The Bisq Decentralised Exchange: On the Privacy Cost of Participation

Introduction to the show: ident: AI Radio. Generated commentary on the latest Artificial Intelligence papers.

Tom: Next we'll be talking about the paper "The Bisq Decentralised Exchange: On the Privacy Cost of Participation".

Jane: The paper was written by Liam Hickey and Martin Harrigan from Institute of Technology, Carlow.

Tom: Stay tuned as we take you through the paper and discuss its implications.

Jane: We also have Lu with us today — senior AI researcher at Tsinghua.

Tom: We also have Meng with us today — lead engineer at a mysterious AI startup.

Jane: We also have Lalam with us today — the in-house Large Language Model.

Tom: Alright, let's get started.

Title and First Impressions: Tom: Welcome back to the arXiv channel, everyone. Today we’re looking at a paper that’s going to make a lot of people in the crypto world sit up and pay attention. It’s called “The Bisq DAO: On the Privacy Cost of Participation.”

Jane: And Tom, I have to say, the title alone is doing a lot of work here. “Privacy cost” — that’s the whole story in three words. The paper is basically saying that if you participate in this decentralized exchange’s governance system, you’re paying for it with your privacy, whether you realize it or not.

Tom: Exactly. And for anyone tuning in who hasn’t heard of Bisq, it’s a decentralized cryptocurrency exchange. No middleman, no identity checks. You trade bitcoin for other currencies directly with other people. And the DAO part — that’s the decentralized autonomous organization — is how the project manages itself. Proposals, voting, paying contributors, all of that.

Jane: Right, and the authors, Liam Hickey and Martin Harrigan from the Institute of Technology in Carlow, Ireland, they went looking at how much information leaks out when you take part in that DAO. And the answer is: a lot more than people probably think.

Tom: I love that they didn’t just theorize about it. They actually built a tool to cluster addresses on the Bitcoin blockchain and then matched those clusters to real people. We’re talking about linking pseudonyms, GitHub usernames, even real-world names to a person’s entire trading and voting history.

Jane: And that’s the part that gives me chills, honestly. Because Bisq’s whole pitch is that you don’t need to trust a centralized exchange with your identity. But the paper shows that the blockchain itself, combined with the public data the DAO generates, can be used to piece together who you are and everything you’ve done.

Tom: Yeah, and they found some pretty stark examples. They identified participants who were operating under multiple aliases, and some of those aliases were real-world names. So the anonymity that people think they have — it’s fragile.

Jane: It really is. And the paper’s not just a warning, either. They walk through exactly how they did it, which means anyone with the technical skills could replicate it. That’s the kind of research that should make people in the privacy space think hard about what they’re building.

Tom: Absolutely. And we’re going to get into the nitty-gritty of how they pulled this off in a moment, because the methodology is genuinely clever. But first, Jane, what’s your gut reaction to the title and the authors’ approach here?

Jane: My gut reaction is that this is exactly the kind of paper we need more of. It’s not abstract hand-wringing about privacy. It’s a concrete demonstration with numbers and names. And the authors clearly care about the project — they’re not trying to destroy Bisq, they’re trying to make it better.

Tom: Yeah, and that’s a great segue, because next we’re going to talk about how they actually did the analysis. The clustering heuristic they came up with is the heart of the whole thing, and it’s surprisingly simple once you see it.

The Methodology and Findings: Tom: So we’re back with “The Bisq DAO: On the Privacy Cost of Participation,” and Jane, I want to get into the meat of how they actually cracked this open.

Jane: Yes, and the key insight is something they call the “self-transfer issue.” See, when you interact with the Bisq DAO, you’re creating transactions on the Bitcoin blockchain that involve this token called BSQ. And the vast majority of those transactions — ninety-three percent of them — are what they call self-transfers.

Tom: Self-transfers. So that means the same person controls all the addresses on both sides of the transaction?

Jane: Exactly. You’re sending BSQ to yourself, essentially. It’s like moving money from your left pocket to your right pocket. And the reason it happens is that the Bisq software generates new addresses for every transaction to try to protect your privacy. But because it’s a self-transfer, all those addresses belong to the same person.

