Comment by jedsmith | Hacker News Reader

Comment by jedsmith | Hacker News Reader

The comments on the story, and the first round here, make me worry that people are missing the point. Particularly this one:

> VeriFone's point (albeit a stupid one) is that the path from the card reader itself to the iOS device is not secure, so it's a hardware "problem."

That point isn't stupid at all, and is actually a legitimate gripe. It is my understanding that the Square device is only meant to be used with the Square payment software. Is that wrong? If that's the case, it shouldn't be as easy as VeriFone claims it is to skim a number from the device. I don't follow card skimming, but I would think that the point VeriFone is trying to make here is that Square is making it easier for low-tech criminals to implement a skimmer. While it is possible today, I've never heard it described as easy.

Another of the commenters mentioned that professional card skimmers have their own devices. That's fine; that isn't the problem here. Professionals will always do things better than the entry-level criminals. However, from my reading, it sounds like there is the capability here to write a shrink-wrapped iOS app that talks to the free Square reader, and you have skimming for a total investment of an iPod touch (or lack of investment -- you are a criminal, after all). Which is the problem that VeriFone is bringing up, and I halfway agree with given the limited evidence presented to me. We're way too lax with credit cards.

Does the Square app really need to know the card number at all, aside from transmitting it upstream? What is the vector here? It seems like the Square device should be a black box that yields only some kind of payment token, and the app only gets the cardholder name (from their screenshots).

Square's own security page[1] makes this ambiguous, too:

> Card numbers, magnetic stripe data, or security codes are not stored on Square client devices.

I've never used Square, so I might be full of hot air. However, I think instantly painting this as not an issue simply because it came from a competitor is doing a disservice to the potential (real) vulnerability here. Read the information, not the source...

[1]: https://squareup.com/security

Are you serious?

The victim in this scenario is handing their credit card to the attacker. That's game over. It doesn't matter whether the attacker has a Square, a VeriFone device, or a piece of paper for writing down the card number. If you hand your credit card to someone malicious, they can figure out how to do bad things with it.

I'm a big fan of Square, but it's important to remember that their solution doesn't remove the traditional risks of credit card usage. But then, there isn't yet a widely-deployed, field-tested alternative, at least in the US. The advent of NFC payments may put reasonable crypto in the POS flow, but I'm sure it's matter of time before side-channel attacks are discovered in the burgeoning NFC solutions.

Security is hard. But VeriFone isn't talking about security. They're using security as an anti-competitive stalking horse, and that's a disgusting way to do business.

> Are you serious?

Very, and I will dig my heels in on this. For the record, your point would have been just as clear without this assertion that my point is stupid.

> The victim in this scenario is HANDING THEIR CREDIT CARD TO THE ATTACKER. That's game over. It doesn't matter whether the attacker has a Square, a VeriFone device, or a piece of paper for writing down the card number. If you hand your credit card to someone malicious, they can figure out how to do bad things with it.

So everybody should just stop doing security, since people will figure out ways around it? Why bother? The Square device, as another commenter points out elsewhere in this thread, sets the expectation of a legitimate transaction. It's pushed, it's marketed, it's being set as the new brand of credit card processing.

As such, Square have a responsibility to make sure that their products do not make skimming easier. And guess what? VeriFone found a way to skim, easily, with their product.

This whole "but you can do it on the black market for $20!" argument is, pardon the Latin, fucking ridiculous. I seriously can't believe there are people in this thread, who identify themselves as IT specialists, siding with Square because credit card processing is a lost cause. Internet Explorer 6 is pretty much a lost cause for Web exploits. Does that mean the rest of the browsers should not protect against Web exploits? This argument makes absolutely no sense. Who cares what others are doing? Do your product right.

> I'm a big fan of Square, but it's important to remember that their solution doesn't remove the traditional risks of credit card usage.

True. But there is NO REASON to make the vector EASIER. PERIOD. Simply because there are real risks with the vector means that we should take strides to make the vector harder.

