Blockchain Analytics is More of an Art Than Science

[LeGaulois]

Blockchain Analytics is More of an Art Than Science

I am not saying their software is so wrong, but knowing there are false positives, it can be very dangerous.
Even if the false positive rate is only about lets say 0.1%, considering the millions address/transactions being tracked, it can make a lot of dammages.
In example: ""led to unjustified account restrictions and – even worse – land unsuspecting individuals on the radar of law enforcement agencies without probable cause.""

- Hello Chainalysis,
Do you know if your software is accurate?

- I have no scientific evidence for the accuracy.

- Really?

- Really, but clients say so

- Clients, you mean governments lying depending on their agenda?

- Yes

Chainalysis' head of investigations doesn't seem to have a great understanding of whether her company's flagship software even works.

Ekeland said Chainalysis’ Reactor is “a black box algorithm” that “relies on junk science.”

Coindesk, the woke media, didn't like the article talking bad about Chainalysis and decided to remove it
Chainalysis>>>Coindesk>>>DigitalCurrencyGroup
so here it is
https://bitcoinmagazine.com/technical/chainalysis-investigations-lead-is-unaware-of-scientific-evidence
You may be interested in Blockchain Analytics is More of an Art Than Science

[Accardo]

Such softwares used for tracing addresses only instigate stress on a victim that got scammed or lost their coin. A scam victim, who wants to track the address where he/she sent a transaction, needs to bring evidences; chats, photos and money to the agency that claim to trace blockchain transactions. Most victims end up losing more money added to the lost money. Its existence also gave birth to fraudulent agencies claiming to retrieve lost coins or tracking a scammer; with the slogan, a person that fell for scam can easily fall again. The results always appear differently based on the software used to track an address. If the software can't retrieve the lost coin, why depend on unrealistic results, in the name of tracing a person who may not be behind the crime or scam?

[LeGaulois]

You're wrong.

Such companies track more than what you may think and not only to help an individual ripped off.
It's called data mining. They collect millions stuff.
Big companies like Bitstamp , etoro, and other CEXs are using it. So what do you think they're doing. CEXs provide them a lot of information about their customers trading on their platform.

It's not so different than a data broker.

data mining is such a powerfull industry, people do not realize what can be done with it... against them...

[Adbitco]

Such softwares used for tracing addresses only instigate stress on a victim that got scammed or lost their coin. A scam victim, who wants to track the address where he/she sent a transaction, needs to bring evidences; chats, photos and money to the agency that claim to trace blockchain transactions. Most victims end up losing more money added to the lost money. Its existence also gave birth to fraudulent agencies claiming to retrieve lost coins or tracking a scammer; with the slogan, a person that fell for scam can easily fall again. The results always appear differently based on the software used to track an address. If the software can't retrieve the lost coin, why depend on unrealistic results, in the name of tracing a person who may not be behind the crime or scam?

Why then the software is implemented?
There's no usefulness of that software since it can't retrieve back also lost coin, as I believe that there are two things possible..
A lost coin may either go an active wallet or a dead wallet (inactive) whereby any transaction that occurs between these wallet address can be seen but can't be touched since those addresses doesn't bear bearer names on them and it could be only possible if they sent to exchange address. For Instance, if lost is mistakenly sent to exchange address at this point it's possible to contact the exchange about the funds that was mistakenly sent at this point their software is valuable but doesn't guaranteed a total dependency.

[hatshepsut93]

Has there been any cases of someone getting convicted with Chainanalysis being the only evidence? Because I think Chainanalysis mostly only helps law enforcement to reinforce the case and lead to stronger evidence, like the illegal goods, laundered money etc.

This is a big deal considering Chainalysis' surveillance tools are used widely across the industry for compliance, and have at times led to unjustified account restrictions and – even worse – land unsuspecting individuals on the radar of law enforcement agencies without probable cause.

Private entities have right to reject their customers for whatever reason they want, including Chainanalysis saying that the coins are dirty. That's how capitalism works. You can try to change their mind and tell them how its unscientific, but in the end it's their decision, and you have a freedom to be or not to be their customer.

