> Server Logs
> Like all web services, our servers may log:
> IP addresses of visitors
> Request timestamps
> User agent strings
> These logs are used for security and debugging purposes and are not linked to your account.
Shouldn't you have spent some time to think through basic things like this before trying to write an opinion piece on anonymity? Certainly it shows a lack of depth of understanding.
In most countries the law doesn't say you have to log everything about your users, but it does say that if you log it and the police ask for it then you have to give the data to them.
Speaking of mullvad. I recently learned about mullvad browser, which is basically tor browser minus connecting via the your network. This is interesting because the tor project has put the most effort into fingerprinting resistance. If you care about privacy and you have a customized browser, you're likely uniquely finger printable [1]. If you don't want to connect via tor, there's no excuse not to use the mullvad browser. (Doesn't require you to use mullvad VPN; comes with the mullvad plugin, disabled by default, to optionally use mullvad encrypted DNS. Last point, I wrote to the tor project and asked "is it possible to use tor browser minus tor network", and they responded "that's the mullvad browser", so this isn't just my recommendation)
This seems like the wrong end of the system to fix the problem. Someone saying "we don't log your IP address" isn't something you can easily verify, so the promise doesn't mean much because if they suck they're just going to lie about it.
What you need instead is to make it easy and common for people to use browsers that resist fingerprinting, VPNs/Tor, custom email addresses per-account, etc. Because then instead of claiming to not log your information, they simply do not have it.
The biggest thing we need is a better way to pay someone over the internet without them knowing who you are.
What scares me is that the more privacy oriented you are, the easier you are to fingerprint. At what point does privacy mean blending in with the crowd and not sticking out?
You're thinking about browser fingerprinting (client-side), but my post is about service-level anonymity (server-side).
Browser fingerprinting: "Your unique combination of extensions/settings makes you identifiable among other users."
Service anonymity: "There are no other users to compare you against because we don't collect identifying data."
When you sign up with just a random 32-char string, there's nothing to fingerprint. No email to correlate. No IP logs to analyze. No usage patterns to build a profile from.
Fingerprinting matters when services collect behavioral data. We architected our way out of having that data to begin with.
>When you sign up with just a random 32-char string...
There's STILL a browser fingerprint, IP logs to analyze, usage patterns to build a profile from. You may claim you don't collect it, but users need to take your word for it. This is just pseudonymity, which (as many BTC users found out) only gets you halfway there. Real anonymity is way harder, often impossible.
Don't get me wrong, it's good to see organisations that care about privacy and in fact this blog post encouraged me to consider your services in the future. We have some use cases for that at work.
Though by using cloudflare you're NOT putting your money where your mouth is.
I was going to say making the platform open source might solve this problem, but then users would have to trust that we are actually running the open source version and not some fork with logging and tracking. This would be an interesting problem / paradox to try to crack.
But you are 100% right, I will look into alternatives for Cloudflare, which we are using because it seems like the cloud hosting industry LOVES to DDoS new players.
Blending in with the crowd doesn't work. If you use Chrome on Windows you're part of a very large group and "don't stick out". But it's also very easy to fingerprint so you're also part of the "theturtletalks" group with the size of one.
Any business that isn’t willing to be as anonymous as Mullvad, I assume has a compromised business model that I don’t really like. Assuming there aren’t obvious reasons for needing the data, like tax filing, or various regulatory requirements.
I don’t understand why any company would want the liability of holding on to any personal data if it wasn’t vital to the operations of the business, considering all the data breaches we’ve seen over the past decade or so. It also means they can avoid all the lawyers writing complicated and confusing privacy policies, or cookie approval pop-ups.
> I don’t understand why any company would want the liability of holding on to any personal data if it wasn’t vital to the operations of the business, considering all the data breaches we’ve seen over the past decade or so.
They're OK with the liability exactly because of this very sentence. As you said, there's so many data breaches... so where are the company-ending fines and managers/execs going to prison?
