the CODIS DNA database.
However, I constantly have to remind myself that not everyone thinks as I do, and that for some people an identity proven by governmental means is the worst possible scenario.
Use of DNA for humanitarian efforts
Take an example that I recently tweeted about.
I recently read an article from Thermo Fisher Scientific, which among other things provides a slew of DNA instruments, software, and services for both traditional DNA and rapid DNA.
One of the applications of DNA is to prove family relationships for migrants, especially after families were separated after border crossings. This can be done in a positive sense (to prove that a separated parent and child ARE related) or in a negative sense (to prove that a claimed parent and child are NOT related). However, as was noted in a webinar I once attended, DNA is unable to provide any verification of legitimate adoptions.
Regardless of the purpose of using DNA for migrants, there is a certain level of distrust among the migrants when the government says (presumably in Spanish), “We’re the government. We’re here to help.” You don’t have to be a rabid conspiracy theorist to realize that once DNA data is captured, there is no technical way to prevent the data from being shared with every other government agency. Certain agencies can establish business rules to prevent such sharing, but those business rules can include wide exceptions or the rules can be ignored entirely.
Therefore, Thermo Fisher Scientific decided to discuss humanitarian DNA databases.
As a result of migration, human trafficking and war, humanitarian databases are a relatively new concept and are often completely separate from criminal databases. Research has shown that family members may distrust government databases and be reluctant to report the missing and provide reference samples (1). Humanitarian databases are repositories of DNA profiles from reported missing persons, relative reference samples, and unknown human remains and may be managed by non-governmental organizations (NGOs), though in some instances they may be managed by a governmental institution but kept separate from criminal databases. Examples of humanitarian databases can be found in the United States (NamUs, University of North Texas HDID), Canada (Royal Canadian Mounted Police), Australia (National DNA Program for unidentified and missing persons) and internationally via the International Commission on Missing Persons (ICMP).
As you can see from the list, some of these databases ARE managed by government police agencies such as the RCMP. But others are not. The hope, of course, is that migrants would be willing to approach the humanitarian folks precisely BECAUSE they are not the police. Reluctance to approach ANY agency may be dampened by a desire to be reunited with a missing child.
And these non-governmental efforts can work. The Colibri Center claims to have performed 142 identifications that would not have been made otherwise.
Reluctance to set national standards for mobile driver’s licenses
Because of my (biased) outlook, mobile driver’s licenses and other applications of government-proven digital identity seem like a wonderful thing. The example that I often bore you with is the example of buying a drink at a bar. If someone does this with a traditional driver’s license, the bartender not only learns the drinker’s birthdate, but also his/her address, (claimed) height and weight, and other material irrelevant to the “can the person buy a drink?” question. With a mobile driver’s license, the bartender doesn’t even learn the person’s birthdate; the bartender only learns the one important fact that the drinker is over 21 years of age.
Some people are not especially wowed with this use case.
The DHS Request for Comment has finally closed, and among the submissions is a joint response from the American Civil Liberties Union, Electronic Frontier Foundation (EFF), & Electronic Privacy Information Center (EPIC). The joint response not only warns about potential misuse of government digital identities, but also questions the rush of establishing them in the first place.
We believe that it is premature to adopt industry standards at this time as no set of standards has been completed that fully takes advantage of existing privacy-preserving techniques. In recent decades we have seen the emergence of an entire identity community that has been working on the problems of online identity and authorization. Some within the identity community have embraced centralized and/or proprietary systems…
You can imagine how the ACLU, EFF, and EPIC feel about required government-managed digital identities.
Is a Non-Governmental Identity (NGI) feasible and reliable?
Let’s return to the ACLU/EFF/EPIC response to the DHS Request for Comment, which mentions an alternative to centralized, proprietary maintenance of digital identities. This is the alternative that I’m referring to as NGI just to cause MAC (massive acronym confusion).
…others are animated by a vision of “self-sovereign
identity” that is decentralized, open source, privacy-preserving, and empowering of individuals. That movement has created a number of proposed systems, including an open standard created by the World Wide Web Consortium (W3C) called Verifiable Credentials (VCs)….DHS should refuse to recognize IDs presented within centralized identity systems. If a standard digital identity system is to be accepted by the federal government, it must be created in an open, transparent manner, with the input of multiple stakeholders, and based upon the self-sovereign identity concept. Such a system can then be used by federal government agencies to view identity credentials issued by state departments of motor vehicles (DMVs) where doing so makes sense. If standards based on self-sovereign identity are not considered mature enough for adoption, efforts should be directed at rectifying that rather than at adopting other systems that raise privacy, security, and autonomy risks.
