Apple is currently embroiled in a legal tussle with the US Department of Justice over a defendant’s iPhone. The DoJ wants to force Apple to unlock the phone, but Apple argues that first, it technically cannot do so, and second, that it doesn’t have to. It’s that second point the DoJ is trying to address in a very interesting way.
Effectively, the DoJ is arguing that because Apple only licenses iOS to its users, and because Apple specifically states it retains ownership of iOS, Apple can be forced to unlock the phone. It is, namely, the DoJ argues, iOS which is thwarting law enforcement’s ability to do its job, and the copy of iOS on the phone in question is, as Apple itself clearly states, Apple’s property. Or, in the DoJ’s legalese:
Apple designed, manufactured, and sold the Target Phone that is the subject of the search warrant. But that is only the beginning of Apple’s relationship to the phone and to this matter. Apple wrote and owns the software that runs the phone, and this software is thwarting the execution of the warrant. Apple’s software licensing agreement specifies that iOS 7 software is “licensed, not sold” and that users are merely granted “a limited non-exclusive license to use the iOS Software.” […] Apple cannot reap the legal benefits of licensing its software in this manner and then later disclaim any ownership or obligation to assist law enforcement when that same software plays a critical role in thwarting execution of a search warrant.
This is a remarkable argument, and it fascinates me to no end – if the DoJ actually manages to pull this one off, it will not only be a glorious case of the insane anti-consumer implications of commercial software licensing coming to bite a large corporation in the ass, but it will also have far-reaching consequences for the power the US government has – in a very, very bad way, as Cory Doctorow explains:
To my knowledge, this is an entirely novel argument, but as I say, it has far-reaching consequences. Virtually every commercial software vendor licenses its products, rather than selling them. If the DoJ establishes the precedent that a product’s continued ownership interest in a product after it is sold obliges the company to act as agents of the state, this could ripple out to cars and pacemakers, voting machines and tea-kettles, thermostats and CCTVs and door locks and every other device with embedded software.
Commercial software licensing is a shady practice that should’ve come under intense scrutiny decades ago, but this is not the manner in which I want it to be done.
I can’t think of a more pyrrhic victory if the US DoJ manages to make such a legal theory stick. What an enormous can of worms they’re trying to open. At the same time, it would suddenly become an interesting time in the US for the software industry as one of their most cherished anti-consumer practices would become a very sharp double edged sword. I’m certainly not going to try to predict how that would play out.
Back in the 1980’s a piece of accounting software failed right before a tax audit. The user sued and a Judge (California IIRC) declared that the applicable law was LEASING for the same reason – the original company retained ownership. And you can’t lease something “as-is”. One of the reasons for the 55 pages is all the disclaiming (to be legal everywhere) any responsibility for damage, limiting liability, requiring arbitration.
If it was merely sold “as-is” and relied on the “first sale” doctrine or merely used copyright – you can’t sue a cookbook author (1st amendment if nothing else) because you didn’t like how the soup tastes. Is it a book or a toaster? You can’t sell a new toaster that burns your house down and disclaim liability. A book (or its author) that describes how to build a toaster can’t be held liable if you build it and it burns down your house. You can’t have it both ways.
But to go further, it is like iCloud. You don’t own anything on iCloud except your own data. It is Apples’s hardware, software, network, etc. Apple owns it.
Ownership has rights and responsibilities. Apple wants to keep ownership, they keep the liability.
If I have a local Raspberry Pi with opensource doing an encrypted TimeMachine, I own everything and Apple, even if I’m using their encryption doesn’t touch it.
Have to agree. Regardless of what you think about it that is certainly an interesting (and quite possibly valid) way of interpreting existing law…
I need popcorn…
Or maybe they’re just trying to force them to go open source. Because at that point, if this argument holds, they either need to open source their operating systems or kiss the vast majority of their foreign sales goodbye. Small government, pro-liberty, Republicans my ass. Civil liberty loving Democrats my ass.
Except that most open source software is also licensed. The only thing that’s not subject to a license is software in the public domain, where the author has explicitly renounced their interest.
I’ve always thought that other copyrighted works (eg. music, video etc) is also subject to a license, it’s just an implied license. Ownership of the underlying work was never transferred, only ownership of a physical object that contains a limited license. That license may not include things like broadcast or public performance, which would require an additional license.