Up to EUR 10,000,000 or up to 2% of the total worldwide annual turnover of the preceding financial year, whichever is higher; applies to infringements such as controller and processor obligations, security of processing, record-keeping, and breach notification duties.
Up to EUR 20,000,000 or up to 4% of the total worldwide annual turnover of the preceding financial year, whichever is higher; applies to infringements of basic principles for processing, data subjects’ rights, and unlawful transfers of personal data to third countries or international organisations.
Sure, in principle. Have you heard of any company that suffered any significant hardship (say, stock price plummeting, personnel reductions, bankruptcy) because of one of these fines?
Here in Japan the government cracks down on it hard. There are fines for every n users exposed and in extreme cases a company can be forced to stop trading for a period of days or weeks. Companies are so scared of this happening to them that a significant portion of orientation for new employees is spent on it. I don't have stats on how effective it is, but I do know that the public is less willing to accept it as they tend to elsewhere.
Is this true? KADOKAWA had a massive hack last year that leaked a large amount of sensitive user data and as far as I know has faced no legal repercussions. Obviously they took a decent financial and reputational hit, but that was just an effect of the hack itself, not any government intervention.
Infra engineer here. The obvious reasons for needing the data is debugging. I collect logs, metrics, traces, and errors from everywhere, including clients. All of these come with identifying information including the associated user. From the perspective of this thread this is a huge amount of data although it's pretty modest compared to the wider industry.
This data is the tool we have to identify and fix bugs. It is considered a failing on our end if a user has to report an issue to us. Mullvad is in an ideal situation to not need this data because their customers are technical, identical, and stateless.
It's not my department but I think we would get laughed out of the room if we told our users that we couldn't do password resets or support SSO let alone the whole forgetting your 'credential' means losing all your data thing.
Sadly, everybody using a browser from a massive ad company and an idp (not to mention a company with an interest in crawling the entire web for AI at the same time site owners are dealing with better scrapers) means the entire web will be login-only over time.
I don't see how those points bolster your conclusion. These pressures predate AI by over a decade and haven't forced a significant tidal change in the way the internet is used.
According to article, the whole authorization system is flawed. But we haven’t invent a new one and the one we’ve got never meant to be private, it is just a way to separate users from each other. We need something unique, a "primary key" for our DB, and that’s email or phone or username that has to be stored somewhere. A server, someone else’s computer, call it what you want. It has good privacy between users, but the admin can see everything, because otherwise management of the service would be impossible.
There is no anonymity, there is always someone you have to trust in the chain of WAN networking (DNS,ISP,VPN). If you want anonymity and privacy, you selfhost (examining the code is also a prerequisite). There is no other way to do it.
> but the admin can see everything, because otherwise management of the service would be impossible.
It depends on what service you’re offering. There are many cases where you can have end-to-end encryption so that you can know who your users are, host their data but cannot do anything with it.
So my understanding is, what Mullvad is to VPNs, and what Tarsnap is to S3 (kinda), Servury is to entire VMs. It's a prepaid model, you get an account identifier, and that's basically it.
This is very cool. I have wondered for a very long time why such a site does not exist. What pops to mind is that you could get better unit economics reselling really small VMs to the privacy obsessed. I know some netizens who would pay a dollar a month for, say, a tiny NetBSD VM and 64 MB of RAM to serve their tiny static demoscene website of yore. There are some real wizards of there.
Not sure if that's in your roadmap but definitely something to consider in this space.
There is no such thing as anonymity. With the number of bits required to ID a person and the fact that you are leaking such bits all the time you can simply forget about anonymity.
Many people online seem to think that they are anonymous and so were emboldened to do stuff that they might not have done if they had realized this. They continued to feel extremely good at this right up until the knock on the door.
Most UK and Australian writers would spell it "realised" so there's a bit right there.
Even if you include no personal information, there is information in writing style.
Stylometry is the study of this. Yes, there's also adversarial stylometry - distorting your writing style to fool an analysis. It's probably effective now, but that could change overnight and every archived post that every OSINT organisation has collected is deanomynised.
