What do you most commonly enter in that username field? Your email, your phone number?
Hold that thought. We will return to it.
In a previous post I explored why my privacy is more at risk when paying for a newspaper than when not. My proposals, however, were
Specifically, I proposed the creation of a payment processor who would double as an identity provider (IDP). As @jackerhack pointed out, this puts too much power in the hands of the payment processor.
In reply, I proposed a flow that seemingly solves the privacy issues with the newspaper and the payment processor. Except, I got there through
That bit about the generated token aside, the rest of the flow is exactly what already happens. So, the privacy "gains" result purely from using a random token to login instead of a username/password combination.
So, that username field - what's your most common id?
Mine's an old email address. It's lasted me well over ten years - a constant companion while I switched countries and phone numbers. And oh, to make it easy for folks to remember, it's just email@example.com.
Given the dominance of GMail, I imagine this is true for most of us. So, there's an
easily guessable email ID that is the thread linking all of my services together.
Besides, it's not far fetched to say that this email address has been my defacto online identity.
We all accept that services need to authenticate us before they will grant access. Most of us have password managers with hundreds of websites, perhaps 10s of passwords, a couple of email IDs and may be one phone number.
So, essentially, the act of logging into a service as me (or you) consists mostly of guessing my password. Which is to say that the username/password form may as well be just a password form.
Of course, this is not without flaws.
My challenge, and perhaps yours, is that we make profiling far too easy by using the same identity across services. Given the context of centralised apps, I see a few potential solutions.
In this scenario, we create separate identities for each service we sign up for. Kind of like
firstname.lastname@example.org except perhaps a bit more clever.
I recently switched to using Tutanota with a custom domain as my email service. So any email@example.com is a valid email ID. With some simple filters, I catch subscriptions, bills and randoms in different folders.
I could then use firstname.lastname@example.org, email@example.com and so on to keep my identities siloed. That is until some clever sod figures out the domain name part of it.
The domain name is also the primary problem with scaling my approach - how many folks have and set up their own email domains?
Alternatively, how about asking our browsers to generate a unique login id along with the password for each sign up? This would work - if we didn't ever need that password reset link sent to a working email.
Typically, authentication systems enforce uniqueness on the username. We could enforce it on the password instead. This makes the username field redundant.
This does not have to be a race where the first person to use
password123 gets to use it. There are simple existing methods that could work.
The most popular of these is the Universally unique identifier
UUID. This is a 36 character string that looks like this: 123e4567-e89b-12d3-a456-426614174000.
While the probability that a UUID will be duplicated is not zero, it is close enough to zero to be negligible. - Wiki Pedia
Assuming we are all using password managers (and we should be), remembering this sequence is a job we leave to tools.
But most apps don't need 36 character passwords to ensure uniqueness. Blokada uses 12 character codes. With lowercase/ uppercase letters and digits, this allows
62^12 = 3226266762397899821056 unique codes. Should suffice for a long while yet!
Assymetric cryptography using public/private key pairs has been used to secure protocols such as ssh, MQTT and even Bitcoin. This post is already too long to go into details here but suffice to say that:
This is perhaps the most secure and private authentication system we know. Admittedly, this is a more complex paradigm shift. Our browsers and web services are not built for this mode of authentication. But, perhaps, the day is not too far where we may need them to be.
For the moment, I have made the following choices:
I am still finding my feet with these topics so if you have insights and reads for me please do share at firstname.lastname@example.org.