Software makes this license explicit. But since the entire industry has grown up this way, it raises the question of what rights would be implied in the absence of an explicit license? Courts have never had to consider that question before. Presumably the rights would not extend to redistribution (prohibited by copyright) but software can’t function without some duplication as part of installation or execution. Even if courts were forced to consider this issue, the original argument may still be valid, that software ownership remains with the author and hence the author can be compelled to act with regard to their property.
This argument appears to be taking us down a very dangerous “nuclear option”-type place, where we blow up the legal underpinnings of software and reinvent them all over again. And just to be very clear, that doesn’t necessarily imply the result will be more consumer friendly. Most likely it would be solved in part by legislation, and legislation is seldom written on behalf of consumers.
But the legalese argument seems to heavily rely on ownership in the licence. As far as I know, no open source licence, not even GPL3, says anything about ownership. In the case of GPLx licences, everyone owns their copy of the software, in a sense it is “sold” and is “paid” for by making the modifications available to the people they pass the software on to.
Edited 2015-10-24 02:22 UTC
The GPL relies on the Copyright of the software to work and, effectively, the Copyright indicates who “owns” the software.
If you are the sole copyright owner, you could relicence it, for example, and sell it. (In the UK, at least, any software written in the line of work has the copyright automatically assigned to the company – which is the way that companies typically own the software and can licence it to people to allow them use of it)
I wonder if FSF software, owned by a single entity, complicates this…?
Yeah, and the GPLx licences effectively hands over ownership to whoever has the distribution, which is my point.
The FSF doesn’t own its software any more than the GPL allows them to.
Yes, it does. (With the small caveat that there are a lot of tarballs out there that the FSF can’t change out)
Just in the same way that Apple can insert new terms and conditions on it’s software, the FSF can relicence their code as being GPL 3.0 or later. (Dropping the terms of GPL 2)
It’s why libsdl can go from LGPL (libsdl 1.2) to zlib. (libsdl 2.0) Or why musl-libc could swap to a permissive MIT license. Or why MAME could swap to GPL.
In all cases, all of these Copyright holders have agreed on a change of licence.
The GPL hangs off Copyright and Copyright is how you own code.
It uses copyright to subvert it.
I would argue that they subvert it far less than some of the big companies have in the past. (but, mostly, barring the way laws like the DCMA and EUCD are implemented, those are individual points rather than being prolonged)
You’re discussing something else entirely as far as I’m concerned. By subvert, I’m referring to the original purpose of the concept of copyright of limiting people’s rights to copy. The GPL subverts copyright by expressly telling someone they have all the rights the original owner has. In that sense, those big companies are not subverting but are carrying on the tradition of copyright by further limiting people’s rights.
Errrr… No.
I, as a user of GPL software, do not have the same rights as the original copyright holder(s). (As a whole) I cannot relicence the code. It would be pointless if I could.
As a user, I have the licence to use it in specifically the way the specific GPL states. The owner can turn around and grant additional rights to other people using the code. It’s how, for example, openRTOS and safeRTOS, which are based on the (modified) GPL’d FreeRTOS, can be sold without being GPL themselves.
On the other hand, some of the big companies have both gone after people who are abusing their copyright – fair enough – but have also ignored other people’s copyright sometimes using the standard technique of “We can afford to put many magnitudes of more expensive lawyer time into this lawsuit than you can and then watch you go bankrupt”.
DMCA takedowns have been used on people which they can’t ignore – and yet the big firms just seem to shrug them off.
(A recent case involving a company sending a takedown notice to somebody with a youtube account who just happened to licence the video in question to the company in question being one example – although somebody sensible at the company happened and it went away)
However, to be fair, unlike your typical EULA, the GPL states that, if you receive something under the GPL and comply with its terms, the copyright holder cannot interfere with your right to use it under the license you received it with.
(eg. If you’re in compliance with the license, nobody has the power to revoke it and organizations like the FSF have no power to sunset old GPL2 copies of code they upgraded to “GPL3 or later only”)
Edited 2015-10-28 02:09 UTC
Errrr… I was obviously simplifying things a bit because I was explaining what I was using the word “subvert” to mean.
The government’s argument is based on ownership of the final product. In the first sale doctrine in the US even if an item is under copyright you still own the product you purchased. You own the physical book, the CD, etc and there are no restrictions on what you can do with them outside of copyright law. That means the original seller can’t legally tell you you can’t use it to build a bonfire, use a CD as a frisbee, etc.
