The State Arrives as a Scan: Building Syria's Open Record, and Making It Forkable

Transcript

hear me? I can hear myself. Thank you so much for sticking around even though this conversation is less about crypto and more about civil liberties and rights. Um yeah, my name is Karam. I come from a very small country um that is in southwest Asia called Syria. And for the last 60 years, I would say we've been living under a dictatorship um that basically took over everything. Um in 2011, we decided that enough is enough. Then we started organizing against the dictatorship. Um and

basically um throughout the uprising um which started in 2011 I I was 20 years old uh when it started. I was studying petroleum engineering. Um and uh basically a demonstration went down our street. Um and I was I I basically decided that I wanted to join it. So I picked up my camera and a mask. Not nearly as cute as Vitalik's mask yesterday. Uh but the idea was that the camera was to try to showcase what's going on in Syria to the outside world. And the mask was

for me to protect myself because in Syria we have a saying that even walls have ears. Um which means that you would not be able to trust even your relatives if you want to talk about politics. One word can throw you in jail forever. Um basically I started uh organizing and helping set up the city's media office in 2012. I was also elected to be within the provincial council which is a mini DAO you can say uh that is taking care of everything related to services in areas outside of the regime control. Um basically we try to keep uh the state records save them and so on.

um and question it as well. In between 2011 to 2014, I basically got arrested by the Syrian government for four times. The last time lasted for one year in a place called Palestine branch, which is a military branch for human rights activists. They drop human rights activists in it. It's two floors under the ground. In one single room of 4x5 m there was 120 individuals with me and due to torture throughout these this time I was in which is the last time between 2013 to 2014 73 individuals

died out of 120 um and this is one story out of over 1 million Syrians that got arrested between 2011 to 2024 There are still 170,000 that never came back and most of them are believed to have been killed by the Assad government. Um, I came out. I was lucky. When I was released in 201 14, the authorities have kept my identity card, my passport, my university transcript.

um and my devices as well. Um I basically had to smuggle myself outside of Syria to Turkey in the north and become a refugee there. But even in Turkey, uh I needed a friend with a passport to help me open or sign a contract for uh a lease to rent a flat or even to register an SIM card. Um, I needed at some point a passport to uh go to the US on a scholarship and I had this to ask the same state that arrested me in 2011 to certify who I was and to say that I'm a human being.

Um, the same authorities that detain me as well still controlled the documents that allows me that allow me to continue my life. This is one way authoritarian powers and dictatorships reach you. Basically, um it reach you through the passport office, through the bank, when you try to enroll in university, you need to state, you need a state to recognize you. Um but also your ability to question what it does to you is quite restricted. So when when a record like that when a passport affects your life, can you read it and can you question the government that is authorizing it?

In Turkey basically I kept working and reporting. I worked as a war journalist or as a journalist. I worked in investigations and I worked in grants management. Um in 2018 I started uh a group called ZFI uh which was aimed at training uh human rights activists and civil society groups in Syria, Jordan, Turkey, Iraq and Lebanon on the use of blockchain. It was open source and free and 42 individuals of course worked with me. We managed to train 132 organizations across these countries. Some of the things that we trained them was very on

the very practical side of things like how do you open an unccustodial wallet? Um how do you start receiving funding without a bank in between when you even cannot open a bank account as a Syrian? Um, and then I also did an Ethereum Foundation fellowship in 2022, which is to ask and to research civil society groups in Syria and how Ethereum and how blockchain can basically and what does it mean for their work. One of the main questions I remember when I was given some of these workshops in 2022 23 in the time of an earthquake that hit both Turkey and Syria is one NGO leader that told me that he was afraid that opening a Metam Mask wallet or an uncustodial uh

wallet address crypto wallet might also put his organization's bank account and PayPal account at risk uh when people depend on that money when they were trying to collect funding for uh attacks that or for the earthquake. Then in December 2024, the Syrian regime, the same dictatorship I was telling you about fell apart. Um families basically started entering the same jails that we were living in and they were looking for their sons and their children and their daughters.

um when they could not find their son or a person um they searched for a name. They looked through lists and ledgers and pages that might say what have happened to their kids. Um this photograph basically shows people reading records at one of the main jails in Syria called Sed Naya outside of Damascus 4 days after the fall of Assad. Um and basically none of them got an answer.

When when a country breaks so to say uh the records of the previous government or the government could break too in 2013. I'm going to just share some examples of what happened throughout the time. In 2013 the land registry in a city called Hums in the middle of Syria got burned. Um in 2015 under siege another city called Dar which was under siege by Assad government as well. The court and notary records also got burned. People lost their papers to prove their homes belong to them or their rental records.

