Hacker News Reader

Shouldn’t it make most sense in contexts exactly like Steam, where the merchant can take the goods back in case of a fraudulent transaction?

It did, and they did. That's probably exactly why they allowed it like that in the first place - if the transaction is fraudulent then simply remove the user's entitlement, but offer instant access for better experience.

“Fraud” and “simply” don’t belong in the same sentence. This is coming from someone who’s spent years working with many fraud departments in major companies. Wack-a-mole doesn’t work, they just scale their account creation.

Well then please explain what is difficult here. You initiate a purchase with bitcoin, valve instantly grants you access, then few hours later valve checks if the transaction has been confirmed in the blockchain or not. If it hasn't, then the entitlement is removed. What's not simple about it?

What about people who do this 10x a day to play free games forever?

How much worse is this than people who do this 0.5x a day with 48-hour returns?

You let the user do it twice then ban them from using bitcoin payments?

They already answered this question in their previous post... The fraudster scales up their account creation. You ban then after 2, they switch to a new account... which you ban after 2, and they switch to another new account, which you ban after 2... and so on, ad infinitum.

Accounts here can be blocked after an average of 10 minutes.

But then what? Someone downloads and plays a game for an hour, gets banned, then goes through the whole process of having to do all that again?

Doesn't seem like a good user experience. All to get a free hour of gameplay?

There are easier ways of pirating games. The only possible usecase for this kind of fraud would be if you can buy digital items, transfer and then sell those digital items.

But even then, that might be difficult to do all in 10 minutes.

I don't care how well it would work, and I'm not arguing any of your points... I don't even understand why you people like these video games, so I'm the last person who should opine on what constitutes a good UX.

My comment was intended to point out that it's parent comment didn't bother to read the grandparent comment, which had clearly answered the question it was asking.

But then the parent comment author edited themself, I assume to avoid looking foolish. So I guess it's irrelevant now.

> I don't care how well it would work

Ok, so then I guess it isn't really a problem then. The whole just ban them "ad infinitum", is totally a reasonable solution to all of this.

...or so say the guys sitting in the cheap seats. In practice, it may well be more complicated than that, and involve all sorts of other tradeoffs. It usually is.

> In practice, it may well be more complicated than that

No, its not. If an account gets banned after an hour, all that happened is that the user got is 1 free hour of playing a video game. There are easier ways of pirating videos games than that.

So really, thats good enough.

You tried to claim "The fraudster scales up their account creation". And I am telling you, that if a fraudster wants to play a video game for free, there are much easier ways of doing so. By just pirating the game, for example.

So that specific point is wrong.

> or so say the guys sitting in the cheap seats

Actually, it seems like you are the one sitting in cheap seats. Because you have already straight up admitted that you don't really know much about the industry. Whereas, people like me, who do work in the games industry, do know about what a good user experience is for people playing video games, actually know more about this than you do.

So, in other words, you actually should "care how well it would work". Because how well it would work, is integral to the attack vector.

I worked five years in the game industry, mostly on (guess what!) security and infrastructure for a Very Large F2P game. I was nowhere near game design or any part of the game loop itself, but I did a lot of work on our account system.

> > In practice, it may well be more complicated than that

> No, its not...

You keep assuming that the information you already have is ALL of the relevant information about this issue. You're just a Monday Morning Quarterback, and you have no way to know whether the situation is more complicated than this. So you can't possibly make absolute statements about this... Unless you happen to work at Valve, on something that has given you actual experience with the Steam account system.

> the relevant information about this issue

I already told you the answer to this.

There are easier ways to pirate a game than this, as getting banned after an hour is not a good user experience, and the user would be better off just pirating the game.

For which you had no response.

The only specific reason that you gave, which was "they could just make more accounts" I refuted by describing how it would be easier to just pirate a game.

And then you refused to address the refutation, because you didn't have a response.

If you didn't have any specifics arguments about that point you could have just said so.

Because I was talking about the actual, specific arguments, about why your previous statement of "make infinity accounts" was wrong, and you are now continuing on being incapable of giving a response to that specific statement.

Making new accounts isn't trivial. You can add friction from suspicious attempts like requiring a phone number or making them wait some time. Some determined attackers might slip through but at the end of the day all you are losing is some bandwidth.

The UX sucks sufficiently that it shouldn't affect their bottom line.

