Using two layers of mixers to further obfuscate address history

[NotATether]

As some of you know, it's not recommended to run a Tor connection through another Tor connection (double connections), because weaknesses in the protocol allow attackers to deanonymize your traffic.

I was wondering if the same concept also applies to mixers. In other words, if coins are sent through one mixer, then through a different mixer, can they be identified using weaknesses such as this? Does it act as a buffer to protect transaction privacy in the event one of the mixers is deanonymized?

[pooya87]

I can't think of any scenario where weakness in one mixer could lead to breaking the gained privacy of the other. The two mixed coins are not connected in any ways (assuming mixers aren't connected, like owned by one company or having the same vulnerability), and when trying to break the mixing transactions it doesn't matter what the source of the coins were. Whether they came from an exchange with KYC, or your wallet or from another mixer, the difficulty of breaking it is the same.

[o_e_l_e_o]

I am considering a scenario where I have sent coins to ChipMixer, and now instead of withdrawing the chips to either send the coins back to my own wallet or to pay for a good/service/other transaction, I instead send them to a second mixer. Even if the second mixer is horrendously buggy or filled with weaknesses, I don't see how any of those bugs or weaknesses could decrease the privacy I have already gained from ChipMixer. The only scenario I can see which might decrease my privacy is that if I am mixing coins a second time, I am probably far more likely to withdraw all my chips at once and send to the second mixer. If I deposit 0.5 BTC to ChipMixer and then withdraw different amounts at a time over a period of days or weeks, my privacy is much better than if I withdraw all 0.5 BTC in one transaction shortly after I deposited it.

[ranochigo]

It would be futile to try to achieve anything if there is any form of privacy leakage while using the second mixer. If either your browser or any modules within your computer is leaking any personally identifiable information, be it having DNS leakage, Torrent leaks, WebRTC, then you should assume your privacy to be dependent only on the second mixer.

[PrimeNumber7]

I was wondering if the same concept also applies to mixers. In other words, if coins are sent through one mixer, then through a different mixer, can they be identified using weaknesses such as this? Does it act as a buffer to protect transaction privacy in the event one of the mixers is deanonymized?

Generally speaking, after the fact, mixer addresses are known to blockchain analysis firms. So if you send coin to a mixer and receive coin from a mixer a period of time later, a blockchain analysis firm will be able to connect both transactions to the mixer. If the mixer sends coin back to you on a not-consistent delay, takes a not-consistent fee, and/or sends coin to you in multiple transactions, it may not be clear that the send and receive transaction(s) are associated with the same person.

My concern is that if on a consistent basis, you send coin to mixer "A", and after x period of time, you consistently send all of the coin you have received to mixer "B", you may give an adversary information they would not otherwise have access to. If mixer "A" does a very good job at obfuscating the link between the "in" and "out" transactions to/from the mixer, but mixer "B" does a poor job, you may end up worse off in terms of privacy.

The above means you need to use the mixer that is best at maintaining privacy last.

If one mixer is actually a government honeypot, or information about its users is otherwise leaked, and you use it last, the privacy you received from the first mixer will be reduced.

[o_e_l_e_o]

If one mixer is actually a government honeypot, or information about its users is otherwise leaked, and you use it last, the privacy you received from the first mixer will be reduced.

I take your point about consistently sending coins from mixer A to mixer B - doing so is a fairly unique transaction and so potentially links all your mixed coins together. But can you elaborate on the part I've quoted above? I assume when you say "information about its users" you are talking about things like IP addresses, browser fingerprints, etc? If so I understand, but if you are referring to blockchain or transaction information, then I don't follow your line of reasoning. If I withdraw (for argument's sake) 0.256 perfectly anonymized bitcoin from ChipMixer, send it to government honeypot (using a perfectly anonymized Tor session), and then receive a different 0.256 BTC from the government to my own wallet, what additional information have they gained?

[PrimeNumber7]

If one mixer is actually a government honeypot, or information about its users is otherwise leaked, and you use it last, the privacy you received from the first mixer will be reduced.