I don't know about you, but when I hand my credit card to somebody, I don't actually check out the brand name of the reader they're using. Whether the little gadget says Square or Circle or Rhombus on it has no effect on my expectation of a legitimate transaction.

If legitimate people using Square could have their numbers stolen because the upstream communication to the server wasn't encrypted, that would be a problem. That would be something to secure.

I mean, if the Square device couldn't read the raw numbers, then people would be complaining about how they own the hardware and they should be able to do whatever they want with it.

"This whole "but you can do it on the black market for $20!" argument is, pardon the Latin, fucking ridiculous."

Nice way to add 'black market' in there to try to dismiss a valid point. You can get card readers from legitimate businesses. They are valid peripherals used all the time. Do a search for 'usb credit card reader' on Google. All those results are not black market.

As an aside, it's funny how you have a problem with GP saying 'Are you serious?' and thinking it implies your point is stupid but then take the exact same dismissive tone for your whole post.

"I seriously can't believe there are people in this thread, who identify themselves as IT specialists, siding with Square because credit card processing is a lost cause. "

The fact of the matter is that people hand their cards over all the time to pay for things and no one ever looks to see what kind of device it is. When you are paying for something in a shop, do you check out the credit card processing machine first to make sure it's secure? When you hand your card over to the waiter to take to the back and pay for your meal, do you follow him to make sure he isn't swiping it into another machine or writing the number down on paper? Nope. And the fact that the square reader is a plaintext device, like a lot of the readers on point-of-sale systems, really doesn't mean anything.

Jed, there are exactly two directions you can go from here, and I advise you to pick one of them instead of standing in the crossroads hollering and waving your arms.

On one road, you make the argument that it is fundamentally unsafe to accept credit cards with commodity mobile computing devices, because there is no way to do chain-of-custody control with insecure magstripe cards. There is a reasonable case to be made here. I'll disagree with it, but you won't automatically lose.

On the other road, you can provide one of more specific steps Square could take to make their commodity mobile card reader device meaningfully more secure. Be careful on this road because it is dangerous; people, myself included, will be standing on the side of the road shooting at you. Remember, anything you come up with must meaningfully improve the security of a device that can only interact with insecure magstripes.

You can also walk your argument back, which is what I'd do if I were in your shoes.

> You can also walk your argument back, which is what I'd do if I were in your shoes.

Good point. I'm just leaving my argument where it is, since I've poignantly lost. The way you summarized it made something clear to me, though, so thank you for your response.

I had a response typed up to the "we need to switch to $X" elsewhere in this thread, then discarded it and went back to work after writing a much smaller one. That's probably the best way forward from the whole mess, since it is indeed true that magstripe technology is outdated and insecure. It's amazing to me that with the prevalence of strong crypto we haven't come up with something better that has been mass-adopted. Until Joe's Gas-n-Snacks in Bimbleberry, Kansas accepts my new-fangled card, I won't consider the replacement mass-adopted.

There's an interesting aside: Square hurts adoption of new-fangled payment technologies because it caters only to magstripes.

However, I still disagree on mere principle with "it's already broken, so no need to do it right for ourselves". That's a dangerous path to take in itself.

I think your perspective on what's broken is an issue.

It's one thing to say "everybody else is crappy, we can be crappy too." I agree, that's bad. Doing better can be a great differentiator.

It's another thing to realize "the input we receive is broken, our output will be too". Garbage in, garbage out.

Square, having decided they are going to be in the business of reading magnetic stripes, cannot make those magnetic stripes any more secure than they already are (not). They could maybe do better/right by being in a different business, but they can't do the business they are in any better.

To be fair to al3x, I was genuinely wondering if you were serious.

(I'm speaking only for myself with this comment).

