5 ms·
Carrier IQ Speaks Out: We ignore extraneous user Data
- salimane 15y agoyou ignore extraneous user data?!??! what!?!?! why do you get it in the first place if you're only interested in a subset? bunch of lying rootkit developers
- derefr 15y agoA topical metaphor: sometimes, to get a toy, you have to buy the whole happy meal.
- hboon 15y agoBut here, you are forced to eat it too.
- ghshephard 15y agoFrom the article: "If during a support call a technician asks a customer to enter a short code, CIQ will be listening for it; when it’s entered, CIQ will relay the appropriate diagnostic information to the carrier. Any keystrokes beyond that are ignored." Their explanation then, for monitoring keystrokes, is that they are waiting for a diagnostic code to be entered by the user. Presumably this can be done from anywhere within the phone, on any platform, without regards to running applications.
- latch 15y agoHow do you capture a subset of data without capturing it all (and discarding what doesn't match)? If you are doing a find in a document, don't you scan through the entire document and discard whatever isn't what you want? It sounds like that's the explanation they are trying to give.
- salimane 15y agothe thing is, given the context, you can do this without looking at everything the user is doing, you can basically wait for the trigger (error, call drop...) to happen before trying to submit error reports. that's the way it's done in software applications
- latch 15y agoJust to be clear, I'm not saying they are right, I'm just trying to look at this objectively. If you read what they said, I don't think your solution works: "During a support call a technician asks a customer to enter a short code, CIQ will be listening for it" Without knowing a ton more, maybe there's a _much_ better way to do this..like having the user explicitly launch a program rather than having something always running and always listening. No argument there. But if you're thinking they should abstract away some "code_entered" event, ultimately something lower-level is going to be listening to each keypress and looking at sequences to raise those events. Some of this stuff happens at a lower level than a lot of us might be use to programming nowadays. I'm not sure that there's necessarily a pretty error event they can hook into. Something about all of this reminds me of using stuf like Spy++ on Windows and you see just how much raw data is available at the lowest levels
- 101001010101 15y agoIt would be interesting to see their agreements with the carriers. Presumably the carriers could request any data, not simply diagnostics. It certainly sounds like CIQ is prepared to deliver whatever a carrier might ask for: CIQ's rootkit can capture everything.
- tptacek 15y agoAt (grave) risk of being repetitive: if you don't trust your carrier with your personal information, you shouldn't be using Android phones on their network. CIQ or not, carriers have vastly more information about most people to work with than e.g. Google.
- 101001010101 15y agoAt least you have an agreement with your carrier. Alas, you have none with CIQ. Even though you're using their software.
- dedward 15y agoAll agreements and contracts aside, a legal court order to help with an investigation is unlikely to be ignored.
- 101001010101 15y agoAnd I'd guess they still charge a fee for that, too. :)
- flyt 15y agoThe key question for them is "Why can't users disable Carrier IQ on their device, and why does it run if the cellular antenna is disabled?"
- tptacek 15y agoTwo plausible answers: (a) Because the carriers don't want you to, if only because it would increase support costs to have to walk users through turning it back on, but probably mostly because normal people don't really care about the stuff CIQ is intended to do. (b) Because the feature for disabling CIQ with cell service disabled makes zero money for any party involved and therefore has slipped every MRD given to its engineering team.
- jimrandomh 15y ago> why does it run if the cellular antenna is disabled? That's easy; they can't reliably tell between disabled cellular service and broken cellular service, and the carriers really want to know about the latter. Well, they probably could figure out whether service is supposed to be disabled, but the mechanisms that control whether service is enabled are, themselves, one of the things that CarrierIQ would be used to detect problems with.
- georgieporgie 15y agoBased on what I've read, none of the investigations actually looked at what data was reported back. My money is on Carrier IQ providing a valuable service in a somewhat sub-optimal way with regards to privacy, and carriers probably doing a really shoddy job of integrating it into the phones they distribute.
- ghshephard 15y agoI don't think any claims were made that data was being reported back. The mere fact that this third-party utility is monitoring keystrokes, SMS data, Web URLs is what has everyone concerned. If it's discovered that the data is being stored _locally_ on the phone - that would turn out to be a scandal. I suspect laws would be broken if any of this data was actually being reported _off_ of the device.
- tptacek 15y agoThat seems like a really unfair argument. Discussion about CIQ has been rife with people talking about CIQ collecting and reporting keystrokes. The supposition has clearly been that CIQ or the carriers are centrally logging private data. Given that the carriers already have all your SMS data and phone usage history, I think you're probably wrong on the legal front, too.
- tptacek 15y agoI'm more or less convinced that you're right about this. CarrierIQ brought some of this on themselves by reacting terribly to a security researcher's claims regarding their software. Trevor Eckhart claimed CIQ was a "rootkit", and they freaked out and threatened to sue him. Naturally, everyone now assumes that CarrierIQ is evil, since suing security researchers is almost invariably an evil thing to do. People are paying less attention to the fact that CarrierIQ subsequently issued a formal apology to Trevor (from experience: companies stung by security researchers virtually never do this; they just back off quietly), and that nobody has really refuted CIQ's claims about what their software does. The "there's this evil big brother company nobody has ever heard of that a security researcher found out and got sued over" narrative is a lot more fun than the banal reality, so that's what we're all going to run with. It's also a little amusing to see people wigging out over the privacy of their SMS messages. Their SMS messages have no privacy. The carriers see and log all of them. Lord only knows what other tools they have on their network explicitly for the purpose of mining data out of those messages.
- jrockway 15y agoThis explanation is very interesting. I'd like to be able to share with my carrier when a call is dropped and when a web page doesn't load, but I don't want that being done automatically. Make a list for me, and I'll send it along to the carrier once a month after making sure there's nothing I want kept a secret there. Also, I should be getting paid to do this, because my job as a customer is not to make some big company's product better. It's a nice try to include software to do this in the background with no way to opt out, but with the cat out of the bag, the carriers are going to have to figure something else out. (I use Cyanogen Mod so I doubt this affects me at all. But it's annoying to discover that the reason behind closed bootloaders and firmwares is to spy on me behind my back without permission. Good thing they employ shitty developers to write that code!)
- djb_hackernews 15y agoThey are hiring an "Extreme Analytics Architect", to work on datasets measured in petabytes[1]. Would be interested to know what they consider extraneous. [1] http://www.carrieriq.com/company/careers.htm#ExtAnalyticsArch http://www.carrieriq.com/company/careers.htm#ExtAnalyticsArc...
- warmfuzzykitten 15y agoWell, of course, they only gather information requested by the carriers, and we all trust the carriers. What could go wrong?