Every time we interact online—whether as customers, gig workers, buyers, sellers, or even online daters—we invest time and effort into building trust and credibility. Our reputation is shaped through reviews, ratings, and interactions, becoming an essential part of our digital identity. But there’s a catch: that reputation doesn’t actually belong to us. It’s locked inside the platform where it was created. The moment we leave, it’s gone.

Think about it—if you’ve spent years as a top-rated seller on Facebook Marketplace and decide to try a new marketplace, none of your credibility comes with you. You’re just another unverified user starting from scratch. The same is true for gig workers, freelancers, and businesses. Our reputations are trapped within centralized platforms, making it difficult—if not impossible—to take our digital trust with us.

Why should something we’ve earned—our reputation—not belong to us? Why is our trustworthiness platform-dependent? And what if there was a way to make it portable?

The reality is that whenever we conduct our lives in the digital world, we’re just manipulating 1s and 0s on servers owned and operated by the platforms we interact with. Not to get too technical, but many of these “servers” are actually virtualized, and the underlying hardware is often owned by AWS, GCP, or other cloud providers… but I digress. The point is that while these platforms may offer some connective tissue to the broader internet, our existence within them is mostly confined to their ecosystems.

There are exceptions, such as APIs and integrations across platforms that enable some degree of connectivity—usually designed to drive value for the platforms themselves and, hopefully, for their users. Examples of this include a Stripe integration with Shopify, where different platforms can be mixed and matched. Social logins are perhaps the most notable form of interoperability, allowing us to reuse existing notions of identity in new contexts. They're incredibly convenient and likely increase conversion rates, as users can skip the tedium of creating yet another password and username.

But what if we could take this concept even further? What if reputation itself became portable?

To better frame the problem, let’s imagine a scenario: I’ve spent years making a living selling on Facebook Marketplace. Now, I want to expand to a new peer-to-peer marketplace. Maybe there’s a restriction preventing me from selling my products on Facebook, or perhaps they've increased my costs. Over time, I’ve built a reputation—a tried-and-true, reliable seller who's responsive, kind, and fully accountable to my customers. Unfortunately, none of my potential customers on this new platform will know any of that. I have to rebuild that trust from scratch.

What if, somehow, my reputation could carry over to a new platform? Sure, Facebook Marketplace would likely never go along with such a fantasy, but let’s explore how it might work. 

First, we'd need a shared digital substrate that no single entity owns, with redundancy built in. Losing our reputation just because a company goes out of business isn’t an option. We’d also need a standardized format for representing reputation—essentially, a way to structure the metadata so that any third party could honor it within their platform.

Turns out, both of these are solvable problems. That shared substrate? It’s essentially a blockchain. The standardization? It’s just a matter of getting the right people together—ideally, those with expertise in reputation systems and metadata standards—to define it.

Before we go any further, let’s address a few things that might be raising eyebrows. Yes, this is a somewhat non-monetary new use-case for blockchains, requiring trivial amounts of capital, and zero speculation. Another key point: you may be tempted to say what I’m describing is essentially an NFT for reputation. And while that might loosely be the case, it fails to capture the many nuances and complexities of the reputational system I’m envisioning that generally aren’t present in your average ERC contract.

Let’s skip over geeky details, in part because I’m not super interested in them, nor am I especially qualified to list them here. Instead, consider how a car title works: you physically own the car and can generally use it as you please, but the car's mileage and title cannot be tampered with. The lender who holds the lien on your car can modify the amount owed, but they can’t change the fact that you own the vehicle. Similarly, in this reputation model, the platform issuing the reputation would be the only entity able to create or modify it.

So far, we’ve been overusing the word "reputation," but perhaps it serves as an entry point into a much broader vision. Imagine if diplomas, car titles, background checks—you name it—were owned by individuals and made portable. Beyond mere convenience, this shift would lay the foundation for an entirely new digital era. And when combined with AI?

Part of why I’m writing this is that my current role has given me the opportunity to help bring portable reputation to life—but reputation is just one piece of a much larger puzzle. AI data labeling today lacks standardized protocols, leading to fragmented workflows, inconsistent quality, and inefficiencies in how data is collected and verified. The tools and processes are scattered, siloed within individual platforms, making it difficult to create a cohesive system that ensures high-quality data for training AI models.

Our protocol aims to change that by establishing a unified standard for data labeling—one that provides clear guidelines, shared infrastructure, and interoperability across different platforms. A key part of this is reputation. Those contributing to data labeling, whether through direct annotation work or by lending their expertise, should have a verifiable history of their contributions. We aim to memorialize their work history, credentials, and, where applicable, validate their identities. This allows for a more transparent, trustworthy workforce—one that can be carefully curated based on meaningful, standardized reputation data.

Ultimately, we’re not just creating an ownable and portable reputation system; we’re setting the foundation for a decentralized standard in AI data labeling—one that fosters openness, efficiency, and accountability across the entire ecosystem.

While our focus is currently in a very specific context—helping train the next generation of AI models—the implications are much broader. We’re particularly focused on extensibility, meaning our standard could be applied well beyond our immediate use case. And while it’s still early days, companies like Coinbase are already working on reputation systems. Yes, it’s an API, which is somewhat antithetical to the decentralized vision we hope to cultivate, but it’s a giant step in the right direction. In fact, we’ll likely be consuming it as part of forming reputation profiles for individuals—and our reputational standard might even become part of what Coinbase supports.