⁠Proof of personhood without scanning your eyeballs

Devcon 8 · 6 Nov 2026 · 7:52

Jay Sojitra, a full stack developer working on Web3 identity infrastructure, asks how a person can prove they are one unique human online without an iris scan. Jay explains the Sybil attack problem in the age of AI, why airdrops, DAO voting, public goods funding and rate limits need one share per person, and why government ID and KYC reveal too much. The talk then examines World's Orb iris-scanning approach and its trade-offs in privacy, security, centralization and accessibility, before time runs out. In the Q&A, Jay favours key-based approaches over biometrics.

Speakers

Topics

From Devcon 8, hosted by Ethereum Foundation.

Transcript

Good morning everyone. So, I'm Jay and today I would like to walk you through one question like how we can prove that we are a unique human online without scanning our eyeballs. So, let's have a quick show of hands like who here had chat with bot last week. It either could be a DM or your response under your post. So, yeah, one person, right? Now, keep your hands up if you are 100% sure that it was a bot.

So, yeah, that's the correct like answer like we don't know like who ex- act as whom. So, here is the core issue that online one person can act as a thousand of people. And in crypto we call it a civil attack. And in this AI era, AI almost made it nearly free to like generate this kind of fake people. So, AI like writes like you, replies at 3:00 a.m. of your any DM and etc.

So, what are we actually trying to build that Vitalik defines proof of personhood simply as a list of two keys without revealing which key belongs to whom. So, two properties he he has focused on. First one is uniqueness like one person should only contains one key. So, you can't sign up twice. And the second is privacy like nobody learns which key belongs to you. And getting both at once is very hard right now to implement.

And question is why should this room care about this? Because anything that counts, people breaks without it easily. For an example, airdrop and UBI. So one person So basically in airdrop, one share per person. It should be one share per person, not one share per wallet because a person can own multiple wallets as well. Second point is voting like Dow governance and quadratic voting assumes one water and one what. Third one is public goods. So funding gives bigger matches to many small donors. So fake donors will drain the pool. So like they won't get opportunity to like raise the fundings. And

anti-spam like rate limit, there should be a rate limit per human, not per account because I have six Gmail accounts. So basically I'm a single human, but I have six accounts. That doesn't make any sense, right? So next is So all of these need a same rule. We are one human and one share only. So what we have today in the system, that is government ID and KYC. So you can upload your password, you can upload your selfie. And it works perfectly, but there is a catch like but every app learns that who are you. And there is the

like we have to compromise with the privacy here. So what WorldCoin is building right now, so previously it was known as a WorldCoin. Currently it is known as a world. So that organization suggested three steps. First is that they will build custom device called orb. So orb will take a photo of our iris. And it doesn't keep the photo of our iris. It creates the special hash from our iris and store it into the database. So one person can't sign up twice because iris we can't change, right? And then with ZK proof, um we can prove that this iris belongs to the database, but not my iris.

So like actually they have done very clever cryptography here. And next is like again show of hand like would you let all to scan your iris for free tokens? No, right? Because once it got leaked then you can't reset your iris. You can reset your password you but you can't reset your iris. So, that's the catch here. So, there are four trade-offs in order to build this kind of um system. First is privacy. Like even with Orbis, anyone who scans your iris can check whether you are in the registry or not.

Uh second one is security like 3D printed fake bodies, government forcing people to register to specific um system. And that the sneakiest one will be in the next slide. Uh third trade-off is centralization. So, Orbis a like specialized hardware. We have to trust like it was not built with any backdoors. We are not sure like who which company has built the Orbis and whether it has any leaky information or not. We are not sure. And the last one is accessibility like there is hundreds of Orbs but like millions of people but there are hundreds of Orbs only. So, no one can reach to the Orb easily. So, the last one is um

like you can't uh uh allow people to rent your IDs. If you have money then you can rent out other IDs as well. It's done? Okay sorry. >> Well, we'll have to move into the the Q&A. >> Yeah, yeah, yeah. Sorry. >> So, very interesting conversation. I believe we can start with this very important question around confidence. So, how is this solution better than if we move away from iris verification? >> So, I would say like if we go with the um public keys and private key setup so, we don't have to leak like we don't

have to share our in our information to any system, right? So, keys belong to us, right? So, that particular key that we can keep it safe by us only. So, not our any fingerprint or body part will get leaked or any other personal information would get leaked. So, that way I believe we can make this solution very confidential. >> Makes sense. Very well. And in regards to the second question, without biometrics, what will stop one person from gaming the system and collecting 100 ZK credentials?

>> Without biometrics, what stops one person from collecting 100 of ZK credentials? So, um I would say um same answer would be there for this question as well because um without biometrics, there would be a option to use keys only. Like either it could be hardware key, your hardware public key or your uh software key as well. So, whether I'm not sure whether I got this question correctly, but uh happy to chat about this uh separately.

>> Exactly, that's the way. To conclude and wrap up, wouldn't biometrics work with ZK proofs? Or wouldn't that um be feasible? And human passport is is an example of that. >> Uh biometrics work with works with ZK proof. Uh Worldcoin has already done it. So, there is no doubt in that. Uh the only thing is why we should uh give our iris to them in order to prove uh generate ZK proof where when we can achieve the same thing uh with set of keys as well.

So, that's the like my take.