Yeah you can say "I change my style". But there's some bits that don't have false positives. If I EVER say "praise the omminsiah" I'm definetly au fait in 40k memes. If I ever say au fait I'm a person who has at least a rough idea of what it means. There's no false positive here, so if you can just find about 29 undeniable uncorrelated bits that are known to not have false positives ... a more advanced analysis could exploit this in a more continuous way (e.g. the likelihood of it being a false positive). I should shut up now.
Well there's anonymity from authorities, and there's anonymity from garden variety lunatics.
There exists a grey area between not getting away with nefarious activities, and not having your life ruined by a lynch mob because you didn't approve their preferred CoC on a hobby project or some other perceived injustice.
And, also not very funny, those corps never tell in advance which data they "require". They grab my mail on "the first page" of the registration form. Then, on "the second page", they ask for my phone and my address. Should I decide to agree to this, they will finally tell me on "the third page", that they only support credit card, no PayPal, no direct payment via Bank ...
One difference with Mullvad is VPN traffic is ephemeral. Here, a VPS has a persistent disk attached, that could contain identifying information (if it is necessary to do useful work).
What I was wondering after reading the article: How does Mulvad actually decouple banking data from the account ID? Or is it as simple as verify transaction once but never log?
So there's no subscription thing going on, you just manually pay invoices?
I once spent an entire year issuing chargebacks on AWS charges coming from god knows what AWS account. Most likely some client project I forgot about and didn't have the login to anymore, who knows. Makes me think about that - for a service where you can't login if you lose the credentials, how do you cancel a subscription? In my case I had to eventually just cancel the credit card and get a new number.
No subscription. It’s pay as you go. You top up $X and you get X months. That’s it. If your month expires, it expires. Just top off and you’re good to go.
I would much rather have privacy with e2e encryption than have anonymity. The way that works is a direct connection between two parties without use of a central server, like webRTC.
tl;dr
“Privacy” = the data is private i.e. only on your devices. Or if the raw data is public but encrypted and the key is private, I think that qualifies.
“Anonymity” = the data is public but not linked to its owner’s identity.
If you’re sharing your data with a website (e.g. storing it unencrypted), but they promise not to leak it, the data is only “private” between you and them…which doesn’t mean much, because they may not (and sometimes cannot) keep that promise. But if the website doesn’t attribute the data except to a randomly-generated identifier (or e.g. RSA public key), the data is anonymous. That’s the article.
Although a server does provide real privacy if it stores user data encrypted and doesn’t store the key, and you can verify this if you have the client’s unobfuscated source.
Also note that anonymity is less secure than privacy because the information provides clues to the owner. e.g. if it’s a detailed report on a niche topic with a specific bias and one person is known to be super interested in that topic with that bias, or if it contains parts of the owner’s PII. But it’s much better than nothing.
Europe is currently being tormented by this exact contradiction: on one hand, it has the GDPR—the world's strictest privacy law, supposedly protecting personal data; on the other, a flood of new regulations under the banners of "child safety," "counter-terrorism," and "anti-money laundering" are systematically strangling real anonymity.
The battle on privacy/anonymity/whatever is lost. Get over it. What we need is a new social paradigm where everyone is happy despite the lack of privacy.
Please provide your full legal name (include any other names you go by), occupation and place of employment, phone number[s], email address[es], usernames on other social media accounts, eye color, height, weight, list of any health conditions. That's just to start, then we can start going over more info.
"privacy" or not sharing your space with a creepy room mate, and reading the internet without adds ar3 parallel
running three flavors of the same off brand browser, each optimised for different segments of online content is what seems to be the minimum.
they are so desperate to sell me something,
(a truck) that it's wild, as it is one of the few monitisable things I consistently look for (parts, service procedures), the ,
pause, when I do certain searches gives me time to predict that yes, the machinery is grinding hard, and will ,shortly, triumphantly, produce, a ,truck.
