Hacker News Reader

Comment by mattlondon | Hacker News Reader

In web3, your wallet ID (used for purchases and proposed for logins too) is an undeletable cookie that tracks you both online and off, and one that anyone can retrieve.

This article seems to think that having a super cookie like this will make advertising less bad. I think it will be the opposite - instead of only Google/Facebook doing it, you're now going to have every single shady operation under the sun now able to use/mine/combine your data however they please with zero controls (and unlike HTTP cookies they'll need exactly zero investment & zero infrastructure to do so, so there is no barrier to entry to start using this data)

Web3 is the advertising world's ultimate dream situation.

It allows significantly more invasive tracking by tying online and offline, cannot ever be deleted or changed, and the same cookie ID is shared across all aspects of your life (unless you use multiple wallets with one-per-site, but my understanding is that an anathema to web3 concepts about data ownership)

View on HN

>In web3, your wallet ID (used for purchases and proposed for logins too) is an undeletable cookie that tracks you both online and off, and one that anyone can retrieve.

Purchases sure on most chains (though not on e.g. Secret) but the chain isn't recording your logins via your address in most cases. You are just showing to the site that you have the key without making any transactions. So no, advertisers won't have much extra data on your site visiting habits that they don't have now.

The thing is, can't they "triangulate" who you are by your transactions?

In the same way tracking your location for sure can determine who you are.

https://www.nytimes.com/interactive/2018/12/10/business/loca...

Based on just a few transactions they can figure out that you're a middle aged guy, living in Nowhereville, Iowa, that's into samba, anime and drives a Toyota Corolla, etc. and then from that it's not hard to actually figure out who that person is.

I don't see a particular reason to conduct every transaction on-chain, or at least on a public L1. At any rate, I was mostly replying about your login data being public which isn't true.

Then how will we conduct our transactions?

Right now I use my country's currency, do you want me to go to a mix of currencies like I'm playing some sort of Android free to play game?

That's just adding complexity to my life for no obvious reason.

Crypto is sort of like the increasing plethora of streaming services, except crypto projects seem to reproduce at a much higher rate and further deprioritize UX. It's not every day, if ever, that I'm dealing with fiat currency conversion. In crypto, it's another one of those seemingly unnecessary hassles that is tolerated as a matter of course.

I use both PayPal and my bank for different transaction with no issues. I won't care to use some crypto app if that's popular, too. At any rate, I'm not personally advocating for crypto here, and the projects in crypto which I'm following are less about pure payments (which I don't mind doing via debit card or equivalent, whether built via banks or on top of crypto) and more about more complex smart contracts (e.g. the dex approach vs traditional exchanges and its assosiated different profit incentives, mechanisms and safety profile).

PayPal and your bank both transact in dollars. In the crypto space, presumably, you’d need to operate with different currencies at varying exchange rates. Or use some kind of “wrapped” currency at significant complexity and risk (see recent Ethereum-Solana bridge exploit and associated massive loss).

End users didn't suffer a loss after that exploit and I expect my risk from holding usdc and it crashing without Coinbase paying to fix things is already lower than the risk of my PayPal account being closed which I've seen happen to people.

I think no one really knows whether DeFi can be really safer than OldFi. Since it's old, we sort of know what to expect. In DeFi we obviously are still discovering.

I dislike OldFi and want to contribute to a new, healthier system to serve people in general. But, short term, risk of DeFi is obviously higher due to its imaturity.

OldFi has been maturing for 100's of years. Perhaps we could argue 1000's. It'll take at least 20-50 more years to reach a similar maturity in DeFi.

By maturity I mean: people understand the risks and know what to expect.

Registration via NFTs is already a thing. You join the yacht club by buying an ape. In this manner, everybody can know who is a member of the club.

If you are just doing login by registering your public key with a service.... then you are just doing signature-based-authentication that you could have done two decades ago.

Except that login is now tied to your transaction history.

I think it's worth puzzling out why nobody actually has been doing signature based authentication for the last two decades though. I'm honestly not sure whether that's a flattering exercise for web3 or the opposite.