I take your point about consistently sending coins from mixer A to mixer B - doing so is a fairly unique transaction and so potentially links all your mixed coins together. But can you elaborate on the part I've quoted above? I assume when you say "information about its users" you are talking about things like IP addresses, browser fingerprints, etc? If so I understand, but if you are referring to blockchain or transaction information, then I don't follow your line of reasoning. If I withdraw (for argument's sake) 0.256 perfectly anonymized bitcoin from ChipMixer, send it to government honeypot (using a perfectly anonymized Tor session), and then receive a different 0.256 BTC from the government to my own wallet, what additional information have they gained?

I was referring to blockchain information. For example, a mixer may (intentionally or otherwise) keep track of what should be the private link between a transaction sending coin to a mixer and the transaction from the mixer.

In your example, you withdrew 0.256BTC from CM after sending a similar amount to CM. My example stipulates that you sent the 0.256BTC from CM to a honeypot multiple times, each time after the same approximately same interval after you sent a similar amount of BTC to CM.

If you do the above one time, there will be many input transactions in which you are sending 0.256BTC to CM. If you do the above five times, the number of potential input transactions irrelevant to you is reduced. Eventually, if you do the above with sufficient frequency, an adversary will be able to reasonably conclude all transaction both to and from a mixer you used.

[o_e_l_e_o]

-snip-

Sure, but I would argue that if you repeatedly perform the same deposits and withdrawals of the same amount with the same delay and the same transaction heuristics then you are going to reduce your privacy, regardless of whether or not you are withdrawing your coins back to your own wallet or to a second, compromised mixer. Using a honeypot might speed up the process of identifying all your transactions for the honeypot operators, but if you repeat the same behavior enough then anyone who can read the blockchain would be able to link your transactions with a reasonable amount of certainty.

[hugeblack]

Since this process is done consecutively and not in parallel, each additional stage will reduce the efficiency of privacy instead of increasing it, especially since most of the mixers contain defects^[1], either the absolute trust in the mixers (for example Chipmixer) or the ease of identifying the source of the coins (Wasabi wallet)


[1] https://bitcointalk.org/index.php?topic=5117328.0

bitmixer.io & coinmixer.se are offline now, however its still possible to use the bugs I describe in my thesis to reverse nearly all transactions which have ever been processed by these services.

[ranochigo]

Since this process is done consecutively and not in parallel, each additional stage will reduce the efficiency of privacy instead of increasing it, especially since most of the mixers contain defects^[1], either the absolute trust in the mixers (for example Chipmixer) or the ease of identifying the source of the coins (Wasabi wallet)

Doesn't this reinforce the fact that we need to use multiple mixers?

It depends though. As far as possible, you should assume that an adversary, if given enough resources can and will expose your identity regardless of the precautions that you take. Unfortunately, there is no way to tell whether blockchain analysis on your TXes are possible, or if your mixers are a honeypot. I would probably prefer using a privacy coins over this, or at the very least choose a mixer that I know and can trust. Else, I'd like to assume that the first mixer would be the most important link and the rest are merely supplementary. It shouldn't really decrease the privacy beyond that provided by the first mixer though.

[PrimeNumber7]

-snip-

Sure, but I would argue that if you repeatedly perform the same deposits and withdrawals of the same amount with the same delay and the same transaction heuristics then you are going to reduce your privacy, regardless of whether or not you are withdrawing your coins back to your own wallet or to a second, compromised mixer. Using a honeypot might speed up the process of identifying all your transactions for the honeypot operators, but if you repeat the same behavior enough then anyone who can read the blockchain would be able to link your transactions with a reasonable amount of certainty.

If someone is not using two mixers, they might not otherwise use a similar time interval to mix their coins. Also, once a person's coin has passed through (a) mixer(s), they might spend their coin at a diverse set of merchants.

So using a second mixer may be providing additional data points to an adversary who is trying to break a user's privacy.

[o_e_l_e_o]

If someone is not using two mixers, they might not otherwise use a similar time interval to mix their coins. Also, once a person's coin has passed through (a) mixer(s), they might spend their coin at a diverse set of merchants.