How tf are you supposed to provide working authentication without storing the email somewhere? Should i just disable password resets and tell the users to fuck off if they forget theirs? Cant even use passkeys as they make users identifiable too.
Users need to have hard memorization or record of a paraphrase, same as a crypto wallet. Or just use web3 for auth, that can work well if users have decent opsec.
I agree, privacy still means a lot. It's a term that's been co-opted by the large tech companies which operate with impunity. It will has meaning that cannot change.
The post also misunderstands privacy
> Privacy is when they promise to protect your data.
Privacy is about you controlling your data. Promises are simply social contracts.
> Server Logs > Like all web services, our servers may log: > IP addresses of visitors > Request timestamps > User agent strings > These logs are used for security and debugging purposes and are not linked to your account.
That's already a huge breach in comparison to mullvad privacy page. (https://mullvad.net/en/help/no-logging-data-policy)
Web server logs were not tied to user credentials in any way, they were used for debugging purposes and could not have been used to identify users.
Front page says "zero logs"
Some logs, including specifically datapoints you have promised not to log, but you mean well (?) is pretty different from zero logs
I'm not here to debate, the reason I posted here is to hear what people thought and see how I could improve my platform based on the criticism.
(Asking because I really don't know)
And the "3 data points, that's it" of the blog post
Web server logs were not tied to user credentials in any way.
Also:
> // What we DON'T collect:
> - IP addresses (not logged, not stored, not tracked)
> - Usage patterns (no analytics, no telemetry, nothing)
> - Device fingerprints (your browser, your business)
so, I've read one blog from this company, and already they're lying or incompetent
[1] https://coveryourtracks.eff.org
What you need instead is to make it easy and common for people to use browsers that resist fingerprinting, VPNs/Tor, custom email addresses per-account, etc. Because then instead of claiming to not log your information, they simply do not have it.
The biggest thing we need is a better way to pay someone over the internet without them knowing who you are.
Cryptocurrency?
Browser fingerprinting: "Your unique combination of extensions/settings makes you identifiable among other users."
Service anonymity: "There are no other users to compare you against because we don't collect identifying data."
When you sign up with just a random 32-char string, there's nothing to fingerprint. No email to correlate. No IP logs to analyze. No usage patterns to build a profile from.
Fingerprinting matters when services collect behavioral data. We architected our way out of having that data to begin with.
There's STILL a browser fingerprint, IP logs to analyze, usage patterns to build a profile from. You may claim you don't collect it, but users need to take your word for it. This is just pseudonymity, which (as many BTC users found out) only gets you halfway there. Real anonymity is way harder, often impossible.
Don't get me wrong, it's good to see organisations that care about privacy and in fact this blog post encouraged me to consider your services in the future. We have some use cases for that at work.
Though by using cloudflare you're NOT putting your money where your mouth is.
But you are 100% right, I will look into alternatives for Cloudflare, which we are using because it seems like the cloud hosting industry LOVES to DDoS new players.
https://news.ycombinator.com/item?id=46334951
I don’t understand why any company would want the liability of holding on to any personal data if it wasn’t vital to the operations of the business, considering all the data breaches we’ve seen over the past decade or so. It also means they can avoid all the lawyers writing complicated and confusing privacy policies, or cookie approval pop-ups.
They're OK with the liability exactly because of this very sentence. As you said, there's so many data breaches... so where are the company-ending fines and managers/execs going to prison?
Up to EUR 10,000,000 or up to 2% of the total worldwide annual turnover of the preceding financial year, whichever is higher; applies to infringements such as controller and processor obligations, security of processing, record-keeping, and breach notification duties.
Up to EUR 20,000,000 or up to 4% of the total worldwide annual turnover of the preceding financial year, whichever is higher; applies to infringements of basic principles for processing, data subjects’ rights, and unlawful transfers of personal data to third countries or international organisations.
https://ico.org.uk/action-weve-taken/enforcement/
Some went to prison, some were fined £14M and it's a mixture of small fry and big fry.
It’s not very hard to handle customer data in a legally compliant way, that’s why you don’t see companies deciding against retaining data.