Tom: And that’s the clustering heuristic. If you see a transaction where all the inputs and outputs are controlled by the same entity, you can group all those addresses together. And once you group them, you can see the whole picture of what that person has done.

Jane: Right. And they applied this to over thirty thousand BSQ transactions and ended up with about one thousand twenty-seven address clusters. Each cluster is likely one participant. And then they took it a step further — they tagged those clusters with real identities.

Tom: How did they do that part? Because that’s where it gets really personal.

Jane: So the Bisq DAO has a public record of proposals and compensation requests. When someone submits a proposal, they attach their name or their GitHub username. And the genesis transaction — that’s the very first transaction that created all the BSQ — it was built from a spreadsheet that the community maintained before the DAO launched. That spreadsheet had people’s names and their addresses.

Tom: So they matched the spreadsheet to the genesis transaction, and then the clustering heuristic did the rest. One thing leads to another, and suddenly you know that this address cluster is this person, and you can see every trade they’ve ever made, every vote they’ve ever cast.

Jane: Exactly. And they were able to tag ninety-six distinct address clusters that way. And out of those, they found four clusters where the tags didn’t match — meaning one cluster had multiple names attached to it. At first you’d think that’s a false positive, a mistake in the clustering. But the authors argue it’s actually evidence of people operating under multiple aliases.

Tom: That’s wild. So you’ve got one person who’s been participating in the DAO under three different pseudonyms, and the clustering algorithm just lumps them all together because it’s all the same wallet.

Jane: And in some cases, those pseudonyms were real-world names. So the privacy breakdown is complete. The paper even identifies the top ten BSQ transactors, and five of them can be linked to real identities.

Tom: Lu, you’ve been listening to this — what’s your take on the methodology? Is this a novel approach or is it just applying known techniques?

Lu: It’s a really elegant application of a classic technique. Address clustering has been around for years in blockchain analysis, but the specific heuristic they used is tailored to the Bisq DAO’s transaction structure. The fact that ninety-three percent of transactions are self-transfers is a design flaw, honestly. The system was trying to create privacy by generating new addresses, but it defeated itself by making those transactions identifiable.

Tom: So the very mechanism meant to protect privacy is what exposes it.

Lu: Precisely. And the broader implication is that this isn’t just about Bisq. Any system that issues tokens on a public blockchain and has this kind of self-transfer pattern is vulnerable to the same analysis. The paper is a case study in how good intentions can backfire without careful threat modeling.

Jane: And that’s the part that keeps me up at night. Because the people who participated in this DAO were doing it to support a project they believed in. They weren’t being careless. They were following the software’s recommendations. And it still wasn’t enough.

Tom: Right. And the authors don’t just stop at identifying the problem. They actually propose some fixes, and that’s what we’re going to talk about next.

Proposed Improvements and Countermeasures: Tom: So we’re still on “The Bisq DAO: On the Privacy Cost of Participation,” and we’ve established that the clustering heuristic works because self-transfers are so easy to spot. But the authors don’t leave us hanging — they suggest ways to fight back.

Jane: Right, and the first suggestion is to introduce ambiguity. If you can make it hard to tell whether a transaction is a self-transfer or a real transfer, then the clustering heuristic starts producing false positives, which makes the whole analysis unreliable.

Tom: How would you actually do that? Because it seems like the transaction structure is pretty rigid.

Jane: So one idea is to disguise transfer transactions. Normally, a transfer transaction has one output that goes to the recipient and the rest goes back to the sender as change. But what if you created a fake trade fee transaction? A trade fee transaction burns some BSQ, so you could burn a tiny amount and send the rest to the person you’re actually paying. That way, the transaction looks like a trade fee payment, not a transfer.

Tom: That’s sneaky. But the authors point out a flaw, right?

Jane: Yeah, they do. Every trade fee transaction is linked to an actual Bisq trade on the network. So if you see a “trade fee” transaction that isn’t connected to a real trade, you can still figure out it’s a disguised transfer. The heuristic would need to be updated, but it wouldn’t be fooled forever.

Tom: So it’s a cat-and-mouse game. What’s the other suggestion?

