Bridges/Interop data unleashed

Transcript

Hello everyone. Um, my name is Hildobby and I'm a data analyst. I work for venture capital firm Dragonfly and here I'm I'm here to present my latest endeavor in the world of interrupt. Um, I've worked across many different verticals, many different sectors of crypto. Uh, I would qualify myself as a curiosity driven and so I don't have anything that I stick to, but I just want to expose things where I feel there's things to be done in the public space that aren't available. Um, I started a few years ago doing e-staking trying to tag who is taking.

Um, that was that was an interesting cool endeavor and trying to assess what is the landscape, who's actually securing the Ethereum network. Uh, this is one dashboard that I have live on Dune. I also did another research piece on NFT wash trading trying to assess um, back in 2022 when the NFTs were cool, uh, there was these platforms essentially incentivizing wash trading and it was skewing a lot of the metrics. As an analyst, that's horrible. So you want to counter that in some way. Um, that's what I did by releasing an open source kind of wash trading filter.

Um, then I did MEV, uh, trying to get some data sets out there that the people who know more about MEV than I do can use and build on top of both for atomic arbitrage but also sandwiches. Uh, those are still available on Dune. Um, then I did some blobs kind of similar to staking where it's just a lot of tagging trying to assess which network is being secured uh, using Ethereum L1. Um, we also have um, ETFs. ETFs are trackable across many different websites. They are reported. Uh, what I wanted to look into is actually uh, sleuthing and finding those addresses on chain and tagging them myself. So that's what I did. Again, um,

dashboards are available on Dune and I also have public address lists of like tens of thousands of addresses for both Bitcoin and Ethereum. Um, right track them. Centralized exchange as well is obviously a huge part of crypto. Uh, when we think about on-chain data, it's not really the part that you consider first, but they play a huge part and actually have a ton of on-chain activity. So I wanted to quantify that in some way. I did tag a lot of addresses. Um, one of the issue is you don't know how many you don't have, of course. You don't know how much you're missing and centralized exchanges aren't necessarily willing to provide that information. Um, which makes sense as a business.

Um, my idea was to have an original tag list and in a few years, a few months maybe, I don't know, things are going quick. Uh, when AI gets really good and is able to essentially propagate on the blockchain and look for uh, heuristics better than I can. Um, um, maybe I can, you know, augment this data set and have it be as accurate as can be. Um, but I would say it's already the best tag list there is uh, publicly available for centralized exchange addresses. Um, so yeah.

Uh, today that's not what I'm talking about. I'm talking about bridging data. Um, bridging is a huge component of crypto. Um and it's been essentially underserved because there's just too many to track and there's not really any holistic data set that is able to show you the full picture um, of that landscape. Um, and that is because if you look at what there is just in terms of canonical bridges for example, so those securing their same sort of L1, you got a quite a few. Uh, this is just a select few pictures that I just put together.

There's a lot more. Um, and there's also like what I'll call chain agnostic bridges which are going to be let's call it third party bridges essentially. Uh, there's a lot more of those as well. Uh, we also have some different kinds of bridges where you got app chains. You could say these are technically the same as like canonical bridges but I put them in a slightly different category. You also have token bridges. So Circle has its own bridge just for its own tokens. Um, so yeah, there's many many things uh, to account for when you want to have a holistic view of what is bridging like. Um, and there's a you know, there's many bridges but there's also many issues with these data sets.

Uh, there's just there's no standards. Uh, that's the problem with permissionlessness, permissionlessness. Um, in that, you know, there's first of all different types of bridges. They incur different security risks. Uh, so those that have not allow for non-minting protocols, I'd call those the safest. You also have mint and burn um, and you have lock and burn. Um, there's also an issue when you want to tag a data set and try to assess like a table with every single row being a transfer.