Yes, that was the point I made in my first reply. If I deposit 0.5 BTC in ChipMixer, and then withdraw a variety of smaller chips over a period of days, weeks, or even months, as and when I require them to spend directly with a merchant or other third party, without ever withdrawing them in to my own wallet, then my privacy is far greater than if I deposit 0.5 BTC and then immediately withdraw 0.5 BTC. If I repeat that action of depositing and immediately withdrawing the same amount, then my privacy is lessened further. If I always withdraw the same amount to an address associated with the same service, be that another mixer or something else entirely like an exchange, then my privacy is lessened further still.

All things considered, I think you are far better sticking with a single mixer. If you are at the stage of considering combining two or more mixers or other privacy techniques, then you would probably be better off just using Monero.

[dkbit98]

As some of you know, it's not recommended to run a Tor connection through another Tor connection (double connections), because weaknesses in the protocol allow attackers to deanonymize your traffic.

Bitcoin obviously works totally different from Tor so we can't compare this two, but mixers are usually centralized and they can store IP addresses and can fingerprint your browser.
Most mistakes are made by people reusing addresses but if we exclude that, and combine Tor, vpn and different browsers with mixers, that would be enough to break any possible connection between addresses.
One thing I am thinking is that new analytics tracking services could potentially spot that you have been using multiple mixers easier, but this is just my speculation.

[o_e_l_e_o]

Most mistakes are made by people reusing addresses but if we exclude that, and combine Tor, vpn and different browsers with mixers, that would be enough to break any possible connection between addresses.

Using Tor with a different circuit each time you deposit, withdraw, or otherwise interact with the mixer is probably better than using a VPN and a variety of browsers. VPNs don't guarantee anonymity by any means, and really you are just shifting your trust from the mixer provider to the VPN provider. Other browsers are also not hardened in the same way Tor is to avoid leaking information which may identify you, such as your real IP address via WebRTC even if you are using a VPN - Tor stops this by having NoScript installed automatically. As soon as you start customizing any browser, be that with add-ons, themes, even resizing the window, you start making yourself more identifiable. Using a clean install of Tor with nothing changed is your best bet to both protect your information from leaking, as well as keeping your browser as common as possible.

[ABCbits]

Most mistakes are made by people reusing addresses but if we exclude that, and combine Tor, vpn and different browsers with mixers, that would be enough to break any possible connection between addresses.

Aside from what @o_e_l_e_o said, combining Tor and VPN have some privacy concern. Tor project made an article about it at https://gitlab.torproject.org/legacy/trac/-/wikis/doc/TorPlusVPN.

Tor stops this by having NoScript installed automatically.

These days, it's only true if you choose "Safer" or "Safest" security option. NoScript on default security option ("Standard") only block few things.

[o_e_l_e_o]

These days, it's only true if you choose "Safer" or "Safest" security option. NoScript on default security option ("Standard") only block few things.

Are you confusing NoScript with Tor's built in security features?

If you go to Options -> Privacy and Security in Tor, you do indeed get the Normal/Safer/Safest check boxes for Tor's own security features, with only the Safer and Safest options disabling JavaScript on non-HTTPS or all sites, respectively. NoScript's JavaScript blocking should override these, but I'm not sure if that option is enabled by default or if the user has to turn it on.

Looking in to it a bit more, it seems that Tor sets media.peerconnection.enabled to false by default, so WebRTC isn't actually a concern when using Tor (unless you choose to enable it for some reason). My original point still stands, though, that a fresh install of Tor which you do not then customize or make unique in any way is your best bet for avoiding browser fingerprinting or leaking other data.

[o_e_l_e_o]

No, Tor's built in security features is tied with NoScript.

Huh, TIL. Thanks!

If you check NoScript Advanced tab, you'll see "Override Tor Browser's Security Level preset" option isn't enabled by default.

I've had everything turned up to max and JavaScript disabled unless I choose otherwise for so long that I couldn't remember what the default options were. I have the override option enabled so I can manually whitelist the few sites I need to allow JavaScript on. Keeping your Tor browser security settings on "Safest" to disable JavaScript by default plays in to what I was discussing above about revealing as little as possible via JavaScript to make yourself stand out.