Seems pretty marginal - you’d have to restart, on average, every 10 minutes.

Are you suggesting that there may be a human being on earth who has the tenacity to enjoy games while creating new accounts, pulling off successful bitcoin heists every 1-2 hours and risking losing their save games? And doing that forever?

This website is just surreal.

> Are you suggesting that there may be a human being on earth who has the tenacity to enjoy games while creating new accounts, pulling off successful bitcoin heists every 1-2 hours and risking losing their save games?

Most of the “tenacity” parts implied there are automatable, so it's mostly about “can people enjoy the paid-for experience of a game with ephemeral rather than durable accounts”, which, well, really depends on the design of the game.

You don't know what implied means.

Force brand new accounts with no purchase history to wait for the transaction to confirm?

What whack-a-mole? Just automate it so if it gets double-spent, you revoke the key. If you really feel like it, cancel the account if they keep doing it, but I'm not sure you even need to bother. What have you lost? The bandwidth to download the game?

I think it does, there are levels of fraudulent activity. Calling "simply" doesn't mean to make it benign.

Then the next question is this: why did those 50% even bother to try? Reselling for less btc (or fiat) keys that would soon be revoked? Has piracy decayed to the point where desperate consumers would really fall for something like that?

> keys that would soon be revoked

The BTC acceptance didn't even vend keys, it was just access on the primary steam account.

>Steam back in the day used to accept 0 confirmation bitcoin spends

This is ridiculous. 1 block confirmation would've prevented near-all transactions from being retracted.

Correct, but why to deal with this at all considering the % of all BTC TXs compared to fiat.

> There are many context where a 0conf tx makes sense, mostly IRL.

I'm curious, like what? To me this is more like "someone wrote a cheque but there's no guarantee that they'll give it to you".

I think more like "someone wrote a check and there's no guarantee it won't bounce". So really, it's quite analogous to how 0conf works in practice.

There are laws in place that specifically handle writing bad checks. While I'm sure double-spending Bitcoin transactiosn is illegal in some way, it is probably covered under a more nebulous fraud-type statute.

I'd be curious to see the percentage likelihood on both, as well, as I suspect they are very different. Would businesses accept checks if half of them were bad?

The general fraud crimes you would get convicted of have a pretty nasty set of penalties. But good luck convincing a federal prosecutor to care about $20 worth of online goods that can be revoked.

I only have a layman understanding of bitcoin, and I've never used a cheque in my life, but would it not be more like "someone wrote you a cheque and there's no guarantee it won't bounce when you try to cash it in"?

Which, again maybe showing my complete lack of knowledge of cheques, is how I thought they worked in the first place.

Personal check is a quasi-promissory note binded whose legal-teeth for recovery through law enforcement. “0-conf bitcoin” has no legal teeth for recovery … yet so there is no enforcement (except for the merchant to chase after the scofflaw). “Money is forthcoming”.

Bank cheque is a bank-guarantee binded by banker self-verification (almost always good) with no takeback (voiding) mechanism. “block-chain confirmed” is always a good transaction with no clawback mechanism so you must wait until it’s confirmed before receiving your merchandise.

“Cash is king.”

> has no legal teeth for recovery … yet so there is no enforcement

In this case, Steam was selling a digital good that they could be fully revoke if the funds never arrived, so the check ability works very well.

It would have been even better on a smart contract platform where the game license would exist on the chain itself and couldn’t be purchased via a double spend, Chain re-orgs would also sort themselves out in a fully on-chain system.

Watch "Catch me if you can" for some fun history on bank cheque fraud.

Here in the UK we had "cheque guarantee cards" - essentially a debit card without a live connection to the bank. When you wrote the cheque the shop took your card details as well and that meant the bank guaranteed to cover the value of the cheque if it bounced (up to a maximum amount). If your cheque bounced your account went into an overdraft for that amount plus a hefty fee. It meant shops didn't need to worry about bounced cheques which customers could pay using them. It worked pretty well.

The US banks I worked with would just overdraw your account (up to an amount), then, overnight, reorder the transactions so the big ones came first and all the little ones came “after” the account went negative and hit you with hundreds of dollars in overdraft fees. I don’t miss the early 2000s.

Don't they reorder transactions even when not overdrafted? I thought my bank did.

That’s because a bank got sued for it, and lost. The banks must now always reorder or keep it in the order it was received. I only bank with the latter ones.

