🎤 Live Worldcoin Orb Disassembly Walkthrough

So, welcome to the Orb disassembly masterclass. We have a lot to cover , and we want to leave time at the end for you to ask your questions. Hopefully we can make it, and if not, you can find us later. So, today we will look at three main points. main points.

First, why did we create Orb in the first place? Secondly, how it works, and that's when we'll take a detailed look at the equipment itself. And finally, what does the future hold for Orb and Worldcoin itself , right? To give you a little context on why we created Orb: crypto world—it’s just great, isn’t it? But today he has two main problems.

Very few people use it, but all the value is concentrated in the hands of a small number of people. So a few years ago we thought: what would happen if Bitcoin was given to every person in the world? And we decided to try to do it. We said: let's try this with a token. We called it Worldcoin.

And for the last few years, we have been working on exactly that . But we ran into a problem: how do we make sure that each person only gets their share once, right? And the more we studied this issue we studied this issue , the more we realized that this is a very common problem for many projects in the world—resistance to Sybil attacks. It's basically the Sybil attacks. It's basically the idea that if there are 20 accounts, you don't know if it's 20 people, each with an account, or one person with 20 accounts, right?

So we started looking for ways to solve this. And there were three things that were extremely important to us during this process. this process. First, no matter how we decided it, it had to be confidential. Secondly, it had to be inclusive, meaning everyone in the world would have the opportunity to use it.

And finally, it had to scale, right? There had to be a real way to provide this to every person in the world every person in the world . So we started with what you might guess is a familiar approach, right familiar approach, right ? Let's just ask people to verify their email addresses or phone numbers. But this, of course, stops working pretty quickly, because it's very easy to get a new email or a new phone number.

Then we said, okay, what about doing something like KYC, right? This is what banks do. Let's just check the documents check the documents that people have. But it turns out that more than half of the world's population either does not have an ID card, or there are no tools with which to verify these documents. Plus, it's extremely expensive and very privacy-invading, right ?

We don't want to collect the names and personal documents of a bunch of people all over the world. Therefore, we abandoned the KYC procedure. Then we came across the idea of ​​a " came across the idea of ​​a " web of trust" . It's the idea where I confirm that you're a confirm that you're a real person, and you confirm me, and through mathematical models, implicitly or explicitly, you can build a system of trust. Where you can have some confidence, and we really think a " really think a " web of trust" is cool, there are a lot of projects that are building it.

The problem with a " building it. The problem with a " network of trust" is that network of trust" is that you need a large group of initial users who can be trusted, right? Because if the early adopters lie to you, you're essentially building a house of cards , right? So we really believe that the “web of trust” is how we will take Worldcoin from one billion to 8 billion people. But the only way to make it work is make it work is to do something different, right?

We need to find another way to reach the first billion users. After much thought and experimentation, we came to the conclusion that the most fundamental solution is biometrics in general, right? If you want to introduce the concept of a unique individual, then biometrics is what distinguishes us from each other. So we also covered a bunch of things , right? Obvious solutions don't work for three main reasons.

main reasons. First, they don't have enough data points or entropy to distinguish one person in a billion or ten billion, right? Something like fingerprints simply doesn't have enough data points. Secondly, the problem we encountered is that we don't try to verify that you are who you say you are. Is n't that right?

So when you go to work and don't say, "I'm Thiago Sada, and I'm showing my biometrics to prove who I am," you're proving that you're someone who's never been here before. This means that even if we had enough data about your palm, for example, by making a small cut, you become a new person. You wouldn't be able to prove that you're Thiago Sada, but you can keep making cuts and be a new person every time. And lastly: there were a lot of really cool technologies that are very new, right very new, right ? And they haven't been studied enough yet.

And we wanted to get something as reliable as possible. So we went through a bunch of options. We looked at fingerprints—they are very unreliable, easy to fake, and there is not enough data. Same with facial recognition. We looked at the palm scan.

Actually, our first prototypes used palm scanning, but it's such a new technology, and there's so much uncertainty around it, and DNA uncertainty around it, and DNA is just creepy is just creepy . This is expensive. This is ineffective. That's why we did n't want to do it. And we stopped at the iris , right?

