‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐ On Thursday, September 10, 2020 12:19 AM, <jamesd@echeque.com> wrote:
On 2020-09-10 02:44, Lee Clagett wrote:
A Monero contributor/developer for a few years now ...
There are lots of cryptographers vastly better than I am, but they tend to suffer from the mighty unbreakable fortress wall syndrome. They build crypto that is utterly unbreakable against the threat as defined, and all the ways around their unbreakable wall are declared to be out of scope.
No actual critique here, some vague accusation of ivory tower engineering.
Monero's blockchain necessarily leaks quite a bit of information. Perhaps the information is completely useless to an attacker. It is certainly useless to an attacker who attempts to to figure out who is transacting with whom by performing the attacks specifically defined and addressed. Monaro is clearly invulnerable to the bitcoin problem.
Invulnerable -> "impossible to harm or damage". "Monaro is clearly invulnerable to the bitcoin problem" ... ? What does that even mean in this context? You seem to be stating that Monaro has solved bitcoin's problems?
Whether the attacker can put the bits and pieces together and frequently make a good guess as to what is happening behind the curtain is unclear to me, and I doubt it is much clearer to the developers of Monero.
I don't see a survey and and analysis looking for the gaps between the mighty unbreakable fortress walls, and am disinclined to perform such an analysis.
If I was a Monero contributor and developer, I would have such a survey and analysis at my fingertips. What can an attacker do by analyzing the blockchain, and what patterns in the blockchain would he notice? In what ways do your activities when you use Monero cause the blockchain to differ from white noise to someone without your private keys?
You've provided no clear examples of data leakage, and assumed that there no people within the Monero community researching such topics. There are several in the #monero-research-lab doing exactly what you describe in various ways. Funding for this is somewhat low so its difficult to have the major press releases and landing pages describing it. There's been a few changes and possibly a few more based on analysis from a people that "hang around" that IRC room. There's a few videos on youtube for this topic as well.
Its worth mentioning that the original cryptonote authors were aware of the twist issues (or blindly/luckily followed DJB's advice) as the codebase has mul8 (the cofactor) in two key areas since the first commit. These authors came up with an entirely new ring-signature design - it was not as simple as "using academic literature"
It is really hard to do new stuff securely while using a group of composite order.
It did not seem to me that they were aware of how hard it is.
The problem was not that their algorithms were necessarily new. The problem was that these new algorithms were necessarily implemented on a group of composite order.
I take exception with the phrasing "really hard", but certainly there is more to think about with non-prime groups. However, framing this as a mishap resulting from ignorant people that should've simply used a prime order group is also incorrect. When the coin first launched, Ristretto didn't exist and Ed25519 was an interesting choice for a curve on a new coin. The original implementation also has no issues resulting from this choice, but unfortunately a modification to the original design did. And changing the curve of a production-used coin isn't trivial. Its easy for outside critics to boast about things "they would've done" when they aren't considering all of the details.
I doubt you can claim more competence than this person, if any. The problem is some operations do not require a sub-group check or cofactor multiply, and either mitigation strategy uses non-trivial number of CPU cycles.
We know he screwed up. Would I have screwed up? You don't know and I am not sure, but I know I would have sweated bullets, being aware of how difficult it is to avoid such screw ups, and having studied other people's efforts to work around the problem.
The problem is no worse than Bitcoin
The problem is vastly less worse than Bitcoin. Monaro is better in this regard than Bitcoin. Whether it is enough better to make a big important difference is disturbingly unclear.
- did you intend to promote Zcash here?
I have not attempted to scrutinize Zcash, which uses cryptography well beyond my limited competence.
But I am sufficiently competent to understand Wasabi, which is what I am indeed promoting.
There's some negatives to that project that can be found via websearch, but the z-address transactions are (assuming no bugs or math errors) not leaking the information you describe.
If zcash works as described, the blockchain should look like white noise to an attacker without the private keys.
Whether it does look like white noise is beyond my abilities to determine.
Wasabi is not an improvement over Monero, there is far more information leakage. I can't even think of a single privacy related benefit to Wasabi over Monero transaction constructions off-hand. Every transaction has a publicly visible amount, which aids in tracing "through" the mixing process - outputs are frequently broken into fixed sized-amounts, mixed, then re-assembled into nearly the same size as the original output.
This is the sudoko attack.
When Wasabi was issued, it was vulnerable to the sudoko attack, which could typically track about half the coins through a transaction, but now countermeasures have been applied against that attack.
The sudoko attack should have been foreseen, and it was not. But now there is an effort to check the mighty fortress walls to make sure there are no gaps between the mighty fortress walls that an attacker who declines to play by your security model can use.
The attack cannot be sufficiently addressed with public amounts in the transaction. The wallet user typically also has to be careful about their usage patterns. The "opt-in" privacy is a primary issue.
Every Monero transaction requires the ring-signature construction, so they do not "stand out" like Wasabi transactions.
Yes, this is an important vulnerability of Wasabi transactions. They can detect that you have laundered money, though once the money has been laundered, they can track it no further. But Monaro itself stands out, while Wasabi transactions are just more bitcoin transactions.
If everyone used Monaro, Monaro would not stand out, and if everyone used Wasabi, Wasabi transactions would not stand out.
This is ignoring my other point - if everyone used Wasabi the efficiency gains touted by Bitcoin are reduced.
To use crypto currency, you are apt to wind up laundering bitcoin by converting it to Monaro, which transaction tends to be highly traceable, and then converting it back to Bitcoin, which transaction tends to be highly traceable, and then actually paying someone with bitcoin, whereas if you are using wasabi, you are paying with the equivalent of crumpled used notes that the mafia collected from the laundry.
I'm not sure what you mean here. Lee