For years proprietary software houses have been doing end runs around the various doctrines that limit control over their product by US copyright law and first sale by selling a license to use the product which is theoretically covered by the UCC (Uniform Commercial Code) and contract law rather than exclusively US copyright law. What that means is that copy of OS-X, iOS, etc on your computer isn’t your property. VIA a EULA which you theoretically agreed to making it a contract under the UCC, the software binaries on the device still belong to Apple (or whatever company sold the licenses). The idea was to limit the ability of people to create compatible third party software, perform benchmarks that may place their products in a competitive disadvantage, limit the ability to resale, and a host of other frowned upon activities.
There are few, if any, open source licenses that retain product ownership over the software being distributed. In effect, the contract here only stipulates what you can do with the source code and says nothing about giving up rights to physical ownership conveyed by first sale doctrine. The US DoJ is only claiming that since Apple still owns the software on the devices its sold under the UCC contract law, it can be forced as the owner to give the DoJ access to those devices regardless of the rights of the defendant because the defendant never owned the software used to encrypt the data on the device to begin with. Since open source licenses like the GPL, BSD, etc make no claims in retaining end product ownership, the theory wouldn’t likely apply.
It’s a novel legal theory, and I’d not at all care to place bets on which way a judge would rule.
Well the thing is that Apple doesn’t really own their own software: most of it is licensed from the University of Berkeley or Carnegie Mellon University.
It will be pathetic to see DoJ demanding OpenSSL to decrypt what their licensees have encrypted but … go ahead DoJ!
No, they are not trying to destroy it, but they are trying to possess it.
If you look at the US economy, most of it prints money for export to other countries for use in commerce.
I would put that as 78% of our economic output, which is rapidly collapsing by the way if you now understand NIRP is almost here.
(Imagine depositing your payroll check, and $200 goes to the bank for the priviledge of direct deposit….that IS coming.)
Don’t even think about not doing direct deposit as these criminals certainly don’t want you to get away so easily.
Cash is BANNED.
If you understand the above implications, ANY cryptographic entity must be owned by the government.
OR
It cannot be permitted to exist and the individuals must be detained or destroyed. (That includes whole countries as well…so this is going to be a very very bad time we are entering into.)
Either by force, or by coercion this is nothing but complete slavery for every individual who buys or sells using the Federal Reserve note.
Obviously this just doesn’t apply to individuals, it also applies to countries as well.
Some of the countries who refuse to use or engage in commerce using Federal Reserve notes are the current darlings of the military industrial complex built up around the preservation of this idea.
But the idea is simple, digitize everything for complete control and enslavement for complete power.
Encryption that isn’t owned by such an authority bent on this madness is obviously going to target such a threat to its power.
I understand that Apple owns the rights to the software on the phone, but why would that lead to the conclusion that they are responsible for providing access to the user’s personal data?
Apple are being asked to provide the keys to decrypt the device. They are not being asked to provide the data (which is held on the device).
and if there are no keys to provide? (as it should be the case if encryption is properly done)
Apple want to be able to decrypt their customers device if they forget their password. That is why they store the keys.
My understanding is, that even Apple can’t decrypt the data because the key is encrypted with the user’s passphrase.
Apple doesn’t have the key. Apple cannot compel the user to disclose the key any more than the government can and that is only getting harder. Precedent now says that you have 5th amendment rights to not unlock your phone for law enforcement.
I think in simple terms you’re probably right. Apple might not have all users Private keys stored on their servers – But, they (often) have full access to users devices..
(certainly non-rooted iDevices ; PLUS arguably the majority of Mac and rooted iDevices that access Apple Services which haven’t been carefully locked down and secured by their device owners against the devices “software owners”)
And while Apple maintain to be the legal owners of said OS/firmware Softwares – then the US DOJ in this case is using a (probably logically sound, but certainly morally bankrupt argument) that they can (again, given their own dubious legal justifications) REQUIRE Apple to hand over the keys to the kingdom or at least the keys their Customer’s kingdoms..
Publicly I don’t think Apple can condone this from happening, rightly fearing a backlash, so that alone will probably force them to do the right thing by Customers and stand up for this legal Bull****
plus the house of License cards would come crashing down, and the mega corps down want to cede all control to the govs any more than us individuals do. But they have real power (chance)
interesting the DOJ had the balls to go for this approach though. They must have unseen leverage (more than I would have expected). High-level blackmail cards perhaps….. /Conspiracy Theory B
“And while Apple maintain to be the legal owners of said OS/firmware Softwares – then the US DOJ in this case is using a (probably logically sound, but certainly morally bankrupt argument)”
Then as the owner of the hardware I can refuse permission for the software I don’t own to unlock or provide anything I don’t want it to.