It turns out that the iris of your eye has enough entropy, enough data, to actually distinguish between many people. And it's also very difficult to change, isn't it? You can get laser correction or something like that, but would you really go for surgery for another personality? It just doesn't scale. And eventually we said, okay, how do we do this?

Our first idea was: let's try to do something smartphone-based, right? Everyone has a camera. But it turns out that the resolution there is simply not enough, right? If you use Face ID, for example, it can only distinguish one in 10 million people, which is great because only one in a million can unlock your phone. But if you try to add everyone to one system, that means after five million people , half of them will be rejected because the system confuses them with someone else, right them with someone else, right ?

Then we said, " ? Then we said, " Okay, so we need specialized equipment." Let's see what's on the market." But the problem was that these devices were designed for one- to-one comparisons, you know to-one comparisons, you know ? They are designed to check: hey, is Philip really Philip?

They're not designed to check: hey, is Philip a new person, right? Also, since we are operating in an environment of lack of trust, we need the system to work accordingly, right? We ship equipment to a bunch of random locations around the world. In addition, there are many measures to prevent fraud: you need to make sure that people are not using prosthetics or simply not showing photos showing photos . And most off-the-shelf biometric scanners simply didn't have such support.

So eventually we realized we were going to have to build our own hardware, which our own hardware, which , to be honest, was n't very good news. We didn't want to build our own hardware. Because creating equipment is really incredibly difficult. But we still decided to do it. And so over the past two years we have developed several generations of ORP.

We'll stop at one of them in a moment. And finally, after a long time, we got this beauty that you saw on Twitter, saw here at the conference, and basically she does three things. First, it checks if the person is real and not trying to scam, right? So it checks the face to make sure the person is alive, that they are breathing, that there is a 3D mapping of their face to make sure it is not a photograph, etc., right?

That you don't wear dentures, that you don't wear contact lenses. The second thing it does is generate your iris code , and by default, that's the only thing that leaves the sphere. And then it is compared to everyone who has been verified before verified before , right? And then, if it is unique, your World ID is registered , yes, it is essentially proof that you are a person that we give you. Now this is the point I really want to emphasize: World ID is not your iris code.

Your identity or your proof that you are human is not your biometric data. World ID—it's not your wallet either, right? We do not verify your wallet. We are issuing you a new type of key, which we call World ID. It is independent, just like the wallet's private key.

So, you can have 100 wallets in the Worldcoin app. You don't have your iris code there because it should n't be stored there. And you also have one World ID key, and it is this World ID key that is whitelisted. That's it. That would be extremely private extremely private .

But on top of that, we actually use zero- disclosure proofs, which means that the key that we give you is not even a public identifier that people can ask, "Oh, I'm number 1 2 3 4 5." It's secretly stored on your phone, and you only use it through zero- disclosure evidence to prove that you're a person who hasn't done something before, right? So we are very excited about this system. If you're curious about how this whole side of the case works this whole side of the case works , particularly with zero- disclosure evidence, we have a post on our blog that breaks it all down in detail. It's called the Privacy-Preserving Identity Proof Protocol, abbreviated as PPP OP, and you can read about it by coming to our booth, where we'll provide more information.

Now let's get to the fun part, shall we? How does Orb work, huh? What we have been creating for two years and how we solve the problems of fraud prevention, rainbow code creation, comparison and issuance of World ID—this will be told by Sandro, Head of Orb at Worldcoin. So he'll walk us through the whole process. Okay, let's begin.

Let's see. Can you hear me? Good. Sorry. This is a joke.

You have been wrapped around a finger. We will not take the Orb apart. We will go the opposite way. We will assemble it from scratch. Let's go.

So, the first part of Orb is the main frame. Let's see. Uh-huh. The most important part here is the round aluminum frame, machined on a CNC machine. Frame.

It gives the device stability, rigidity, etc., and, very importantly, it is spherical. This was not an easy task for the engineers. Believe me, making the Orb round is a real pain. But ultimately, we wanted it to be something unique.

Something that never existed before. Make it very recognizable. As you can see, it is even located at an angle. This is a small symbol. The angle of inclination of the Earth's axis is 23.