[franky1]

data is only as good as the source

chain-analysis gathers data. but it can only go on the data it gathers. nothing is suggested that the data it gets is correct.
for instance someone could easily scam an exchange using false ID. meaning the exchange registers a withdrawal address as belonging to a false ID victim. so while the scammer is shuffling funds. the ID theft victim becomes the suspect when data is passed from an exchange to chain-analysis.

it is then for the victim/suspect to prove innocence in court. and not chain-analysis to appear in court to prove the victims innocence/exonerate the victim.

Private entities have right to reject their customers for whatever reason they want, including Chainanalysis saying that the coins are dirty. That's how capitalism works. You can try to change their mind and tell them how its unscientific, but in the end it's their decision, and you have a freedom to be or not to be their customer.

very true alot of people think using a service is a "right" when its not. businesses can ban users for any reason. but that business has to allow the customer to leave with their property(for the unwarranted reason). only a court order would allow a business to withhold a customers property. so only if a customer is deemed as doing something suspicious, where the service reports suspicion to the authorities and the authorities then investigate and deem it enough evidence to get a court order to then allow a service to seize funds.
without court order a business can just ban/suspend the customers use of service and offer a withdrawal/refund procedure to exit the service and not return, for even lame reasons like deeming funds are dirty but not evidentially dirty enough to suspect a crime by the customer themselves

[Don Pedro Dinero]

Blockchain Analytics is More of an Art Than Science

I am not saying their software is so wrong, but knowing there are false positives, it can be very dangerous.
Even if the false positive rate is only about lets say 0.1%, considering the millions address/transactions being tracked, it can make a lot of dammages.

This is interesting as I was not aware of this aspect. I thought that blockchain analytics gave results similar to mathematical certainty, not probabilistic certainty, even if it is 99.9% certainty, which leaves that margin of error down to the 0.1% point.

It seems that the error could be even higher, according to the first article.

Chainalysis' head of investigations doesn't seem to have a great understanding of whether her company's flagship software even works.

Elizabeth Bisbee, head of investigations at Chainalysis Government Solutions, testified she was “unaware” of scientific evidence for the accuracy of Chainalysis’ Reactor software used by law enforcement, an unreleased transcript of a June 23 hearing shows.

Has there been any cases of someone getting convicted with Chainanalysis being the only evidence? Because I think Chainanalysis mostly only helps law enforcement to reinforce the case and lead to stronger evidence, like the illegal goods, laundered money etc.

This is how I think it should be but we are not talking about condemning alone, we are talking about targeting and passing restrictions, like blacklisting a number of addresses accordingly.

[nutildah]

From the article:

For Chainalysis' part, the company noted in a July 18 court declaration there is no doubt about the Reactor software's findings. Although not "peer reviewed" in an academic sense, the company's clustering heuristic, the algorithm used to find relationships between blockchain addresses, comes to "deterministic" conclusions that can be independently verified and reproduced.

The difference between what Chainalysis does and what armchair blockchain analysts do is that the results they submit to clients are designed to hold up in court. They are constructed to withstand the scrutiny of any cross-examination.

The US government wouldn't go forward presenting a case if it relied upon analysis that had a chance of containing false positives, as any competent defense attorney could use this "shadow of a doubt" to get the case dismissed or ruled in their favor (meaning it would have been a waste of resources, which believe it or not, is still frowned upon by many departments of the federal government).

There is indeed an "art" to it but it lies in knowing what data or conclusions are relevant to the particular case in hand.

[Poker Player]

There is indeed an "art" to it but it lies in knowing what data or conclusions are relevant to the particular case in hand.

I take this as a very qualified opinion on the subject.

[NotATether]

I am not saying their software is so wrong, but knowing there are false positives, it can be very dangerous.
Even if the false positive rate is only about lets say 0.1%, considering the millions address/transactions being tracked, it can make a lot of dammages.

It's more similar to AI actually, which also has false positives and negatives because it's a large data-driven guessing game.

In fact, I wouldn't be surprised if machine learning is what is being used to power these blockchain analytics services.

This is interesting as I was not aware of this aspect. I thought that blockchain analytics gave results similar to mathematical certainty, not probabilistic certainty, even if it is 99.9% certainty, which leaves that margin of error down to the 0.1% point.

Nope not at all. The only way you can be certainly sure is if the address is posted on somebody's social media page as an "ownership" or if its attached to KYC documents.

[o_e_l_e_o]

As I've said before, blockchain analysis is based on guesswork.