You know it's not a real security concern because it can't be mollified. The kernel of Verifone's argument requires that nobody ever accept credit cards with little white card reader dongles, because a little white card skimmer dongle is trivial to construct. If Square dongles had googley eyes and pink antennae, Verifone's argument would involve the ease with which criminals can procure googley eyes and antennae.

What's pernicious about this argument is the way they've tried to make it sound as if there's a real technical concern here, when in fact the real issue they're articulating implicates the entire business model.

Thomas,

You're sorta not understanding what goes on in a verifone device. I would advise actually constructing and inserting a rogue verifone PED, it is non-trivial.

First, understand that when you swipe your card at mcdonalds, you're not trusting the guy behind the counter, the manager, the district or anyone else physically present or even mcdonalds. you're not trusting their malware laiden office computers, nor their network, et cetera. You're trusting the terminal. And the terminals, depending on model and service/etc, all have cryptographic MAC addresses "that are burned in" (lol), and they're externally by verifone. So first hurdle is that you can't just get a job at mcdonalds and ebay your way to a better life, when you switch out the device, it gets detected and blocked, and i imagine once the burger flippers figure out that its not broken and someone tried to own them, they pull the cameras, et cetera.

Next, the device has a myriad of tamper-detection qualities which make it non-trivial to get into and still have the damned thing still function. This is again, dependent on which device is in play and so on-- getting in successfully obviously can be done, but you're getting sorta specialized. Next, the card details/pin/etc all get encrypted at the device never to see plain-text again until it's "safely in the cloud".

That's the theory anyways, there's some problems but it's not that joe bubblegum is gonna download an 'app' and extract my credit card details from his employers iphone. If pressed for an opinion, I'd say the entire idea of getting people used to paying by swiping their cards through some guys phone is maligned and going to be horribly problematic. This is sorta like SSL, there's another maligned trust heirarchy that exists and makes sense sorta.

So, basically my point is, verifone is not entirely incorrect here. It's not that "a little white card skimmer dongle is trivial to construct" or any other hyperbole. A PED or other input device that I can dump my card info from is a problem.

There's another entire trust relationship going on here which everyone seems to skip: the most common use-case of dealing with a representative of a third-party who is less trustworthy than the third-party itself ('Alice meet George, embassary of Bob, don't mind the collasped veins, he's legit!').

This is all interesting but it has nothing at all to do with the point I'm making. What I'm saying is, Square could come up with an encrypted obsfuscated reader-to-iPhone protocol that, like Skype, would take the better part of a year to fully reverse. Verifone would then say: "yeah, but how does anyone who looks at the reader know that it isn't a fake Square dongle that doesn't really run that protocol?"

In that respect we are saying the same thing: the entire platform is, by Verifone's standards, unsuitable for accepting cards.

But that's not what Verifone is actually saying. They picked a much more specific argument against Square. There are two reasons that makes sense for them:

(1) because there will be another 57 similar arguments they can make if Square does something to mitigate that first concern, and

(2) because if they made the intellectually honest argument ("It's not safe to accept cards with any iPhone app, Square's or otherwise") the competitive issue driving this complaint would be too transparent.

I think you sorta misunderstood, and but also possibly hold some misguided beliefs (like i dont?).

The first thing is, there's a lot of things Square could do, and we could both live on the moon eventually also. Their current implementation is a skimmer with all of the logic and safety in the software with all the guarantees of PCI network security beyond that (lol, i bet its salesforce or ec2). On Verifone's end, it's not like we're talking some software-based encraption. Think hardware based asset tags to compliment client certs. Surely it's self-evident that the ease of obtaining what at the moment at best could be a client certificate from a phone and recovering the information from a tamper-resistant device that goes dark if you take it offline too long (rendering the information useless). If they took their device and made it less like a mouse and more like something that tried to tunnel through the stupidest medium ever, I could live with it a little more and see the 'they are stupidly insecure' argument as fud.

Second, you realize it took 1 person about a week to reverse, document and develop the PoC for blocking skype (me) and not much longer for a crypto-library from it (accuvant ryan)? A year is a vast over-estimate and im pretty slow at it.