5 degrees. It is very important for us to convey this. We intend to significantly expand the scope of this project and go beyond the community of crypto enthusiasts like us. To truly introduce this to the world and bring over a billion people into the crypto world. And all we see is a cooling system for heat dissipation, speakers, and a slot for a removable battery removable battery .

Next we have the main board. This is where many critically important functions are concentrated. There are capacitors for the backlight. We'll get to optics in a moment . We have amplifiers, a connector for the front module on top.

Other printed circuit boards. The liquid lens driver is a very cool autofocus system. We'll talk about that in a moment, too. We have a microcontroller that controls the peripherals: cameras, LEDs, device power, etc., a SIM etc.

, a SIM card slot, and an SSD card slot, and an SSD drive. It is very important to note that by default, no images ever leave the device unless users agree to have their biometric data backed up. So, it's extremely confidential. And, of course, everything is encrypted. On the other hand, we have a connector for a liquid lens, autofocus, which I'll mention in a moment, GPS for monitoring, a connector for the main infrared camera infrared camera , and also the LTE module on the bottom.

We also have Wi-Fi up here. Expect it to be very easy to connect the power electronics for the front unit. The battery connector is here at the bottom. And it is very important that the "brain" of the bullet is inserted here . This is truly where all the magic happens.

Let me see. This is Nvidia Jetson. This is an incredibly powerful system on a module. It runs on a special Linux distribution. The main program is written in Rust and supports multiple streams from cameras.

It is incredibly powerful, which makes it very flexible. The reason it had to be so powerful is that many neural networks run in parallel on this tiny chip in real time to guarantee local processing without ever having to load the image. Everything can stay on this device. Your iris turns into an iris code. All images are deleted and only the iris code leaves the device for verification purposes.

Steeply. So Jetson, this "heart", plugs into the motherboard here. See? Let's insert it. That's it.

And then we assemble this into a main frame to the cooling system to remove all the heat. Let's see. Done. This is the first part of the collection. Let's move on.

Next we will talk about image formation image formation . Why do we actually care so much about image quality? Thiago already talked a little about this. Good image quality is very important, but? Why exactly?

Let's take a bird's-eye view of the situation. What are we worried about? The system must be secure, inclusive, and scalable to become truly accessible; This has never been done before on such a scale in an uncontrolled "field" environment. So what we need to do is really something new. To minimize the level of errors, because, as Thiago said, during facial recognition, over time, new people begin to be confused with those already in the database, and the system mistakenly rejects them.

Of course, this is not the best option if you want to create something fair and scale it to the whole world. We need to create something extremely precise, and the basis for that is the ability to capture and capture a large amount of entropy. For this, we need very high image quality, as this is ultimately the limiting factor for all algorithms running on top of it. So, this is for all the mathematicians watching us. Here are a number of distributions.

You can see that with poor image quality in real-world conditions, we get significant overlap of these distributions, which means a high error rate. Good image quality is the basis for truly distinguishing these distributions. Make sure that the same person is really recognized as the same person, and a new person is recognized as new. Since it is biometrics, there is always a statistical component to it. Therefore, we need them to be very far apart .

Well, let's delve into the optics. Ahem. Thank you. OK. Therefore.

Yes. We have an optical housing We have an optical housing that makes the entire system dust and waterproof. So it can work anywhere. Not only indoors, but also in any outdoor conditions. Wherever it is, it is protected from dust and water.

And the heart of this visualization system is this little guy. We use a specially designed lens combined with an extremely powerful image sensor. We had to make this lens to order. We collaborated with an external expert in the field of machine vision. About a year passed from the moment the requirements were submitted to receiving the finished product.

We tried hard to use off-the-shelf components, but ultimately nothing met our requirements for scaling to a billion people. This lens is optimized for the infrared spectrum. Let's see if you can see it. It is actually blue. The coating is blue.

The coating is blue. So it works best with infrared light. It also has a special liquid lens built into it. This means that usually in order for camera systems to focus on the subject, you move the lens using a motor. And this, of course, is not the most reliable or fastest solution.