Bitcoin, by design, is fungible. As soon as a transaction has more than one output, it is impossible to say which bitcoin ended up where. It cannot be done. Everyone who claims to be able to do it is guessing, lying, or both. All blockchain analysis companies, all centralized exchanges, and now Wasabi too (which is particularly hilarious considering they base their whole existence on coinjoins). They have made up a system based on guesswork, and have successfully marketed it for their own profit to large parts of this space as some infallible law. It is not, and the only way to get rid of it is for the community to agree to shun companies and entities which support and enforce this made up nonsense.

As soon as a transaction has been made, it is impossible to say that those coins haven't changed hands. As soon as a transaction has more than out output, it is impossible to say which bitcoin ended up where (and indeed, "which" bitcoin doesn't even exist at a protocol level). It is trivially easily to fool many of the heuristics blockchain analysis uses, such as script type matching to identify the change output, or inputs being spent together to identify co-ownership. And not just to fool them as in "they can't draw any conclusions", but to fool them as in "they actively draw the incorrect conclusion". And of course one incorrect conclusion leads them to build more and more incorrect conclusions on top, building an entire chain of nonsense which they then pass off as irrefutable fact.

I've said for a long time that blockchain analysis is provable nonsense with no scientific basis. It seems even the directors of blockchain analysis companies agree with that. But of course they will continue to peddle their nonsense to centralized exchanges and governments alike because it pays handsomely to do so.


I did a small experiment some time ago regarding blockchain analysis: https://bitcointalk.org/index.php?topic=5395035.msg59905002#msg59905002

One particular piece of blockchain analysis software put a significant amount of coins in the wallet of various centralized exchanges in one of the categories of scams, hacks, or blacklists. Obviously the blockchain analysis software being used by these exchanges did not classify these coins in this manner, otherwise they wouldn't have accepted those coins. The fact that two different pieces of software can come to completely different conclusions about the exact same coins should be more than enough to tell you that blockchain analysis is made up trash.

One of the core principles of any piece of science is that its results are repeatable and independently verifiable. If I come up with a process to say, isolate gold from an alloy, then I publish my methods and other people perform the same steps, end up with the same results, and verify my process works. If I come up with a process to say some coins are tainted, and other people do the same thing and end up with completely different results, then my process is bullshit.

[ABCbits]

Instead, Chainalysis reportedly judges its software’s accuracy using customer feedback, she said.

This is ridiculous. Customer usually expect or assume analysis would be accurate enough. In addition, on some cases customer have no way to verify whether the analysis result is accurate or not. I hope more people switch to wallet with better privacy feature to make their analysis become even less useful than now.

Such softwares used for tracing addresses only instigate stress on a victim that got scammed or lost their coin.

It's not the only reason to use blockchain analytic service though. For example, many such service claim their service is useful to help comply with law.

[franky1]

the data which different transaction analysis sites produce is solely based on the credibility of the data provided to them..
for instance chain-analysis is sister companies of several dozen exchanges and services. so they share data about their customers.
this means if they have a customer on exchange A with email X and exchange B with the same email X. they can then try to find coin spend linkages

some crap analysis sites just do taint analysis without CEX/service customer data. so the results are subjective to the data they are given

[DaveF]

I'm just about 100% sure the 3 letter agencies of the US government have a lot better accuracy in determining what coins belong to who and where they go.
Companies like Chainalysis are guessing at best. BUT and this is the big BUT if the companies that USE their service are happy and pay on time that is all that Chainalysis (and the others like it) care about.

It gives them the ability to tell regulators that they did their job the best they could in making sure they did not take any funds in from people or places they should not have.

That's it, over and out. Do you really think that ANY company wants to spend money on a product or service that they did not have to? It's just checking off the box that says we did what we had to do.

-Dave

[Lucius]

I'm just about 100% sure the 3 letter agencies of the US government have a lot better accuracy in determining what coins belong to who and where they go.

We can be sure of that, given that they have been working on it for at least 10 years, and this was confirmed by someone who had an insight into all that. If we look at the actual volume of their work, which included time, money and specific knowledge, then it is clear how much of a priority this issue was for them.

Companies like Chainalysis are guessing at best. BUT and this is the big BUT if the companies that USE their service are happy and pay on time that is all that Chainalysis (and the others like it) care about.
~snip~

Guessing pays very well today, and that's why many people started to engage in this business. It is obvious that all the companies that pay for their services need some kind of justification for the purpose of disclaimer in order to fight against money laundering and of course the state's enemy number 1 - terrorism.