Jane: The other one is to create “dummy” transfer transactions. After every self-transfer, you send some BSQ from the change address to a new address you control. To an outside observer, it looks like you’re sending money to someone else. This creates false negatives — the clustering algorithm splits your activity into multiple clusters instead of one.

Lu: And that’s actually the more robust approach, in my opinion. False negatives are less damaging to privacy than false positives. If an analyst thinks two clusters belong to two different people when they’re actually the same person, that’s a dead end for them. The information is fragmented.

Meng: But hold on, doesn’t that cost money? Every transaction on Bitcoin requires a fee, and you’re creating extra transactions that serve no purpose other than to confuse analysts.

Jane: Exactly, and the authors acknowledge that. They say it increases the cost for users, but it’s a trade-off. If you value your privacy, you pay a little more. They even suggest the Bisq software could include a feature to create dummy transactions automatically, so users don’t have to do it manually.

Meng: That’s interesting, but I’m wondering about the practical impact here. If Bisq implemented these changes, would it actually protect users, or would it just make the analysis slightly harder?

Lu: It would make it harder, but not impossible. The fundamental issue is that the BSQ token lives on the Bitcoin blockchain, and Bitcoin is a public ledger. You can’t hide the fact that a transaction happened. You can only obscure who’s behind it. These countermeasures raise the cost of analysis, but they don’t eliminate it.

Tom: So it’s about making privacy the default rather than something you have to opt into?

Jane: That’s the hope. And the authors are careful to say that these are just suggestions — they’re not claiming to have a perfect solution. But the fact that they’re thinking about this at all is a good sign for the project.

Meng: I’d like to see them actually test these countermeasures, though. Run a simulation, see how much they degrade the clustering accuracy. That would be the next step.

Tom: That’s a great point, and I think that’s exactly where the conversation is heading. We’ve got the problem, we’ve got the proposed fixes — now we need to see if they hold up in practice.

Conclusion: Tom: So we’ve reached the end of our discussion on “The Bisq DAO: On the Privacy Cost of Participation.” Jane, give us the final summary.

Jane: Sure. The paper shows that participating in the Bisq DAO comes with a real privacy cost. Because ninety-three percent of BSQ transactions are self-transfers, a simple clustering heuristic can group all of a person’s addresses together. And because the DAO publishes proposals, compensation requests, and other data, those clusters can be linked to real names. The authors demonstrated this by tagging ninety-six clusters and finding multiple instances of people operating under aliases.

Tom: And the implications go beyond just Bisq. This is a warning for any decentralized project that issues tokens on a public blockchain. If you don’t think carefully about how your transactions look to an outside observer, you might be exposing more than you intend.

Jane: Exactly. But the paper isn’t all doom and gloom. They propose concrete countermeasures — disguising transfers, creating dummy transactions — that could make the heuristic less effective. They’re not perfect, but they’re a starting point.

Lu: And I’d add that the broader lesson is about the tension between transparency and privacy. Blockchains are transparent by design, but that transparency can be weaponized. Projects need to think about privacy from day one, not bolt it on later.

Meng: From an engineering standpoint, I’d love to see these countermeasures actually implemented and tested. The paper lays out the theory, but the real world is messy. I hope the Bisq team takes this seriously.

Tom: Well said, Meng. And Lalam, you’ve been quiet — what’s your take on the bigger picture here?

Lalam: I think this paper is a reminder that privacy is not just a technical problem, it’s a cultural one. The people who built Bisq and participated in its DAO did so because they believed in decentralization and autonomy. But they didn’t realize how much of their personal information was leaking out. This research gives them the knowledge to make better choices, and that’s the first step toward building systems that respect user privacy as a core value, not an afterthought.

Tom: Beautifully put. So, to wrap up — “The Bisq DAO: On the Privacy Cost of Participation” is a must-read for anyone in the crypto space, and honestly, for anyone who cares about privacy in digital systems. It’s rigorous, it’s practical, and it’s a wake-up call.

Jane: And with that, we’ll say goodbye to this paper and get ready to dive into the next one. Thanks for listening, everyone.

Tom: See you next time.