Last in December, as I just showed you, um, in detention branches or outside of detention branches, many of the files and documents got looted and burned. Um, could be by individuals who wanted to hide the fact that they were spies for the previous government. NRC, which is the Norwegian Refugee Council, in a survey mentioned that over 62% of households in Syria have at least one member without any civil documents. Um, in Syria, we have something called the official gazette or the Syrian gazette. And this is basically where the

Syrian states old one and new one uh publishes its laws, its decrees, its appointments under authoritarian regimes as well. Lows and paperwork still exist. Um but a decision basically a decision can be officially published while the people affected by it don't have access to it or struggle to find it or understand it or challenge how it's applied. Basically, imagine for example, an official tells you that you owe a fee or a tax. Where's this decision authorizing it? Is that an amount that actually says what it is? Do you need you need to be able to check out without depending entirely on one person asking you to

pay? And even if you find the rule, challenging an official can also carry risks. You might be put into jail. A document cannot really give you an independent court or protect you basically from retaliation. But without the document, you may struggle even to establish what that published rule say. So the gazette gives you kind of a record that we can examine and deal with. This is basically um this basically the gist of what no the project I'm I'm talking to I'm talking to you about is doing at this point.

Um it the gazette was printed and published and cubbies several cubbies of it across Syria were there. So they were kind of protected because they were not centralized in one building. Um we built something called sigil which is um a one window within and no means windows. Um for centuries Syria's courts, kept registers sales marriages depths lands. Um Syria has always kept records, but the question is who holds them and whether anyone can actually check them.

So we try to have different records um that basically all live onchain as well in addition to living on web two on servers. that are uh around the world. And um when you look into the records that we have, there are different records and each record carries different duties. So yes, we have the detention or like we can take a look at some detention records that belong to the families and the investigators who can protect people named in them especially post crisis especially when the government is moving

or the country is moving away from crisis. Um but the records that we follow within Noahid are mostly the ones that the state choose or chose to publish in the gazette or in the media or in uh with agencies and open source agencies. All of it is public. Most of it is on paper. Um we needed to try to gather all these documents and papers to put them all together. um and basically work on OCR in them uh and building frameworks to extract that data from them.

This is how no it looks like at this point. Uh it has um the app uh it has lots of the information about basically what the state publishes the official gazette the companies and organizations that got registered before. How much do they connect to the previous government to the authoritarian regime has tenders jobs scholarships research and news media basically aggregating everything about Syria in one place. Um we but we wanted to also take into account um basically uh ideas that we have in the blockchain space in the web 3 space. So it's not super centralized and the idea and you have we all have

words for it. So don't trust verify credible neutrality and fork and exit. So for a lot of people these are kind of design design references or uh choices or preferences good engineering something people argue on on a forum or in a referendum in a proposal and so on. The country I grew up with these um these rules basically are there are no permissionless kind of alternative no right to dissociate no right to exit. In Syria, these ideas decide whether the record survives the next government on whether a citizen can check it without asking anyone's permission.

I picked a small decision that uh come up in one of the decrees and basically it's it's ironic because uh it's about the price of a photo copy a document one page from the gazette uh which is 20 Syrian bounds to copy one page uh even the price of reading the government and the state's record is written in the record but to question the decision you first have to find it and this is most of uh what basically no it is doing. Um, our software, our AI, our locally trained models basically uh try to pulse out the numbers, the dates, the ministries, which ministry set up this fee. Um,

you don't need Arabic to read it, but uh, basically our record says, and this is a bug that I want to talk about, especially when dealing with algorithmic frameworks, our record says it's the Ministry of Energy. When you look at the other page, it's the Ministry of Finance that set up that record. At the top, it's basically what you call a a page break bug. The kind any document parser for those of you that deal with OCR can hit. The decision sits under finance and our LLM thought it's under energy.

Um, if you want to challenge who sets a fee, basically you should also be able to catch it without taking my word for it or Noid's word for it. But when we publish, we we have dropped uh that page, the lines and the link to the scan. So to question us, you'd have to have had for the evidence or to have the PDFs that we have and so on. So kind of making us another gatekeeper. Um, and that's why we're trying to work on another framework or a correction trail. So when it said what was checked, how it was checked, when and why.

Um, I spent a year within a system that I could not question. So that's why I owe the people that use the system that we're working on the right to question it. Um as of now this is why we built sigil which I just told you about which is the proof that you can check without even trusting us or the state. This is the first release which has all the documents that we have between 1970 to 2026. Um basically we photographed within Damascus most of it and we also obtained the BDFs for it. The red box as you can see is the missing archive between 1972 to 1979.

It had a Syria had a coup a military coup during that period. So we could not and there was no release of the officer gazed then. But the collection also goes back way further um to 1919 when there was a French mandate in Syria, bulletins and then other uh documents we that are all printed in Damascus. We managed recently to obtain the official gazette from 1919 until now. So we gave every scan we have uh a fingerprint which is one bite.

Anyone can basically it I mean if anything changes or anyone changes one bite changes us included basically. Um and the fingerprints goes into one list uh and then it goes into a release manifest and the manifest fingerprint sits on one Ethereum transaction. Um basically you can check the block number on the screen. The main idea here that I wanted to talk about is of course there are thousands of computers that keep the same record or nodes or and so on. Um if you hold one of these scans you can open the website and check for it.