We really want to make sure that Orb devices are extremely reliable. They must work in various conditions and not fail, for example, from a fall from a height of 5 centimeters. So we chose something extremely sturdy. It looks like a membrane. It contains liquid oil inside.

It is incredibly fast and durable. This whole structure comes together like this and is inserted here, into the back of the optical housing, just like this. OK. So the just like this. OK.

So the next part is that this lens has an incredibly narrow field of view. So you don't see the person completely or anything like that. Orb only sees a tiny part of a person. So now the question arises: how to properly align a person who is trying to get verified on Worldcoin for World ID? How to correctly position it relative to the device?

This is a very difficult task. It took a lot of time and many attempts. The first thing we came up with was: let's just take two separate cameras. Let's put the mirrors on top. They are transparent to infrared radiation but work like a mirror in visible light so that people can hold the device in front of them, see their eyes, and position themselves correctly.

It kind of worked, it was easy to create an MVP. But it wasn't very good. This was economically unprofitable. Lenses are quite expensive. And then, alignment, especially for people who wear glasses, of which there are a lot in the world, became a problem when positioning.

So we thought: damn it , this won't work. It looks good, but it's not an option. So for future versions, we developed a two-dimensional suspension. Essentially, this is a camera you've seen before. She's here, in the back.

Then we have a static mirror from which light is reflected onto a 2D is reflected onto a 2D suspension. This two-dimensional suspension can be moved in two directions using two motors: up, down, left, right. So regardless of the user's position, we can point the lens directly at them. It's a good thing we didn't create this for the first prototype. It was very difficult to make it work.

But now the verification process is incredibly simple. You can check it later. It takes 10 seconds, uh, and it's done 10 seconds, uh, and it's done . And it doesn't matter where you are in relation to the device. So, here we are.

OK. So the 2D suspension is inserted into, uh, let's see, into the optical housing, like this . This is a bit complicated. That's it. It's almost done.

What's going on top? We'll see. The front printed circuit board is attached from above. There are many very important things here. important things here.

First, we have a visible spectrum camera. It can be compared to the camera on your iPhone, etc. We use it to scan the QR code in the user's app, as well as to point the 2D to point the 2D gimbal we talked about earlier to make sure it's pointing correctly. Then we have a lot of LEDs in the near infrared range. They are located around, in all these sections.

Why do we need so many LEDs? Each of you has probably been to a photo studio at least once. You've seen photographers use a lot of light. This is not for fun, but because it significantly improves image quality, prevents blurring, etc. You know how hard it is to take a good quality photo on your phone at night.

So light really helps to get quality images. That's why there are so many LEDs here. Last but not least, there are several fraud prevention sensors here, which Thiago talked about earlier. We need to make sure that people are not only unique, but that they are real people that they are real people . That they are not trying to tamper with the device, etc.

So we have a 3D etc. So we have a 3D camera that checks camera that checks that it's not a display, a thermal imaging camera, and another near- infrared camera . It is important to note that all of these images do not leave the device's boundaries. They are only processed locally unless users choose to back them up. Uh, okay, this is put on top, and here we have a fully assembled optical unit.

Uh, let's see. Now let's connect this to the rest of the device. It goes in here, like this . Let's see. It's shifted a bit.

No, he doesn't want to get up. Oh, that's why. That's it. And that's it . We're almost done.

And now, um done. And now, um , as the next part, um, one thing that's missing is a spectral filter. So, we talked about the Orb taking pictures in the near infrared spectrum. Um, this means that, um, there are different parts of the electromagnetic spectrum, and only an extremely narrow part of it is what we humans can see with our eyes. Well, this is what we call the visible spectrum.

It's here, this, um, tiny, tiny bump in the middle. And the camera can actually see much more of the electromagnetic spectrum. Uh, a lot more light, and we really want to make sure that no visible light, uh, gets into the imaging process because it interferes with that process. What we need to do is create a laboratory environment somewhere out there in the world, uh, regardless of where the Orb is. And that's actually the very, very difficult part.

And to do that, we separate out a very large portion of the spectrum of sunlight, uh, using this filter. In the visible spectrum, um, it's black. Um, that's why you perceive him as black. It does not let in light. Um, but it's actually transparent, like a window, in the near infrared spectrum.