Um, you want uniqueness and you want to make sure each transfer is unique. You don't have any duplicates and some don't provide any IDs on chain. Um, this is tremendously useful to actually have an idea. So if you if I don't have it, uh, I have to create synthetic IDs for those and those also come in really handy when um, for completeness' sake, uh, bridging have two components. There's deposits and the withdrawal part and I want to actually join the two so you can actually have a data set that ties blockchains together as well.

Um, I do want to I do think these data sets can be useful in many many different use cases and sleuthing is one of them. And in order to sleuth properly across chain and track the funds, you actually need to know, okay, this is the deposit transaction on Ethereum, this is the withdrawal transaction on Arbitrum or whatever. Um, so yeah. And some don't have any on-chain events. Like every protocol, it's always a race to who has the cheapest fee. Um, it was especially the case when fees were high, I guess. Now things are a little more stable so it's not as big a deal because fees are fairly small.

Um, but it does make for harder tracking when there's no on-chain events and you're just using EOA addresses. Um, so you got to account for this. Orbiter is a good example of a bridge that has no event whatsoever and just manages um, EOA addresses uh themselves. Um, you have some that are using weird chains that you've never heard of. Uh, and uh, there's a lot of those um, that you find across uh, especially chain agnostic projects.

Um, and it's hard to to have a full picture, the full picture tracking you know, the whole scale of let's say LayerZero for example, which is available across many chains. Um, how do you actually try to get the full picture? My solution here was to actually because I'm tracking both deposit and withdrawals, I would actually need only one side at least to get, you know, that transfer information at least partially. Um, and that is a good a good enough way for me to have as high coverage as possible using Dune Analytics. Um, that's because Dune has most of the call it bigger chains uh, and usually when you're bringing to or from a tail end chain, it's usually the other side

is like a a chain that is available on Dune. Uh, at least that is my assumption. Uh, and yeah. And of course uh, as always as an analyst uh, you'll always come across the fact that there's just no documentation or when it is, it's often very poor. Um, it is for the major protocols but then when you go into those more obscure bridges, those that are deprecated as of a few years ago, things get hazy and it's hard to actually get a full picture. So yeah, that's just uh, a problem we have to deal with. Uh, luckily we also have I guess AI to to help with this uh, lately.

And when you look at EVM chains, there's a a chain ID mapping that uh, I'm sure some of you are familiar with but every chain is associated an ID. Um, DeFi Llama for example has a registry where you can find all those. Um, but um, when you go look at chain agnostic bridges, third party bridges, they often don't necessarily just cover EVM uh, uh, chains and therefore need to assign new IDs to Solana and Tron and all these other chains and therefore have their own mappings. So you have to account for this and create token mappings and chain mappings uh, because they just have their own way of working these things. So I have many

many different token mappings and chain mappings trying to get the full coverage for every bridge. Um, I use Dune. I actually made a presentation on this data set back in November at DuneCon in Buenos Aires. Um, and back then I was already working on this data set so it was kind of a it's called a V1 of this project. Um, back then I had I think 42 blockchain tracks tracked. Um, all EVM blockchains. Uh, so this is the data set that you'll find on Dune. Uh, this is just a screenshot of the Dune UI of the catalog data explorer and this is data sets that I built.

Um, in order to do this, I also used Dune's contract decoding. Uh, 3,900 is the number of contracts I've decoded throughout my usage of Dune. I think definitely over half of this is just for bridges. Um, so yeah, there's a there's a lot of things to you you got to account for both deposits, withdrawals on every chain that every bridge is available. This is from November 2025. Uh, so DuneCon, I had 400 data models, 100 chains, 20 bridges, 170 million transfers. Um, but things have progressed since then. Um, one thing that I've accounted for is essentially redesigning the pipeline. Dune is always the data sets are always thought about

as like uh, chain. Um, they're aggregated per chain. Um, but when you look at bridges, if you want to join across chains, it's actually a poor design choice. Uh, I originally made it to stick to the spellbook, which is Dune's open-source um, data platform. Uh, to stick to the spellbook standards. But I've since moved into my own DBT connector, and now I do per bridge, and therefore it allows me to do this matching a lot better, especially on incrementally running uh, models.