But what does the match actually tell you? It just simply tells you that it's the same bytes. It does not tell you whether the state actually told the truth, whether our OCR reads the page correctly or whether the archive is actually complete. Um I think Ethereum Foundation worked in 2025 with the US Commerce Department to put a fingerprint of GDB of a GDB release on Ethereum. I assume it has the same limit. A match cannot tell you the number is right. A minister can still lie in the official gazettes that I talked about. What a match actually catches is that someone quietly

swab in a page tomorrow basically. But why why why are we using the fingerprints on Ethereum and not actually the actual scan? We put already some of the text on our wave as um as f files that could be accessed later on if someone gets the private key. If we trust in the new government, we can offer it to them. But a fingerprint uh and not a and not the actual page on Ethereum is because um it holds an address, an ID, um a founder's share, etc. uh blockchain is permanent, but personal data living on a on a permanent ledger cannot be taken back. And that would turn this accountability tool that we're building into something that could

be used for surveillance kind of. Um now imagine if we disappear um someone can still copy our archive, copy our website, copy our pages, manifests and code and rebuild the site from how it is. Even though we're trying to also make sure that we have our data, our scans, our PDFs, all the documents divided across multiple locations and servers. Um, yeah, a state that is coming out of a collapse can still run the same check, including Syria. I believe people have the right to know what their state decides about them, whoever is in power.

Um, I'm running out of time, so I'm gonna move on quickly. Um, to the idea that basically, again, I started with a camera with a question, who holds the truth? Um, I wanted people outside of my city to see what's going on inside my city in terms of the destruction and the killing and the death and the detentions. Um, and I left Syria without my papers in 2014. Today, my work is really about reading what the state publishes about itself and trying to make sure that the next person and the next government does not become the old one, does not become the

dictatorship that we went against. Um, I'm just going to leave it now for questions and uh the links for our project. But thank you so much for staying here and for listening to my talk. Thank you very much, Karam. Yeah, we have a lot of questions. First of all, the audience wants to thank you for telling your story and uh well, it's awesome how strong you keep after all that happened to you. You're inspiring us.

>> Thank you. And uh I believe the most interesting question here is yeah what are you doing if uh the claims or records are conflicting and what do you do about that because as you say well technology can help you here right and you kind of have to handle it yourself. Yeah, exactly. Now, this is a very good question and this is why no have grown into not only trying to figure out what the government is saying about itself but also what the media is saying about the government, what independence media and local media is talking about. So when we have when

we have a claim by the government about something that the governments have done, we usually do triangulation of information and we offer all of it together. So the user, the reader, whatever can also figure out and build up their own uh decision credible neutrality again. So you give all the information and then the user, the reader can make up their mind. We don't try to uh basically analyze or say that this news is fake or whatever. We just aggregate at this point.

>> Okay. So you openly say that we don't know for example. Yeah. And great. Uh another one would be you already told about the architecture a little bit of onchain records. So you're using arviv, you're saving hashes basically. So uh but we would probably want this to be covered as well. How do you use uh AI in this project and uh where AI fails? Uh >> I I talked a little bit about the one example of one bug where AI fails. It mostly fails in our case in reading Arabic because it's not fully trained on Arabic sources and RTL. So it's been really hard try to find the right OCR uh

models for it. So we ended up trying to train our own local models by basically taking an open source OCR model and trying to work on training it using Nvidia and so on. um and giving it all the documents that we have, disconnecting it from the internet to try to reach a stage of like marginal error to be smaller and smaller. But yeah, AI fails in this also like in terms of bias uh and in terms of figuring out uh if this news is said uh in an honest in an honest way. Um it also sometimes brings information that are not or news or media that is not about Syria and it's about a different

country completely. Um so we have to always have a human in the loop uh and someone that checks all the news that are published on a daily basis and also like to reread uh the OCR data after they are being scanned and so on. And I believe our our last >> last question would be so what challenges did you encounter when designing the onchain architecture of the storage volume? Was it easy? Was it hard?

>> It was it was quite easy but it was quite limited as well because I mean we're talking about thousands if not hundreds of thousands of PDFs and and scans and putting it on chain is quite expensive. Um and also like altering or changing or uploading new files in the future is also super expensive. So we resorted into rather than putting the BDFs themselves, the scans and the text files themselves on Arweave. Um and then also like the fingerprints or the finger like the the hashes on on Ethereum the manifests. Um it's complicated. It's not easy to figure out like say in the

future the new government becomes good or it's good and we want to basically give the private key of these f of these folders to the new government or like just keep it away from it. How can we do it? How can we transfer ownership of something like this that lives on chain to a new DAO for example and so on. So all of this is quite still complicated or it's really trying to figure it out. But other than this, putting files on chain has been quite easy and putting the hashes and also like designing a UI UX where you can boot the file and then scan if this manifest actually exists on chain have been quite easy. Odds to you guys and all the developers in the blockchain space who are making it easy

for us. >> Thank you very much Kan. That was