I have a Windows phone. MS clearly state that encryption is only for privacy if my phone is lost or stolen. They can decrypt the device if I forget the passphrase.
My Linux laptop and Android phone are also encrypted. Only I know the passphrase(s). If I forget the passphrase(s) there is no way to decrypt my devices.
I’m pretty sure iOS follows the Windows Phone protocols ie Apple can decrypt the device remotely.
With US Gov leading the charge against privacy, against pretty much all consumer oriented data storage and comms (outside of banking)..
they’re trying every avenue to force the companies holding large silos of data (FB,MSFT,GOOG,AAPL)
to basically pass them whatever they want whenever they want it..! (well, maybe not quite yet – but it’s sure going that way)
– the “legal” yet shameful frameworked path with CISA (which could yet be fought) alongside the underhand under-the-radar path with all the myriad Snowden exposed NSA/GCHQ “tools” (new versions of which will ALWAYS exist)
And now here it looks like they are demanding device unlocking/encryption keys on their want also (via this supremely morally deluded premise being taken here)
Sadly (IMO) even if the device manufacturers architected their devices to be more privacy conscious (and required customer to procure their own SSL certs, generate their own encryption keys (and not even allow themselves access)
—BigGov will go after the certificate providers and will strong-arm them too
On the Secure website side of things:
In future I predict Certificates chains of trust will have to be distributed and work/compute-based (blockchain like or similar) / I mean rather than having “trusted” Root CA’s.
On the p2p/individuals Private communications front:
I think we’re just about nearing the point where every individual (with the help of technologically literate friends probably) and a huge dose of yet to be developed wonderfully easy-ish-to-use tools from the OS and hacker communities – will be responsible for their OWN online security and privacy – because I don’t think anyone else will in the future [not REALLY – and that’s the point]
I only think banking gets (and will always get) a special pass as:
(a) it’s not providing communications facilities or data storage outside of “numbers” and (b) those “numbers” are universally important to everyone including all the powers that be!
Because any CA can issue certificates for any domainname.
They already wanted something more secure.
So http://www.certificate-transparency.org/ was created, it uses some of the blockchain technology.
This records which CA issues which certificate for which domain. This means you can monitor when a CA issues a certificate for a domain they shouldn’t issue a certificate for.
It’s supported in Chrome:
http://www.certificate-transparency.org/certificate-transparency-in…
Firefox developers are working on it:
https://bugzilla.mozilla.org/show_bug.cgi?id=944175
CAs are already required to do this for their Extended-validation certificates (greenbar).
Lennie,
I hadn’t heard of this, very informative!
It’s a very serious problem, I consider the CA trust model to be completely broken, theoretically as well as practically.
Hmm, I get why they do this. But it absolutely contradicts the “No trusted third parties” and opens up new avenues for abuse. They should explicitly disclose these drawbacks on the website.
Anyways, this is very neat, thanks for sharing!
Browsers and standards aren’t standing still at all.
Here a list of protocol updates in recent years from the top of my head.
In reverse order:
– Certificate Transparency
– HTTP Public Key Pinning (makes the CA-system ‘irrelevant’ for frequent visitors on a website)
– OCSP stapeling
– Firefox adds OneCRL, a Sub-CA blacklist
– HSTS
– old but people have been deploying it more: Forward Secrecy
Crypto:
– SHA1-certs are going to be blocked next year or the year after.
– no more SSL (blocked by browsers)
– less use of TLS/1.0
– less use of TLS/1.1
– more use of TLS/1.2
– development of TLS/1.3 is ongoing at the IETF based on some of the ideas from: Google’s QUIC
– better understanding of DH parameters and updates of webservers.
Not to forget support for HTTP/2 for better performance.
Thanks Lennie and Alfman both,
some useful info there. I think I like what the Certificate-Transparency.org effort is doing on the realworld level – I wasn’t up to speed on that at all.
But it still looks rather like a half-way house to an ideal system
i.e. looking to add a reasonably transparent and auditable monitoring layer to things — rather than actually re-architecting the whole chain to be (near)fool-proof. And non-hierarchical
Anyway – that’s probably a job for rather further into the future isn’t it
Probably something that’ll come about in concert with a technology like Namecoin reaching critical mass in response to overreach by people trying to make DNS registrars responsible for copyright enforcement.