I don't think it is puzzling.

1. Users understand passwords. While password stuffing is a concern, almost nobody actually makes product decisions based on rational security analysis.

2. Only a small number of users are concerned about centralized login via OpenID Connect. The privacy and account access weaknesses of logging in with Google or Facebook aren't concerns for the huge majority of users.

3. Creating and maintaining a key pair is less ergonomic than creating and maintaining a password.

#3 is now different for people who are regularly using blockchain systems so it doesn't surprise me that uptake is larger than now than in the past. More people have a Coinbase account than use PGP. But I am not convinced that people actually care enough to swap if they aren't already ideologically aligned with blockchain technology.

This is all reasonable, but you're a lot more certain it's the full answer than I would be.

For #1: New people come to technology all the time. They don't understand passwords yet, so if something is better than passwords it can make inroads.

For #3: It's not clear to me why this is the case, besides just that it has always been the case. It's really un-ergonomic to create and maintain passwords. It just seems the alternatives have been even more so.

(Interestingly, I don't think Coinbase actually gives you a keypair you can use to login to stuff...)

I think the underlying technologies behind web3, i.e blockchain cannot really evolve beyond ponzinomic pump and dump schemes without some kind of legitimate commercial purpose, maybe that's some kind of incentivised advertising.

People always say the likes of Google and FB treat us as a product, so perhaps we can at least get some kind of magic bean tokens as a reward :D

It's been tried and tried and tried. Generally when they want to increase profits the same outcome - they either flame out of VC money or start selling your data.

~1998 Beenz.com launched

https://en.wikipedia.org/wiki/Beenz.com

ahh that brings back memories. i had an account.

The real problem in Web3 (as EVM does it) is that almost all wallets lets you export your private keys. This leads to a lot of scams that trap people. Got rugpulled? Joined Metamask chat on Telegram? Hi, I’m an admin. Just paste your 12 word phrase into this official-looking site and we’ll help you. There’s even a site that will generate vanity addresses for you! (Don’t use it.)

Instead, all this could have been avoided if they took the Keybase approach: keys never leave the device. Just have a keychain of devices which you authorize or repudiate, and use THAT to retrieve your actual accounts from the network and sign things.

This would work best with BLS signatures without a trusted dealer. I am not sure it can work with ECDSA or Shnorr signatures if you want to collapse all multisigs down to one signature.

Identity and Authentication are related but separate topics. Identification is uniquely associate a user in a system or application while Authentication is proving that this identity is yours.

This can easily be solved with multi Sig wallets

Yes, those are better at least. But only if “this” is your individual use case, not if “this” is the general state of crypto custody in EVM ecosystem.

People think they need fancy airgapped computers and hardware wallets when all they need is to copy their 12 word phrase three times on paper, cut into 9 pieces and bury in 9 different places.

Better yet —- wallets normally shouldn’t allow people to export keys.

>Instead, all this could have been avoided if they took the Keybase approach: keys never leave the device. Just have a keychain of devices which you authorize or repudiate, and use THAT to retrieve your actual accounts from the network and sign things.

This is what modern smart contract wallets like Argent do. There is no 24 word phrase or long private key to write down, remember or lose. Instead you set up guardians which is a group of people, organizations, and devices you create that allow you to recover your account if you drop your phone in the ocean so there's no practical way you can lose your money or identity. An example guardian setup might be your mom, best friend, lawyer, your hardware wallet and a distributed paper wallet. To recover your wallet you would need 3 out of the 5 to agree. The number of guardians and other details can be as secure and exotic as you desire.

The account is also protected by configurable safeguards that mimic your existing bank account. Sending over $500/day? Guardians must approve. Your tokens or NFTs are being sent to a non-white listed address? Guardians must approve.

The crypto ecosystem is still in the early phases of being built but these problems and ideas have already been discussed and debated for years and the foundational infrastructure that people will end up using in the real world is being built and deployed today after years of hard work. Using crypto today is a lot like using the internet in the early nineties: it was hard, easy to break, and you generally had to be good at cutting your teeth on new and quickly evolving tech. Nowadays you mostly just forget you're using it.