Anything where the buyer doesn't leave 15-30 mins after making the payment or the purchaser is id'd.

This is why it's always been absurd to believe that standard retail purchases would be regularly done with layer 1 base chain transactions. You can't even say that you're ok with waiting for 1 block confirmation, because the time between each block is only an average of 10 minutes, with high variance. It's fairly common to experience over an hour between blocks, or as little as a minute.

The security and confirmation of a transaction is not a function of number of blocks, but time since inclusion in a block and total amount of work (hashes) piled on top of your transaction. Generally, 6 blocks (1 hour on average) is considered sufficient certainty of immutability for normal functioning of the blockchain (i.e. not under active attack). One hour for final, and completely irreversible, settlement is absurdly fast in the context of the traditional financial system, which takes days at least for final settlement, and can be much longer for some international transfers.

Small, individual scale purchases only really make sense to do on higher layer transaction protocols, like lightning network (non-custodial and open) or some custodial networks (like a bitcoin backed Visa credit card).

The only model that makes sense for bitcoin is the layer one base chain is a settlement network, with payment networks build on top that aggregate many transactions into one on chain transaction. Think OSI model for networking.

> The only model that makes sense for bitcoin is the layer one base chain is a settlement network, with payment networks build on top that aggregate many transactions into one on chain transaction. Think OSI model for networking.

Do you think this applies to cyrptocurrencies that has faster confirmations/block generation? Or is only important when there is a lengthy confirmation?

The only thing that really matters is how much work has been built up on top of the transaction. Shorter block times doesn't change that function at all. In a version of bitcoin with 1 minute block times, the security guarantee of 60 blocks is identical to the current security guarantee of 6 blocks, because it would be exactly the same amount of energy going into mining to create immutability.

The amount of work (hashes) represents a real world cost (electricity and mining chips) that can't be undone without massive commitment of these real world resources.

This is the foundation of why bitcoin works.

So basically, people who say their blockchain with shorter block times is "faster" are either fools who don't understand what's going on or conmen trying to pump and dump their bags. Layer one transaction confirmation is about settlement finality, not how fast you can pay for coffee.

I think the faster block times help with conceivably making it possible to use 1 confirmation instead of 0 confirmations for in-person transactions, as opposed to than 60 instead of 6. That tilts the odds of winning in a double-spend fraud attempt in favor of the merchant quite a bit, compared to 0 confirmations, although I don't know how to model by how much.

(You could empirically estimate by how much by looking at abandoned chain history in a real network, though.)

Great answer. This is related in my mind to the framing of crypto "pointlessly wasting" electricity - it's actually a fundamental part of securing the network by making it prohibitively expensive to reverse or modify mined blocks. I'm open to alternatives to PoW but so far it's been the most successful tool to solve this problem.

Proof of work is an alternative to military spending.

Proof of stake doesn't solve this problem, as the top stakers end up needing to control (or get coopted by) the monopolies on violence. The traditional banking system is effectively an obscured proof of stake system.

Proof of work also has the nice feature that every watt of energy used to acquire and secure bitcoin is a watt that can't be used for violence.

You might like the work of Jason Lowry, a US Space Force Major who's currently studying bitcoin conceptualized as a new kind of military technology, https://twitter.com/JasonPLowery

> Proof of stake doesn't solve this problem, as the top stakers end up needing to control (or get coopted by) the monopolies on violence

How so? Assuming (pseudo-)anonymity for stakers is possible.

Long lasting anonymity of stakers is impossible. Eventually, everyone will be tracked down. There's too much power at stake to remain hidden from establishment power forever.

And anonymity of stakers who have unassailable control of the network consensus is exactly the problem. If you don't know who the king is or where he lives, then the peasants have zero ability to countervail his control authority when he starts abusing his power. No sovereign nation or person will sign up for a system like that without being violently coerced, like the US does to keep the rest of the world on the petrodollar standard.

Instantaneous payments exist, I use them several times a day, with traditional banking. I never quite understood what problem bitcoin was supposed to solve beyond anonymity.

These are instant payments, not instant settlement.

Your bank abstracts the underlying inter bank settlement complexity and aggregates all the minor transactions and transfers that individuals do to a small number of larger settlement transactions, likely on the SWIFT network.