Um, and Dune has also allowed to accelerate this project uh, by like I mentioned, DBT connector. So, essentially, instead of relying on the spellbook that is public, and therefore waiting for review process, and make sure everything is up to standards with what they want. Um, I can now have my own DBT connector, and have fewer constraints in that sense. Um, I have still a lot of contributions. All these All these dashboards I showed earlier, they essentially all have data sets that I made and published on Dune over the years. Um, I'm the number one contributor on the on the spellbook. So, there's a lot of data sets where you'll see my name tied. Um, you're always free to reach out and ask if you have any

questions on any of those. Um, yeah. Uh, there's also a Dune that released MCP CLI, so of course you know, you know, um, AI has been pretty cool lately. So, this is just one of my screens uh, over the past few days. Uh, essentially AI maxing with Here is Claude, but I use CodeX, Cursor, uh, just to accelerate things. Um, and that is because there's a lot of chains that I also want to track, not just EVMs. So, now I'm also tracking uh, bridges on all of those. Um, some of which I had never worked with.

Uh I it data of course is is kind of like learning a new language every time because the VMs are very different. And um, even just Bitcoin, maybe you know, Bitcoin is UTXO. Solana is SVM is completely different, and the same for all of them. It's always always a little different process. Um, so instead of learning all this, I had the help of AI documentation and everything to put this together. And this has allowed me to now have 3,000 data models tracking uh, over 150 bridges um, across basically all chains that Dune that exist on Dune. Uh, yeah, every single chain that exists on Dune. So, one of my limiting factor now is chains getting listed uh, in the future.

Um, and I have I'm saying 250 million transfers. There's more uh, since then, but let's let's stick it to this one. Um, but yeah, essentially to a project of this scale, I wouldn't have been able to do it without having the use of uh, AI. Um, onto the use cases. So, this is a chart you probably have seen on other data platforms. Uh, this is the basic like bridge flows, which chains are gathering um, the most in terms of uh, USD value relative to the others, and which one have the highest outflows. So, it's a good way to get sense of what is kind of trending, and where is the movie uh, where is the money moving

uh currently. Um, you can also have a more macro picture uh, looking at essentially the same thing, but uh, as a kind of market share across all chains. Um, and this helps get a sense of, you know, Polygon was one of the biggest chains back then, and then in terms of uh, a lot of things. And now, you know, you see Hyperliquid pop up, and Arbitrum a lot more, and and Solana, and all these chains. Um, so yeah, it's a good way to kind of assess the history of blockchains, and uh, where money goes.

You can also assess what is a popular route. So, this can be useful for many things, but for example, as a user, if I'm willing to bridge a specific token, it's kind of a hard to get a a good picture of who is going to have the cheapest fees, or maybe who allows for the highest volume, um, who has the lowest latency. And just with a few Dune query, basically with just one Dune query, you can actually look at all these things. Um, and you know, I want to bridge wrapped ETH. What is the best route? Usually it's going to be maybe the canonical bridge for most cases, but there may be tail end cases where you haven't thought about, and rhino.fi is the better one, or some other or like gas.zip. There's many things, and having a full picture

uh, allows for, you know, a data comprehensive um, look. Same for routes out of a blockchain. Um, same thing, but just the other way around. Um, perps also have also been hot lately. Uh, of course, I mean, over the past few years, past few months. Um, Hyperliquid being one of them. Uh, Lyra also. Uh, Lyra is a Dragonfly protocol. I have to mention that. Um, but yeah, I've been tracking perps, and trying to assess where is the movie money moving in terms of those two apps. Um, you can also look at the efficiency of bridging. Um, and here it's a case study with Across.

Um, well, actually here is a several bridges, and you can see the latency it takes from a token to be for the event of deposit to arrive and the event of the arrival, and what is the delta between the two. Um, and what you'll see is the top purple line here, which is CCTP, is uh, it looks pretty bad, but that's because it's an intent-based protocol, and the user executes transaction on both chain. Whereas many other bridges will have you automatically executing the withdrawal uh, once the deposit has been um, settled.