> you set up guardians which is a group of people, organizations, and devices you create that allow you to recover your account

And you call this an improvement over the current state of affairs?

This provide you the choice of who (if anyone) you want to trust.

You can choose to only use yourself as a guardian, a group of family members/friends, a set of financial institutions, or a mix of all of those.

You can set it up so that it’s very hard for individuals or governments to take control of your assets.

Yes? I think so?

The problem is that this model is basically impossible with the current level of fees on ethereum. It isn't clear to me what the solution to that is supposed to be.

The solution should be to not use monolithic blockchains, at least not for key management :)

Look at Sidetree protocol for instance

If I understand correctly, you're proposing that we trust a set of individuals and hope that they'll cooperate when asked to, instead of relying on a single party which can be made to cooperate whether they like or not by the judicial system. That doesn't seem like an improvement to me.

> we trust a set of individuals and hope that they'll cooperate when asked to

Yes, in most cultures these individuals are called “family” or “friends”. It’s a rare concept in the crypto world though, yes.

Now imagine if some app replacing 911 broadcast an emergency message to your friends or neighbors, with your location.

Who would you trust more to come to your aid when there is a situation?

How about in a rural area or one where police are overworked?

How about if each neighborhood could hire its own police agencies and police agencies would compete on how well they keep people safe and maintain order without excess brutality? If they got to know the people they are policing, and worked with the clergy and others to help them holistically?

That is how libertarians think.

I'm really not sure what this comment is about, sorry. Are you offering a critique or just throwing out ideas for someone to ponder?

If you're critiquing the idea that family/friends won't be able to respond when you need them the most, not everyone has family/friends. So there will naturally be companies that act as account recovery services as well.

I think the critique goes deeper than that. An informal agreement between you and your circle of friends and family is not a legally-binding agreement, and is not enforceable at all. A modern economy needs private contracts that are fully enforceable by courts of justice, it cannot rely on informal agreements like this. Further, you're asking people to become custodians of your on-line accounts? I don't think anyone wants to become custodians of someone else's on-line accounts, even if they are your friends. I find super weird that someone genuinely thinks this a workable idea or that this is something that would be appealing to the general public. It reminds me of survivalism.

The point is YOU have the choice of who YOU choose to fulfill role X, and you can spread the tiny responsibility across multiple entities, while removing the liability that one of them could just impersonate you anytime. It’s totally up to you, and you can recreate the legal custodian relationship if you wish, but then they or their employees would technically have full power over your account.

Are you the kind of person who LOVES signing in with Facebook, and trusts them to safeguard your keys and identity in their nice centralized database — if anything goes wronng you’ll have the legal system, another nice top-down system of laws and enforcement, to correct everything, right?

I don't have Facebook, but from what I gather they are not in charge of safeguarding anyone's keys or identity. So... yeah, I have trust in the rule of law. For example, banks pay millions every day to customers in compensation for improper charges as a result of court rulings. If you can't trust the rule of law you have bigger problems than worrying about Facebook stealing your identity... but I'll leave it here.

> If you can't trust the rule of law you have bigger problems than worrying about Facebook stealing your identity

Most individuals cannot bear the weight of corporate lawfare. Congrats on winning a nice case and getting a good pay out. but that is survivorship bias.

You got it the other way round. I haven't won any case.

This is just a particular implementation of "reset password". I don't have legal recourse if the sites I use don't implement a good account recovery protocol... They could implement it as "if you lose your password you're SOL" if they want. It's just a nicety for me to be able to "implement" this myself by configuring a set of people or of technical steps I can use to recover my account.

> An informal agreement between you and your circle of friends and family is not a legally-binding agreement

You’re right, it’s much deeper and more significant than laws (which are broken literally every day).

> A modern economy needs private contracts that are fully enforceable by courts of justice, it cannot rely on informal agreements like this

That may be what an economy needs, but what a society, culture, or civilization needs is the deep bonds of human relationships. Those things are built on families.