You can do everything right and still have a data breach, and in that case nobody is fining you.
This data is the tool we have to identify and fix bugs. It is considered a failing on our end if a user has to report an issue to us. Mullvad is in an ideal situation to not need this data because their customers are technical, identical, and stateless.
It's not my department but I think we would get laughed out of the room if we told our users that we couldn't do password resets or support SSO let alone the whole forgetting your 'credential' means losing all your data thing.
There is no anonymity, there is always someone you have to trust in the chain of WAN networking (DNS,ISP,VPN). If you want anonymity and privacy, you selfhost (examining the code is also a prerequisite). There is no other way to do it.
It depends on what service you’re offering. There are many cases where you can have end-to-end encryption so that you can know who your users are, host their data but cannot do anything with it.
This is very cool. I have wondered for a very long time why such a site does not exist. What pops to mind is that you could get better unit economics reselling really small VMs to the privacy obsessed. I know some netizens who would pay a dollar a month for, say, a tiny NetBSD VM and 64 MB of RAM to serve their tiny static demoscene website of yore. There are some real wizards of there.
Not sure if that's in your roadmap but definitely something to consider in this space.
Many people online seem to think that they are anonymous and so were emboldened to do stuff that they might not have done if they had realized this. They continued to feel extremely good at this right up until the knock on the door.
Most UK and Australian writers would spell it "realised" so there's a bit right there.
Even if you include no personal information, there is information in writing style.
Stylometry is the study of this. Yes, there's also adversarial stylometry - distorting your writing style to fool an analysis. It's probably effective now, but that could change overnight and every archived post that every OSINT organisation has collected is deanomynised.
Yeah you can say "I change my style". But there's some bits that don't have false positives. If I EVER say "praise the omminsiah" I'm definetly au fait in 40k memes. If I ever say au fait I'm a person who has at least a rough idea of what it means. There's no false positive here, so if you can just find about 29 undeniable uncorrelated bits that are known to not have false positives ... a more advanced analysis could exploit this in a more continuous way (e.g. the likelihood of it being a false positive). I should shut up now.
There exists a grey area between not getting away with nefarious activities, and not having your life ruined by a lynch mob because you didn't approve their preferred CoC on a hobby project or some other perceived injustice.
be confident that the service is not keeping logs? JÁ!
I once spent an entire year issuing chargebacks on AWS charges coming from god knows what AWS account. Most likely some client project I forgot about and didn't have the login to anymore, who knows. Makes me think about that - for a service where you can't login if you lose the credentials, how do you cancel a subscription? In my case I had to eventually just cancel the credit card and get a new number.
But in order to read the article you need to enable JS. What a joke.
“Anonymity” = the data is public but not linked to its owner’s identity.
If you’re sharing your data with a website (e.g. storing it unencrypted), but they promise not to leak it, the data is only “private” between you and them…which doesn’t mean much, because they may not (and sometimes cannot) keep that promise. But if the website doesn’t attribute the data except to a randomly-generated identifier (or e.g. RSA public key), the data is anonymous. That’s the article.
Although a server does provide real privacy if it stores user data encrypted and doesn’t store the key, and you can verify this if you have the client’s unobfuscated source.
Also note that anonymity is less secure than privacy because the information provides clues to the owner. e.g. if it’s a detailed report on a niche topic with a specific bias and one person is known to be super interested in that topic with that bias, or if it contains parts of the owner’s PII. But it’s much better than nothing.
running three flavors of the same off brand browser, each optimised for different segments of online content is what seems to be the minimum.
they are so desperate to sell me something, (a truck) that it's wild, as it is one of the few monitisable things I consistently look for (parts, service procedures), the , pause, when I do certain searches gives me time to predict that yes, the machinery is grinding hard, and will ,shortly, triumphantly, produce, a ,truck.
The post also misunderstands privacy
> Privacy is when they promise to protect your data.
Privacy is about you controlling your data. Promises are simply social contracts.