When you make a payment, that transfer of money isn't instantly settled. The instantaneous feeling is a service that the bank provides you, but it's an illusion. There are similar experiences being built now on top of bitcoin, like http://strike.me, although the lightning network that Strike and CashApp are building upon actually is instant finality, unlike Visa/Paypal.

This is why comparing bitcoin to Visa or PayPal is fundamentally flawed, since those are both layer 2/3 networks on top of SWIFT.

The proper comparison is with other settlement networks that create finality, like SWIFT. Where SWIFT may take days to create final settlement of a transfer, bitcoin takes ~1 hour depending on how much confidence you want. Where SWIFT is secured by nation state monopolies on violence and global banking gatekeeping, bitcoin is secured by the most powerful computer network on the planet that no single nation state has the power to disrupt. Bitcoin is open and permissionless and anyone and everyone can join. SWIFT is tightly permissioned because it relies heavily on exclusive networks of humans who trust each other.

At a high level this comment is correct, but SWIFT is not a settlement network. A message on the network does not actually represent the movement of money between accounts.

Bitcoin only gives you a pseudonym wallet ID. I'd say it's less anonymous than cash, since everyone has the ledger.

I'm not convinced bitcoin solves _any_ problems at all. It isn't anonymous at all, exchanges are the gatekeepers of the whole system and they have to comply with local laws. Canada was found to be directing exchanges to not allow cashing out of certain coins and putting the coins through a mixer just automatically blacklists them from being cashed out.

Bitcoin does not provide sufficient privacy which allows exchanges to freeze tainted coins and authorities to track down real identities.

so anywhere where a check makes sense, in a trusted or doxxed environment for example like IRL

BCH is trying to keep 0-conf alive by eliminating the artificial block space constraint, removing RBF (which is really dangerous when you have an RBF-flagged grandparent tx that maybe your wallet is t looking for), and implementing double spend proofs to warn tx recipients as soon as a double spend tx appears on the network. It’s obviously impossible for an unconfirmed Bitcoin tx to be guaranteed secure, but the risk can be minimized, which is nice for everyday, low value transactions.

With no RBF how will you prevent TX's from getting stuck in a high fee environment? I know BCH thinks block space should be unlimited which means TX fees will never get high but i don't think the economics works out there.

Miners can set their acceptable minimum fee levels and orphan the blocks of miners who accept transactions with fees that they think are too low. This is essentially cartel behavior with an incredibly strong enforcement mechanism. There should never be a “high fee environment” due to an artificially low block size cap.

And why can't miners do this on BCH?

That’s the point. They can. Are you reading this as me thinking the collusion on minimum fee levels between miners being a bad outcome? I think it’s going to become necessary as the block reward goes to zero, and therefore it’s necessary and good for the system. It’s nice that Bitcoin (including BCH in “Bitcoin”) has a the block orphaning mechanism for enforcing cooperation in this manner. It’s much stronger than enforcement mechanisms available to cartels such as OPEC.

You don't even need to do anything on bitcoin side if you value the 0 confirmation. You offer the goods or services before you receive the payment. It's like a restaurant that serve the food first and receive payment later. The restaurant do that because they trust the customer to pay up.

This doesn't work,people don't know the amount to pay before they finish their meal.

You pay up front in most cafeterias for example.

More likely the reason this isn't done is to allow customers to buy more (drinks/dessert, do so in a single transaction, and tip post-service. Not because the amount is unknown.

Yeah but they verify that you pay before giving you the food. That's exactly the opposite of what op was talking about.

> and you don't trust you customers

Seeing as “remember the password” does about nothing in the desktop app on my own machine near which only I ever come—Steam's attitude toward the users is crystal clear, and I'd say it's weird that they lived almost two years with that arrangement.

Convenience and security are always in tension - and steam account takeovers are very common, as they can have tradable assets worth hundreds or thousand of dollars.

In the same era steam stored your password on disk if you used "remember my password".

Depending on what it is, 3-6 is a _lot_. For something like a Steam game, it seems like waiting for 1 should be more than enough.

I say "more than enough" because can't they just revoke the key of whatever was purchased if it gets double-spent? Seems easy enough.

> it seems like waiting for 1 should be more than enough.

Not in bitcoin land, orphaned blocks happen a few times a day: https://bitcoin.stackexchange.com/questions/2170/how-often-f...

If you're willing to tolerate this risk then by all means go for it.

(I'm with you I think steam should be able to tolerate this risk.)