> I don't think anyone wants to become custodians of someone else's on-line accounts, even if they are your friends.

Have you ever helped your friends or family? Have you ever been helped by them?

I find this a necessary question to ask because the only people that critique this idea are those that do not have any real friends or family. And to be fair, I already addressed that in my original comment (there will be services / companies to help people in that situation).

My family and friends aren't qualified and don't have the necessary medical equipment to keep me alive on the way to the hospital. Also, they're busy during the day and sleep at night and I don't want to burden them with being available 24/7 to come to my aid. Also, like most social situations, people generally assume that others will care enough to respond, which leads to no one responding.

Emergency services are generally funded to be able to respond to an emergency, are available 24/7, and know that if you call them, they're the ones that need to come. I trust them nearly absolutely to come if I call them, though I do know that if they are funded poorly, they may come too slowly to help me, which is why I want them to be funded properly, and in a way that operates at a loss (I don't believe these services need to turn a profit).

If I'm in a rural area, I know that response times are going to be very slow, but they will also be slow if it's friends or family, because it's a rural area, and everyone lives far away from each other. If it's the middle of the night, my assumption is that my friends and family won't be reachable at all, but maybe the police will be.

Neighborhoods funding their own services mean that rich people get great service and poor people are on their own.

I understand that this is how libertarians think, and this is why I think that libertarians live in a fantasy world where they believe things will work out exactly the way they've planned them in their mind.

Sounds good, until you get to the last two hypotheticals, which sound awful.

You choose your set of individuals. I trust my sister far more than any corporation or judge.

Judges know a thing or two about family disputes. Maybe you'll find yourself in need of one, eventually.

Yes, and lots of people end up on the bad end of the judicial system as well. There is no perfect solution for this, it isn't computer code, it's people, it's all just shades of gray. I think the "expected value" of having people I trust help me keep my accounts safe but recoverable is much higher (and much much cheaper) than that of thinking I'm going to successfully sue someone to achieve the same thing.

Well, there's a reason why civilisation is built upon private contracts and the rule of law, and not upon unenforceable agreements between individuals. It's because, even though none of the solutions may be perfect, one is clearly better than than the other.

Civilization is built upon both familial / social bonds and legal rules. Neither system would work well on its own.

> There’s even a site that will generate vanity addresses for you! (Don’t use it.)

Asymmetric split key generation is a well understood cryptographic technique, sure, don't trust a service that just spits one out but if you give them a public key to use in generation then you can verify the result only works with the paired private key you control.

Nitty gritty of the elliptical curve cryptography involved:

https://bitcointalk.org/index.php?topic=81865.msg901491#msg9...

https://bitcoin.stackexchange.com/a/75523

Intriguing. How do you do backups of your keybase keys? My understanding is that this is the purpose of being able to export private keys for blockchains.

Just replace the backup with what I said: an “account” you pre-created, which you don’t use until you need it.

You can create N “accounts” which are not tied to devices. In fact, that would also be good for a quotum to “outvote” whoever found M devices, to repudiate them. (Otherwise it would have been a stalemate.)

Using “virtual accounts” you can bootstrap from one private key + U of V keys from a set that you gave to friends + family, to reconstitute your account should something go wrong and you lose ALL your devices.

So, just to be clear: either a private key is stored in a device, or it would have to be paired with U of V other keys that you gave out, in order to activate a new device one time with that key.

But the real provisioning and repudiation of keys would be in a smart contract on the blockchain. That is — pardon the pun — the key to the whole thing :)

Most of crypto is very bad for privacy. Somehow people think all of crypto is private because someone used Bitcoin to buy drugs 7 years ago.

I’m not looking forward to a world where everything about everyone is carved into the blockchain forever.

Alternately, Web3 marks the moment that advertising transforms from an industry into a product feature.

Consider that our basic premise of advertising is that we must compete for attention to influence consumers. This premise requires a "captive audience", and must insert the ad into some span of time and space where it is viewed.