This is the situation when someone says "based on the collected data, we did not think that we should take any actions to prevent...", and I guess that's exactly what such companies are for.

[hatshepsut93]

very true alot of people think using a service is a "right" when its not. businesses can ban users for any reason. but that business has to allow the customer to leave with their property(for the unwarranted reason).

When a user opens an account, they enter a user agreement. If the user agreement says that the service can freeze the account and seize funds, then there's no violation of law. If the rules don't mention such cases, then the users should be able to take their money. But most of the time the rules are very detailed and cover all potential cases.

People need to read what they sign.

[Faisal2202]

Such companies track more than what you may think and not only to help an individual ripped off.
It's called data mining. They collect millions stuff.
Big companies like Bitstamp , etoro, and other CEXs are using it. So what do you think they're doing. CEXs provide them a lot of information about their customers trading on their platform.

It's not so different than a data broker.

data mining is such a powerfull industry, people do not realize what can be done with it... against them...

You are totally right; I might not be new to the cryptosphere, but I was new to the use of data and how valuable it is when I started to read about IOT. Because in IOT we value data and read how important it is to make decisions, increase sales, and send proposals to someone who is in need. It's like when your bank balance is low, and then you get to receive some messages from many lenders and other platforms that provide some type of insurance or loan that could help you out when you are broke.

The thing is, once they know you have no money or have money, they will start to advertise their services and products accordingly just to lure you into becoming their customer.

PS: Is it possible that the investigator who knows nothing is just a human error from her side while the system does have a lot of ongoing in that context? Just asking.

[franky1]

very true alot of people think using a service is a "right" when its not. businesses can ban users for any reason. but that business has to allow the customer to leave with their property(for the unwarranted reason).

When a user opens an account, they enter a user agreement. If the user agreement says that the service can freeze the account and seize funds, then there's no violation of law. If the rules don't mention such cases, then the users should be able to take their money. But most of the time the rules are very detailed and cover all potential cases.

People need to read what they sign.

business policy does not trump law.
a unregulated shady boiler room scam service could put bad terms into agreements to try to make scammed customers not chase their funds.. but a regulated exchange has to follow the regulations and cannot seize funds arbitrarily without a court order.

sometimes although no one likes regulations. its worth reading them. even if its to learn the limits of their powers and thresholds of authority

as for the unregulated scammy shady exchanges.. well the risk is on the customer obviously which is why its best in regulated and unregulated to never leave funds in an exchange eitherway for many reasons, but more so when it comes to unregulated exchanges as you have less legal protections and defenses and less methods to pursue recompense

my common rule is this: if you cant physically find someone your handing funds to, to slap them with a wet fish should they wrong you. dont hand funds to them in the first place

[Latviand]

When a user opens an account, they enter a user agreement. If the user agreement says that the service can freeze the account and seize funds, then there's no violation of law. If the rules don't mention such cases, then the users should be able to take their money. But most of the time the rules are very detailed and cover all potential cases.

People need to read what they sign.

If people start reading the terms and conditions before signing up on something, the faster we can point out terms that we all agree isn't agreeable towards the consumers. As much as I agree that the services aren't at fault when they do something to the account because it's in their "terms and conditions", we have to also call out those services that put unfair terms and conditions just to use their service, both sides should experience some change.

[satscraper]

Blockchain Analytics is More of an Art Than Science

I am not saying their software is so wrong, but knowing there are false positives, it can be very dangerous.

That is why when building transaction one should operate on the base they know  all his addresses. There is a nice guide ( divided into four sections ^{[1] , [2],[3] and[4]}. )which sheds light  on heuristic used by those software. Thus, knowing that heuristic one can develop his own strategy against chainanalysis' clustering technique.
::::::::::::::::::::::::::   ::::::::::::::::::::::::::::  ::::::::::::::::::::::::::::::
[1]. Understanding Bitcoin Privacy with OXT — Part 1/4
[2]. Understanding Bitcoin Privacy with OXT — Part 2/4
[3]. Understanding Bitcoin Privacy with OXT — Part 3/4
[4]. Understanding Bitcoin Privacy with OXT — Part 4/4