We just can't see it because, as we said before, humans can't, can't see that part of the spectrum, and it's just put right here on top and seals the optical unit so that it's dustproof and waterproof. So, we're almost done. Ahem. The only part missing is a battery, uh, to power the Orb. So, in the beginning, we started with, uh the beginning, we started with, uh , built-in batteries.

We just took the batteries from the radio-controlled cars, uh, put them in. Well, that was a quick and easy solution. It worked solution. It worked , uh, but it wasn't very good because it wasn't enough for a full day. However, the advantage was that you don't need certification, you don't need all this fancy stuff, but you can actually just, uh, test everything, uh, go out into the "wild ", talk to people, etc.

But eventually we realized: okay, we really need Orb to be able to, uh, run all day long to reach one billion people. Well, we can't put it on charge for three hours charge for three hours and then, and then come back. So we had to create a special replaceable battery. Um, fun fact: it's built around the 18650 cells used in the Tesla Model S. Um, and then that battery is inserted here, um is inserted here, um , from the bottom, and we have a fully functioning Orb.

Well, not exactly, because this Orb doesn't work. Um, we removed all the cables to make disassembly, or should I say assembly, very easy. However, we have one instance where you can still see what's inside. And it really works. You can come over and look at it.

We have 10 orbs: clear, chrome, purple from EAL, a bunch of orbs here for the first time at a conference on such a large scale. 10 orbs. Come and see Come and see . Um, yes, I'll pass the word on to Thiago about what happens the word on to Thiago about what happens next. OK.

Very cool things. This is much cooler than what I'm about to tell you. But, um, okay, before we get to the questions, let's just summarize: what's next? First, a few weeks ago we announced that we are making everything we do at Worldcoin open source. So we've already started, if you go to worldcoin.

org/opensourcing worldcoin. You can read the blog post and see parts of the protocol that are already working, and there is a roadmap for opening everything up, including the hardware. We hoped that the hardware would be ready for the conference. From our side, everything is ready for opening the code. The tricky thing with hardware discovery is licensing, because it's really important to make sure that the hardware isn't being misused.

So we're working with leading open source advocates on a new type of custom license custom license , which we'll then also make open, so that we can release hardware release hardware , be very transparent about what we're doing, allow the community to contribute, but limit or eliminate the harm that releasing this into the world could cause. Isn't that right? So this is extremely exciting. The rest of the Worldcoin stack is also becoming open source. You can see a graph of this in the blog post.

Next, we tested the orbs along with the entire Worldcoin project. The project is being launched in the USA, at least the orb verification is orb verification is yes. And what we found in the field is that each of these orbs is doing 300 to 1,000 verifications per week, which is really, really cool. This means that in the last few months we have conducted over 600,000 verifications worldwide, with less than 30 orbs in operation. Now these orbs were quite clumsy.

We built them in the lab by hand, and there were very few of them. So we're constantly moving them around the world to understand how they work in different conditions: in heat and cold, in sun and snow, in rich countries and in developing countries like developing countries like Mexico, where I'm from. And it was very interesting to watch all of this. The good news is that although we did a lot of verification with 30 orbs, we now have a really cool new factory. Last month we received our first batch of orbs straight from the factory, about 250 pieces.

So all the orbs you see here at the conference are from this batch. This allows us to now produce hundreds of orbs per month, and soon thousands. So you'll see us significantly scale our operations by the end of the year, which is very exciting because we want to provide World ID to everyone in the world. The penultimate thing we're working on—we're, of course, constantly thinking about what's next. You saw that this is the fifth great generation of the orb.

We, of course, are thinking about the future. Both in terms of user experience and in other aspects. We experimented a lot with using light feedback instead of voice feedback. Today the orb is talking. If you've been to a conference, you may have noticed that this creates a lot of inconvenience in a noisy environment.

So we worked a lot on this issue. We experimented with colors. We have a prototype of the white orb on display if you want to take a look. Additionally, we are already starting to think about long-term alternative form alternative form factors, but this will likely take time. And one last thing about the orb: as you may have heard at the conference yesterday, this weekend we released an SDK that allows any developer to use the World ID verification that we provide to users.