Because these businesses will also operate on Web3, their data is also ripe for the picking. Not all of it, but enough that users will have products available that evaluate the business and make suggestions about the course of action to take, or even select a purchase given a rough specification. If insufficient data is available, independent reviewers will endeavor to create it. And businesses that resist will become dinosaurs. The influencing function of advertising will then begin to collapse as more people delegate decisions to an AI, adding a privacy barrier. Increasingly, businesses will build to the specs generated by analysis of this request data. There will be no market for selling the data, except that of voluntary user disclosure.

I wonder how all this Web3 thing works with GDPR, especially with part 17 - "Right to erasure". Seems like if you go down the blockchain route to store any user data, you're pretty much non-GDPR compliant from day 1. I have a feeling the judge will not like the "it's a feature, not a bug" argument...

As long as the blockchain doesn't have "personally identifiable data" you're fine.

When a web service is requested to "delete your data", all they need to delete is your name, email, phone etc. and still keep your activity data.

That's really not how GDPR works. If it's possible to derive PII from some other data (like activity data), then it's also considered PII under GDPR.

And it's definitely possible to derive PII from activity data.

You can fight what is PII in European courts. I'm just saying how things actually work.

When a company deletes you as a user under GDPR, it does not delete your purchase history for example, because it needs to link the transaction that happened, what was sold, for how much, from which merchant etc. Just because you ask to delete your data does not mean everything associated with you is deleted. The merchant for example will still need to see that the item was sold to someone for a given price etc. That data was not just "yours".

That's because there are other legal reasons to keep that kind of data.

But that isn't a reason to keep all kinds of data related to a user account.

Monetary transactions, claim data for insurances, but not support requests for instance.

Support requests can be kept as long as you anonymize them. This is true for everything else. Look up "GDPR anonymyzation".

Yes, if you mean true anonymization – i.e. not pseudonymization (e.g. replacing names with a pseudonym) or in any other way retaining personal data (internal identifiers, usage patterns or attributes that combined result in a unique enough fingerprint to identify a specific person, etc.)

Why does it need to link the transaction that happened to a specific account? They certainly need to keep inventory records, but it's enough to note that an item was sold without linking that to a specific account.

Are we using different definitions of the term "purchase history"? To me, that means a list of purchases by a single person. You can derive PII from that, so that's a no-no under GDPR. If you just have transaction records but can't group them by customer, then that's (probably) fine.

Also note that sometimes you're legally required to keep PII for a certain amount of time, e.g., invoices with PII for tax reasons. GDPR says that's fine but you have to keep them for no longer than needed.

> Also note that sometimes you're legally required to keep PII for a certain amount of time

I'm not talking about legal reasons. You can keep the data regardless for any purposes as long as you remove the "personally identifiable information".

For example, you will be required to remove their exact address but you can replace that with the general area (eg. post code) as a part of your anonymization.

I don't see anything so specific as "personally identifiable data". The GDPR Article 17 says all personal data. What makes you think activity data is fine to keep?

Look up "GDPR pseudonymization".

Thanks, I have. The first result on Google[0] says that pseudonymized data is still considered personal data.

[0] https://dataprivacymanager.net/pseudonymization-according-to...

Yes, pseudonymization is reversible because you keep the PII separate from the rest of the data associated with the user.

The reason you pseudonymize in the first place is that you can delete the PII part, after which the data becomes anonymized, which you can keep.

That's incorrect. If the psuedonymized data is enough to fingerprint a specific individual (maybe with some additional data), the GDPR says you still can't keep it.

Correct! Which is why you need to pseudonymize correctly.

Good luck truly anonymizing the activity feed of any user on any social network. Good luck anonymizing support tickets that include stuff like "I'm running this version, on this device, and trying to do that". Good luck anonymizing messages between users that includes nicknames.

Blockchain is such a terrible idea because even if today you're thinking "Well, I guess storing X is fine, it can't identify a user", tomorrow you might learn that it can. And there will be nothing you could do about it. And we haven't even started talking about bugs, which all software has (including DAOs and smart contracts...), that might cause you to store stuff on the chain, publicly and irreversibly. The future is going to be fun!

