Somewhere in a filing cabinet — or more likely a scanned PDF nobody has opened since — is the first nutrition label ever printed on a box of food. It didn't exist because a cereal company decided one morning to be more honest. It existed because people asked, for years, to know what was actually in the thing they were eating, and eventually the asking turned into a standard nobody questions anymore. You look at a label without thinking about it now. You didn't always have that option.
We're at the equivalent moment for algorithms, except most people don't know the fight is happening, because there's no box to read yet.
The thing nobody discloses
Every recommendation you get today — what to watch next, what to buy, who to swipe on, what song plays after this one — is the output of a system that learned something about you first. That's not controversial; it's how these products work, and there's nothing inherently wrong with a system learning your preferences to serve you better. What's missing is the label. You don't get told what it learned, why it ranked one thing over another, or whether the answer changed because it was actually right for you or because someone paid to be in that slot. There's no ingredients list. There's just the output, presented as if it fell from the sky.
That gap isn't a technical limitation. Plenty of systems already know exactly why they recommended what they recommended — the reasoning exists internally, every time. It's a business choice not to show it, because showing it invites questions the business would rather not answer, like "why does this favor a sponsor over the thing I'd actually like best?"
Why this is a consumer-rights question, not a tech question
It's tempting to file "algorithm transparency" under nerdy internet-policy territory — something for regulators and technologists to sort out eventually, somewhere far from your actual day. We think that's backwards. This is closer to truth-in-advertising or truth-in-lending than it is to a technical spec. You already accept, as basic and unremarkable, that a loan has to disclose its real interest rate, that a menu has to list what's in the dish if you ask, that a contract can't bury the important part in six-point font on page 40. Algorithmic ranking systems now make decisions about what you see, what you're offered, and what you pay just as consequential as any of those — and currently disclose less than almost any of them.
The system knows a great deal about you. You know almost nothing about it. That's not a stable, healthy relationship between a person and the tools they use every day.
What the pledge actually is
The Own Your Algorithm pledge, live now on alignalgo.com, is deliberately small. Signing it doesn't hand over your data, sign you up for anything paid, or commit you to using any particular product. It's a name and an email added to a public, honest count of people who believe a simple thing: if a company profits from learning about you, it should have to say so, plainly, and show its work.
That's it. No fine print, because the whole point is that we're asking for the opposite of fine print.
We're building toward something bigger — a plain-language model policy that any company or lawmaker could actually adopt, alongside a public list of the companies willing to stand behind it. That piece is still being built; it isn't live yet, and we won't claim otherwise. What's live today is the starting count: real people, added one at a time, saying this matters to them. No inflated number, no fabricated momentum — just the actual count, however small it starts.
Why a grassroots pledge, and not a lobbying push
We could have hired lobbyists. We didn't, on purpose. A policy written by the industry it regulates tends to end up regulating very little. A pledge that starts with ordinary people naming what they want — before any company gets to weigh in on the wording — has a better chance of actually saying something real. Grassroots is slower. It's also harder to quietly water down in a closed-door meeting, because there's no closed door; the ask is public from day one.
This is also, not coincidentally, exactly what AlignAlgo itself is trying to prove is possible: a system that discloses instead of hides, because disclosure was the design decision from the start rather than a concession extracted later. Align Heart, the first product built this way, shows its compatibility scoring on named dimensions instead of handing you a black-box number — that's the pledge, expressed as a product, not just a paragraph.
Adding your voice
You don't need to understand taste graphs, ranking systems, or anything about how any of this works under the hood to sign. You just need to agree with one sentence: the systems that learn from you should have to tell you what they learned, and why it decided what it decided. If that sounds obvious once it's said out loud, that's the point — it should be obvious. It just hasn't been standard yet.
Nutrition labels took decades. We'd rather this one didn't.