FYI the answer in that link is from 2013, not sure how much it's changed since then but I doubt it's the same today.

A feature of bitcoin is it’s stubborn system stability and resistance to change :)

One of the odd benefits of this is that old forum posts and discussions have a good chance of being relevant.

Although I will say lots of progress has been made in p2p message propagation (since this isn’t directly a part of consensus) which could definitely prevent orphan blocks, so you may be on to something in this case.

Yeah, the biggest effects on the number of orphaned blocks should be the number of distinct mining pools and the latency between them. I could see some of those numbers changing enough to matter since 2013, but I'm not all that sure.

But it is much riskier for the fruadster after 1 transaction. With 0 confirmations it is almost certian that a new transaction with a higher fee will replace the older one. With 1 confirmation you have to either swing a lot of mining power or hope that your block becomes orphaned. The fraud likely doesn't make sense anymore if you lose your money 24/25 times instead of almost never.

Then Steam can accept that lower risk and mitigate further by then revoking game access.

> Steam back in the day used to accept 0 confirmation bitcoin spends. This means that the transaction has been gossiped on the bitcoin p2p network but had not yet been mined into a block and thus had minimal finality guarantees.

Why? Why would anyone ever accept this?

Because I am pretty sure it isn't true. I am one of the few people who used bitcoin on steam during this time and I don't remember it ever being instant. I would like to see clarification to the top point because I don't think it is true...

In my opinion when he says "fraudulent" he is probably talking about how people would use it to avoid bans. Steam would track banned users that remake accounts by checking their CC. They would also verify the people by making sure the address on the CC was close or the same to the address on the account.

With Bitcoin you could avoid any tracking from remaking an account which leads to more 'bad actors' using Bitcoin. For reference you needed to spend something like $5 - $10 to enable trading on the platform. From their people would phish, scam, and break the TOS on the account.

I used bitcoin all the time on steam and there was a time when they were using 1 conf but the majority of the time they used 0 conf with bitpay. Couldn't find anything official with a quick goog but here is a reddit comment from 5years ago corroborating 0 confs on steam: https://www.reddit.com/r/Bitcoin/comments/6arpvq/comment/dhg...

It is possible I am incorrect then. I remember it always taking 5 - 10 minutes before any funds hit my account. I never actually counted the confirmations so it is very possible I am just wrong and they were doing 0 conf. At the same time I am surprised BitPay would have been structured this way and not experience a major attack...

Because they can revoke your game from your steam account if the transaction doesn't get picked up.

Alright, that is a point I did not consider. This makes me feel that it is difficult to use Bitcoin for expedient transactions, due to the fact that you'd need to put high fees on each transaction.

It was my understanding that RBF does not work by double spending, but by making a transaction with a much higher fee that depends on the parent transaction which means you would need to mine both to get the higher fee.

This is known as CPFP (Child pays for parent)

> The bitcoin community moved away from accepting '0 conf' transactions pretty much everywhere because of this reason.

They moved away from 0 conf because of the high fees, making payments sometimes get stuck for days or weeks, making 0 conf much less trustworthy.

Appreciate the explanation, thank you.

I was imagining it was something on the line of drug cartels and human traffickers laundering money through trading game assets.

How would you prevent the fraudulent transactions you just posted if you accept 0 conf proof?

>SSH caused serious security problems

>I don't know what "null cipher" means but it sounds very technical and is probably the best one.

The whole UX excuse is a bit silly considering modern Steam games take an hour to download anyway.

You say wait 3-6 blocks -- which makes sense for large transactions, but are there cases where 1 blocks have been getting reversed? Is this happening now because some parts of the network are on stale or diverging info?

These are called "orphan blocks" and they happen very regularly, like everyday.

https://bitcoin.stackexchange.com/questions/2170/how-often-f...

That’s not really a double spend issue, though. Usually the blocks are found within a very short time between them. Unless the double spender is coordinating with a miner, orphaned blocks are a non-issue. It’s 51% attacks that would be the real problem.

This is the answer to what I was asking. Not sure who I offended to get downvoted.

edit: I guess my point is that if you're selling coffee and making 100 $6 transaction, going 1 block deep probably isn't a big deal. If you're selling one car a month, you might not want to risk the small chance of an orphaned block -- but a one hour hold on title is a whole lot better than waiting for a check to clear.