If the overall debate pertains to whether verifone did this right or not, I'd agree with you. Although, I have to admit cheering for them with their 'and we will prove it with an app@#@!' part-- if they were consultants they wouldve gotten invited to speak at blackhat.

If I had to take a stab at it, after looking at square a little bit. I'd say that if they made the honest argument, Square would try to throw lawyers at them. The honest argument is, 'have you looked at these guys? they're bordering on a criminal-enterprise-by-design, albeit rubber stamped by PCI'. I mean seriously? a point and a half more if I don't have the card and want to enter it on your website instead?

(Also the hubris in their research & disclosure policy made my blood boil)

Anyone in the world can make a little white dongle that reads credit card magstripes. What is the point of making an ultra-secure white dongle? How would anybody know you're using the ultra-secure one?

Also as an addendum; the thought process about hardware reversing is not entirely correct. You will run into things like can't build a device that can sniff that fast without a substantial budget; how much is that microscope? That just have no real parallel in software

I've never used the Square device either, but I don't see this as a hardware "problem". This is no different (to me) than a USB card reader that operates as a keyboard. You could scan the card into the vendor software, or into Notepad. Either way, it isn't a hardware flaw.

Again, please answer this: why does the Square app need to know the plaintext number? That is what makes it different from a USB card reader that operates as a keyboard. One is designed to accept payments, and the other is designed to steal numbers. We're comparing apples and oranges here.

The path from the Square device to Square's servers should be completely untouchable to any app. That's what VeriFone is saying. Simply because there are alternative black-market products that encourage crime (and cost money, by the way), there shouldn't be a free product that simply hands you card numbers if asked nicely.

It's a matter of cost vs. benefit. Designing an iPhone peripheral that has built-in crypto is extremely expensive (or any embedded system with strong crypto for that matter), especially if the information you're trying to protect is printed on the other side of the friggin card. The fact is, as a consumer, you can't know if the device that's going to read your card is legit. This is true even if you're looking at a state of the art IBM POS system or a tiny plastic widget with a coil of wire attached to a headphone jack.

I think this is the most important point in this thread. Verifone (and the other big players) makes expensive card readers. Square is disrupting that with a card reader so cheap that they can give it away for free. Verifone is using a standard anti-disruption argument: we're not over-engineered because everyone needs this level of complexity (crypto in this case). This would potentially break Square's business model.

Aside from the part where a vendor is required to keep your card number for refunds...

Square (as I see it) is a two part solution: hardware to read your card and software to handle the transaction. The hardware is talking to whatever software is running. As a customer, you hope that software is legit, but you never know. How is this any different from handing your card to anyone else?

USB card readers aren't black market by any means. This is how most POS devices read credit cards. It used to just be straight serial connections. It seems you're under the false assumption that the hardware path from your card to the software was ever secure.

If Verifone really cared about a completely protected path for card numbers, they would disallow all USB card readers and require a secured path from reader to their clearing house. Unfortunately, that would remove credit card processing for many (all?) vendors.

They are blaming Square for a security "flaw" that is present in their own system.

> Aside from the part where a vendor is required to keep your card number for refunds...

Vendors aren't required to retain the full card number. They are required to retain the last 5 digits (or last 4, I'm not 100% positive), as well as the transaction time. When a chargeback or refund comes around, they'll cross reference multiple forms of identification on the transaction, but they'll never have to handle the full card number.

All card readers are a two part solution. A hardware reader (USB, serial, whatever) is useless without some sort of {firm,soft}ware behind it. With Square, consumers will be trusting whoever they hand their card to be running current and legitimate software. How this is any different than me handing my card to an entirely random person at the corner store I don't understand, except this time I could write an alternate application to process the transaction myself without having to use a JTAG cable as I would for one of Verifone's closed systems.

edit: clarification of last paragraph.

Reply on news.ycombinator.com