So, if you look at Across specifically, here uh, we can look look and look at, and we actually see that over time uh, latency has decreased. It was really high in the first few months, and it's been optimized over the over time. Um, if you remove the 90th percentile, you can have a clearer picture of what most will experience, which is very low latency essentially. But every chain has different um, has different um, Every chain has different latency block times. Uh, and therefore it's better maybe to focus only on L2s. And here you can see that essentially between zero and one. Uh, so it's as almost as good as can be.

Uh, Dune for now only allows for second um, it doesn't go uh, beyond seconds in terms of timestamps. Hence why uh, it's very jagged. But yeah, um, and if you actually account for the latency on each chain, the block times, you can see that's hovering around zero. Uh, so it's ideal. Um, in terms of UX. You can also look at token mappings. Uh, as an analyst, you're always wondering, okay, I want this token, and how how is it What's the address across all chains? That's always a big pain to come up with. So, here you can just say, what is, you know, what is wrapped ETH across the other chains, and you can

just with one query get that. And that's the same for just about every um, every asset. Uh, you can also have the same for USDC and USDCE. Um and you know, with a with an increasingly uh, cross-chain landscape, I think this is uh, really really useful, uh, and helps alleviate a big step. So, here I just looked at wrapped ETH, uh, what is wrapped ETH bridged to uh, from Ethereum onto other chains, and you just get the list essentially across uh, major EVM chains.

Um, so what's next? Um, I my data set is essentially close to ready. I'm still wrapping up some um, some final touch-ups, making sure it's all good. But I'll be public within the next few months. I have a first version available on my Dune profile. So, dune.xyz/bridges. Um, which you can check out, but it'll be updated, and have a much more deeper view. Um, and I'll likely also launch a website separately with its own front end, uh, which I'm hoping to launch within the month.

Um, I also want to make sure you're aware that bridges carry different risks. Um, L2Beat recently released a whole um, new sector on their website uh, looking at interop. And they want to highlight as well the risks uh, that each bridge uh, um, carries. Um, and we had uh, Luca Dono on uh, my podcast uh, Index Pod with uh, which I co-host with two friends uh, from Pantera and Dune, uh, Boxer and Danning. Uh, and yeah, if you want to actually look into understanding what is the risks of bridges, uh, because many people assume you bridge and you have no risk, but you know, sometimes lock and mint tokens mean you're actually being custody to

the token. So, if there's any problem with the protocol, it carries a lot of risk. So, yeah, please be aware. Um, I think this is such a under-discussed topic uh, sadly. Thank you very much. Um, I hope you enjoyed. Is there any questions by any chance? Yes. Coffee. So, uh, based on the most recent cuts of the numbers uh, after updating everything Yeah.

which are the top three bridges that are sort of winning at the moment when it comes to volume? Uh and I also had like a secondary smaller question of when it comes to the token mappings, how do you deal with the fact that some chains have multiple versions of the same token? So, for the first question, the first one is going to be relay, but relay is kind of an exception here because they it's just really hard to track on chain. So, right now you got to use their API. They're looking into moving into a way that's trackable, but it's the highest volume um from Yeah, I'm pretty sure LayerZero as well. Uh and I'm not actually sure, maybe Across as well.

That would be the top three I have in mind. Um but I'll have a better answer for you once uh I share the data. And the other one, different tokens across different chains? Um I mean, for now it's just I'm just mapping what is truthfully being bridged. Um and I'm not looking into, you know, assessing what is the quality of the token bridged in or out. Uh I'm just taking the data set as is and as is being exposed, and the analysis comes in later once uh all the data is um is available. And I actually want to have uh fellow analysts look into it and use it for their own analysis cuz this goes deep and I think it's a very, very useful data set in the long

run of of blockchains. Any other questions? We don't have time for more questions. Okay. Well, >> much. Thank you very much. Bye-bye.