For GDPR, that could not be enough. If the data you hold can be used to indirectly identify someone, that information is subject to GDPR. The main example is when information can is combined with other information that allows individuals to be identified.

The GDPR very broadly defines ”personal data“ as: data relating to an identified or identifiable natural person.

So no, deleting you name but keeping your activity data is not sufficient.

Most likely, the upcoming generation web3 big tech companies won't be founded in Europe. The US could go either way. Singapore, India and much of SE Asia will probably do well, though.

Where a company is founded is irrelevant with regards to the GDPR. If they have users in the EU, GDPR applies.

Yes but it matters for how the law is capable of being enforced. The EU does not have as much leverage over entities that have no legal presence there. This is not a value judgment, it is just descriptive.

That's not how sovereignty works.

The EU has no jurisdiction whatsoever over a service outside its borders. It can ban various apps and then penalize its citizens for using them (or just block them) but it can't dictate to companies in other countries what they can put on the internet.

In the extreme case, consider that Tencent and other Chinese companies with social media platforms are required by law to collect user information that violates the GDPR. Do you think they'll follow the laws of their own country where the business is based or EU regulations?

“An undeletable cookie” seems seems like a contrived usage of the word “cookie”. That’s like saying an email login is an “undeletable cookie” because people typically use the same email to login to multiple sites. Just like with an email, an address used for different websites can be changed and ether can be transferred to new addresses (in a way that reduces the chance of linking the old and new accounts) using mixers.

With regard to data ownership, I wouldn’t say the requirement of that is linking everything you have to one wallet. Many people typically have a private wallet and a public one. One that’s used to publicly attest ownership of things for bragging rights and may be sued to link some sort of personal identity, and one that could be used solely for defi applications or tied to another social identity completely.

Another cool thing about using wallets on the web is that it’s always been a design requirement to always obtain user consent when sharing address info and of course when making transactions as well. At the very least when you’re sharing your information it has to be intentionally done unlike with cookies that I can set on any visiting user’s browser (GDPR regulation doesn’t technically stop me from doing it non-transparently, it just has legal consequences if I do)

Further, anyone can put anything into your “wallet” without your consent or approval, and removing something horrible added without your consent can then trigger a smart contract that can do even worse things to your unmodifiable “profile”.

Dumpster fire of dumb.

There's no such thing as putting something "into" a wallet.

But wallets can be swapped all the time, even moment to moment.

Each time you are adding more immutable data to the Blockchain, which cannot be undone. Each of your identities - to which all of your digital content is tied - is permanently associated with that ID.

So if you just have throw-away wallets, how do you use that with your content? You cannot "own" your content if you keep having to abandon wallets to get a fresh advertising ID. And if you keep several different compartmentalized wallets around, as soon as you move something from one to the other or use a different wallet with the same login, you have permanently and irredeemably linked those identities.

Yeah but at that point it’s up to the user. There’s a way to stay anonymous but just like anything relating to privacy, users have to be well informed about their risks and mitigations to be able to manage their data effectively.

You can generate a new public key at any time.

Wouldn't doing so detach past activity I do want aggregated?

How is a site I'm interacting with supposed to know it can use my previously-configured settings if I'm showing up with a new public key every time I authenticate?

I read this as essentially a single origin supercookie. So each origin (web site) has a unique public key for you, but it’s immutable.

So, you want privacy but without privacy?

Yes. I get that in the current ecosystem via doing business with sites I trust, and ceasing to do so when I cease to trust them. It's not impregnable (sites get hacked, companies get sold), but it's preferable to an alternative where I must either stay one-time-padded continuously or have my activity recorded in a global public ledger.

You could use a public blockchain key to log in to a site you trust which stores all of its data somewhere private and have this same functionality. It is only transactions you write to a public blockchain that are public. The act of authenticating using a key stored there does not write anything down.

Privacy as in having control over who gets what information, and being able to change who and what over time

If you want activity aggregated, use the same address. Not sure what the complaint is.

Reply on news.ycombinator.com