For all practical purposes, the chances of the ACLU/EFF/EPIC convincing the Department of Homeland Security to reject government-proven identities are approximately zero. And since DHS controls airport access, you probably won’t see an airport security agent asking for your Verifiable Credentials any time soon. Self sovereign identities are just as attractive to government officials as sovereign citizens.
Who issues Verifiable Credentials?
As ACLU/EFF/EPIC noted, Verifiable Credentials are still under development, just as the centralized system standards are still under development. But enough advances have been made so that we have somewhat of an idea what they will look like. As Evernym notes, there is a trusted triangle of major players in the Verifiable Credentials ecosystem:
There are a number of directions in which we can go here, but for the moment I’m going to concentrate on the Issuer.
In the current centralized model being pursued in the United States, the issuers are state driver’s license agencies that have “voluntarily” consented to agree to REAL ID requirements. Several states have issued digital versions of their driver’s licenses which are recognized for various purposes at the state level, but are not yet recognized at the federal level. (The purpose of the DHS Request for Comment was to solicit thoughts on federal adoption of digital identities. Or, in the case of some respondents, federal NON-adoption of digital identities.)
Note that in the Verified Credentials model, the Issuer can be ANYBODY who has the need to issue some type of credential. Microsoft describes an example in which an educational institution is an Issuer that represents that a student completed particular courses.
Without going into detail, the triangle of trust between Issuers, Verifiers, and Holders is intended to ensure that a person is who they say they are. And to the delight of the ACLU et al, this is performed via Decentralized Identifiers (DIDs), rather than by centralized management by the FBI or the CIA, the BBC, B. B. King, Doris Day, or Matt Busby. (Dig it.)
But NGIs are not a cure-all
Despite the fact that they are not controlled by governments, and despite that fact that users (at least theoretically) control their own identities, no one should think that digital identities are the solution to all world problems…even when magic paradigm-shifting words like “blockchain” and “passwordless” are attached to them.
Here’s what McKinsey has said:
…even when digital ID is used with good intent, risks of two sorts must be addressed. First, digital ID is inherently exposed to risks already present in other digital technologies with large-scale population-level usage. Indeed, the connectivity and information sharing that create the value of digital ID also contribute to potential dangers. Whether it is data breaches and cyber-intrusions, failure of technical systems, or concerns over the control and misuse of personal data, policy makers around the world today are grappling with a host of potential new dangers related to the digital ecosystem.
Second, some risks associated with conventional ID programs also pertain in some measure to digital ID. They include human execution error, unauthorized credential use, and the exclusion of individuals. In addition, some risks associated with conventional IDs may manifest in new ways as individuals newly use digital interfaces. Digital ID could meaningfully reduce many such risks by minimizing opportunity for manual error or breaches of conduct.
In addition, many of these digital identity initiatives are being pursued by large firms such as IBM and Microsoft. While one hopes that these systems will be interoperable, there is always the danger that the separate digital identity systems from major firms such as IBM and Microsoft may NOT be interoperable, in the same way that the FBI and DHS biometric systems could NOT talk to each other for several years AFTER 9/11.
And it’s not only the large companies that are playing in the market. Shortly after I started writing this post, I ran across this LinkedIn article from the Chief Marketing Officer at 1Kosmos. The CMO makes this statement in passing:
At 1Kosmos, we’ve taken our FIDO2 certified platform one step further with a distributed identity based on W3C DID standards. This removes central administration of the database via a distributed ledger for true “privacy by design,” putting users in sole access and control of their identity.
1Kosmos, IBM, and Microsoft know what they’re talking about here. But sadly, some people only think these technologies are “cool” because they’re perceived as anti-government and anti-establishment. (As if these companies are going to call for the downfall of capitalism.)
Which identiy(ies) will prevail?
Back to governmental recognition of NGI.
Don’t count on it.
Anticipated DHS endorsement of government-issued digital identities doesn’t mean that NGI is dead forever, since private companies can adopt (and have adopted) any identity system that they wish.
So in truth we will probably end up with a number of digital identities like we have today (I, for example, have my WordPress identities, my Google identities, and countless others). The difference, of course, is that the new identities will be considered robust – or won’t be, when centralized identity proponents denigrate decentralized identities and vice versa.
But frankly, I’m still not sure that I want Facebook to know how much I weigh.
(Although, now that I think about it, Apple already knows.)
Bredemarket 