This means that just like we, Worldcoin, used World ID outside the US to determine, "Hey, is this a real person and has they received Worldcoin before?" Now before?" Now any developer in the world can, using the same zero- disclosure evidence in a super-confidential way, verify any action. That is, it could be: "Hey, I'm a real person and I've never voted in this poll before", "Hey, I'm a real person, I've never received this NFT before", "Hey received this NFT before", "Hey , I'm a real person, I've never joined this Discord before". So, I know many of you are working on this.

I'm really excited to see what you're creating, so that's all. We want to express our huge gratitude to several open source projects that we use for all of this. And we are also very actively hiring throughout the company. Of course throughout the company. Of course , this applies to hardware, firmware, AI systems that enable all of this, as well as the entire product stack: Solidity product stack: Solidity engineers, product designers for both designers for both hardware and software — hardware and software — in short, if you can come up with something, we are looking for specialists for it, including the community.

So that's all, actually. This, or Orb, helps us verify people and issue them World IDs to provide protection against Sibyl-type attacks against Sibyl-type attacks on a large scale on a large scale . This is private. This is intended as a private tool private tool . It must be inclusive and scalable.

Hardware, as you saw with Sandra, is pretty cool engineering. And all of this is combined with zero- disclosure evidence for even greater privacy, which you can read about on the site. So, we have about 10 more minutes. We wanted to leave some time for questions, because we know there are probably a lot of them. So yeah, that's it .

So, are there any questions? Don't be shy. Express your bold thoughts. Yes. Pink Orb.

Soon. TM. Yes. No. The difference between Clear and Worldcoin, and Orb in particular , is that Clear is designed to re -identify people.

He should say, "Hey, I'm Thiago." "I'm going to get on this plane." " get on this plane." " Make sure I'm still the same person I was two weeks ago." Isn't that right?

As you've heard, one of the pros and cons of our approach is that we can't uniquely identify people identify people , right? And this is built into the design. We don't want to be able to re -identify people haphazardly. What you can do with the SDK, which we also use for our own products, is know that someone is a unique person who hasn't done this before , but you don't know exactly who that person is, right? So I don't expect us to use hardware or software for any purpose related to such things.

Clear is a great company, and they can do it themselves. What else? Yes. If you go to worldcoin.org, there's a list of ideas.

But generally speaking, it means anything where you want to make sure that each person only does it once, right? So, there is management, right? There could be a one- person-one-vote system, conviction voting, and much more, right? In DeFi, you can do undercollateralized lending , right ? Because you can't, uh, guarantee that if you lend money to someone who doesn't pay it back , they won't be able to borrow again , right?

You can do, of course can do, of course , a lot of things in a quadratic format: quadratic funding, quadratic voting, a bunch of things where ten people doing one thing is worth more than one person doing it ten times. There are also interesting uses for NFTs. There is a lot of buzz around soulbound NFTs right now . Thanks to Worldcoin, they can be tied to an individual, not a wallet. In Web2, as I said, we have Discord, we have a Discord bot, and it's open, so you can install it on a server to give users a "real person" role that allows them to write in certain channels and gives them additional rights.

This means that for moderation, if we block them, they won't be able to come back and get that role again because they've already used it. Overall, there are a ton of opportunities, and you know how to use them better than we do . What else? What else? What else?

Miguel, can you show me where the door is? So, as a symbol of our first visit to the US with orbs, we made these wonderful " these wonderful " homemade" orbs that you can "adopt you can "adopt ". If you see the orb and get your World ID, you can find them at the conference and pick one up for yourself. You will receive an NFT adoption certificate and have your own little orb. You'll probably never have a real one, so it's best to get one.

Are there any more questions before we finish? Okay, that's all then. If you have any more questions, just look for us at the stand look for us at the stand . We will be there too . Be sure to get your World ID.

This is our first visit to the USA. I don't know when we'll be in the US with orbs again. So seize the moment. You can do really cool things with it, and it's fun things with it, and it's fun to play with the orb. Well , thank you very much.

See you later.