DNS for the USPS

Nothing in life is certain but death and taxes… and probably updating one’s mailing address a few times before the end. We probably won’t solve death or taxes in our lifetimes, but eliminating change of address forms for all time is within our grasp, and to do so hardly requires standing on the shoulders of giants.

I’ve moved a lot — big moves coast-to-coast and small moves as close as the unit upstairs. Regardless of the distance, the mail game is the same: log into a few dozen websites (banks, brokerages, health providers, The New Yorker) to update my address, setup mail forwarding which is only good for a few months, try to alert any friends and family likely to send me mail, and assume anything that I might miss is replaceable. Even all that hasn’t stopped me from missing out on a Christmas card or two or misdirecting the odd package with an inattentive auto-filled online order. But the stakes aren’t always so trivial either, at least according to one relative who learned of an undeliverable medical bill only after it had gone to collections.

In USPS’s broken mental model, it’s mailboxes that get mail, not humans.

Electronic mail isn’t nearly so tedious: if Chase Bank wants to send me an e-statement, or a college roommate wants to send me a digital holiday card, they can use my email address. I created my current email address probably 15 years ago and I don’t expect it will ever change. “Got my email address? You’re set, no doubt about it.” This certainty is afforded even though the actual servers that are transmitting, receiving, storing, and serving me that email message are changing all the time — perhaps by the minute somewhere in the delivery chain.

We’re starting to pull at a thread that asks us to distinguish between the targeted recipient and the routing instructions which detail precisely how to deliver that message to the targeted recipient. Email (and in fact most human-directed web traffic) distinguishes between the two — I am a unique email recipient jamie@somedomain.com and always will be regardless of whether I choose to change the technical destination among inboxes with Gmail, Microsoft Office 365, or Apple Mail, or even run my own mail server.

At a conceptual level, web routing works because very reliable systems are able to translate simple target identifiers such as jamie@somedomain.com or Medium.com in your address bar into finer and finer routing steps which ultimately connect you with the right end-point. In fact, right now there are evidently about 10 steps between my home wifi network and Medium’s servers, but who cares — it’s invisible to me and I like it that way.

Someone hoping to send a card to Jamie Campbell at 123 Fake Street isn’t so lucky. Firstly, composing an address is an exercise in tedium: an address is a long, disjointed string of numbers, abbreviations, and made up words — e.g., Squeezepenny Lane — that quite literally give the instructions for how to find the physical building I’m (hopefully) to be found residing in. And that effort is only rewarded if Jamie Campbell actually ever did live and still lives in that physical building. The mailing address itself describes routing steps, not a recipient: the ZIP code characterizes increasing degrees of geographic specificity, then a street, then a number long that street corresponding to a building, then possibly a unit or apartment number within that physical building. It’s like addressing an email with an IP address.

At a fundamental level, USPS barely understands that the entity “Jamie Campbell” exists, even though he is semantically the whole reason for the mail system to exist at all. In USPS’s broken mental model, it’s mailboxes that get mail, not humans. If Jamie Campbell leaves House-1 and moves to House-2 without telling each and every sender, mail for Jamie will simply pile up at House-1. Further, if I fabricate an identity “Bob Nobody” with an address at House-3, mail for Bob Nobody will pile up at House-3 even though Bob doesn’t even exist. USPS offers some helpful features to mitigate this brokenness such as temporary mail forwarding, but this addresses a symptom while leaving the root cause unexamined.

What would it look like for USPS to shift their mental model and distinguish between recipients and routing? Would it be better? Almost certainly yes for all parties involved: recipients, commercial and consumer senders, and even USPS itself.

Consider an alternative inspired by internet routing methods: each individual who desires would be granted one or more unique identifying codes such as jamiecampbell@usps. With proper permissions, as a mail-receiving userwould be able to define the precise physical address that the code corresponds to and would be free to change that configuration over time. By addressing a letter to jamiecampbell@usps with absolutely no other information, a sender would be assured that the letter would be routed to the current physical end-point currently configured by Jamie Campbell. At each step in the chain, automated systems (which already substantially route email through the use of OCR and barcodes) would translate the address to my physical address and route the item one step further. If I as Jamie Campbell choose to move then I can simply update the physical address that jamiecampbell@usps points to in USPS’s system, and my letters are promptly directed to the new physical end-point without senders even knowing that I had moved. Borrowing features from Gmail, I could even split out variants: `jamiecampbell@usps` would refer to my default end-point and is routed to my home building, while `jamiecampbell+office@usps` is routed to my current work building.

This isn’t rocket science. The Domain Name System is a wonderful inspiration, and is likely many times more complex than a USPS address lookup system would need be. DNS coordinates the behavior of many many separate entities across networks, corporate boundaries, and international borders— USPS, on the other hand, is a single organization that can control its entire operational chain. A simple (robust) database would suffice to intelligently route mail as described. USPS could even sell access to its lookup service to UPS and other private mail carriers as a revenue-generating exercise.

The mail exists to deliver mail items to People who move, and the mail system must accommodate this reality.

So who cares?

Commercial senders would win: undeliverable mail is a dead weight loss to advertisers, and high rates of undeliverable mail actually disqualifies mass mailers from the best USPS rates.

Personal senders would win: no more address books written in pencil — one address for each contact for all time.

Recipients would win: A) Never again do I have to update my address with dozens of senders when I move, B) I am more confident my mail will actually find me, and C) I feel more secure — I’m not handing out instructions for how to literally knock on my door to all entities wishing send me a flyer.

And USPS would win by delivering a more user-friendly, more efficient, and more modern mail experience.

Fundamentally, it’s about the USPS shifting its mental model and becoming “Software People” — dispositioned toward building systems which can be adaptive to changing requirements. Mailboxes are not mail recipients; People are mail recipients. Mailboxes may not move; but People do move. Combined, the mail exists to deliver mail items to People who move, and the mail system must accommodate this dynamic reality. Software points a way.