Edited 2015-10-25 19:11 UTC
mistersoft,
Haha, yea that’s the way I feel about a lot of things. Certainly there’s lots of friction between legacy standards and modern requirements. Sometimes a clean break away is good, but often breaking compatibility is not an option, which is why we end up in these spots…
Lennie,
To be honest, I’m not impressed with our progress thus far because these problems are 15 years old. Most corporations involved have been pushing solutions designed to fit their business interests rather than the public good. Even mozilla is guilty of making self signed certificates a pain to use legitimately by treating them worse than plan HTTP. I feel strongly that every website should be entitled to deploy HTTPS without going to 3rd parties. Discouraging the use of self signed certificates is a big culprit in why we haven’t made big gains in moving towards HTTPS by default.
Solutions like Dane in combination with some other solutions too would be great, however they’re absolutely pointless if our browsers won’t support them. Looking at what to do in the future is well and good, but it’s just inexcusable that each and every connection isn’t already being encrypted out of the box today by default.
I agree that technology can solve most of our problems, but I think we have a long ways to go before overcoming the bureaucratic tendencies that got us here to begin with. And I think it’s justified to question models that play into the hands of those most responsible for this mess (ie certificate authorities) rather than shedding them.
Sorry about the rant Lennie, I’m with you in supporting rollouts to fix these problems. As with IPv6, I’m just tired of waiting for something that should have arrived years ago.
IPv6 usage (comparison to IPv4) is growing similar fast now than the growth of the IPv4 internet:
https://www.google.com/intl/en/ipv6/statistics.html
http://www.cidr-report.org/cgi-bin/plota?file=%2fvar%2fdata…
If you can think you have a way to deploy DANE which works everywhere without causing more problems than it solves I’d love to know.
Because many browsers in the field don’t have reliable access to DNS-servers which don’t break DNSSEC-queries. And nobody wants to host a DNSSEC-fallback server. And anycast doesn’t work either, we’ve tried that with for example IPv6 transition mechanism. It was a total performance mess.
DNSSEC/DANE is actually being used for securing e-mail more and more though: https://www.ietf.org/mail-archive/web/dane/current/msg06294.html
I’ve actually been thinking about for example DNSSEC/DANE and similar problems/solutions for browsers from time to time for many years, even contacting some of the people who actually work on these kinds of things with some of my ideas.
So the real problem is: finding solutions which work in the field without breaking things. This is no easy task.
Think of it like replacing the engines of an airplane while it’s in flight (or actually multiple types of plains).
Take for example this solution, this solves the broken DNS-servers which break DNSSEC-queries:
https://www.imperialviolet.org/2011/06/16/dnssecchrome.html
This isn’t compatible with the browsers which use the CA-system. There are browsers that won’t support a certificate chain which includes a DNSSEC-stapled certificate and a CA-certificate in the same HTTPS-connection. Some old browsers (think like IE) break if the DNSSEC-stapled certificate is inserted at the start of the chain and other old browsers break if they are appended at the end of the chain.
But this is better than self-signed certificates. But the people who like self-signed certificates are probably the same people who dislike DNSSEC. Because there was no adoption.
DNSSEC still depends on third parties and not everyone trusts those. The DNSSEC root signing keys are in the US and indirectly ‘controlled’ by the US government, they don’t like that.
If you want something else, you’ll need to download a full blockchain on your device running the browser. That is probably not a general practical solution either.
I really don’t think the problem is the vendors thinking about their own agendas. Even worse, vendors try to solve things and what they get is hate.
Have you read the (let’s call them uninformed) comments on this blog ?
https://blog.mozilla.org/security/2015/04/30/deprecating-non-secure-…
Edited 2015-10-25 21:11 UTC
No, I confess, there are nearly 300 posts there. However I will say that I strongly disagree with the OP regarding discontinuing HTTP at a specific date, that would be terrible and likely result in cases where we can no longer connect to our network appliances. Rather than making HTTP less accessible, We need to make HTTPS more accessible. Ideally that means removing some of the difficulties in using websites with custom CAs/self signed certs.
I mentioned one thing above and Let’s Encrypt. Mozilla is one of the main founders and organizations working on Let’s Encrypt.
Maybe I’m wrong, but I think Mozilla is already are doing a lot of things to improve that situation.
Firefox and Chrome actually have the best HTTPS protocol feature support (OCSP stapeling, HSTS, HPKP, etc.).
Lennie,
I know. Ultimately if too few are going to support it, then it may never reach critical mass. I think it’s sad, but it is what it is.
Edited 2015-10-26 06:29 UTC
Edited 2015-10-26 10:09 UTC
Lennie,
As I indicated earlier, I support this development, but it just sucks that browsers are not building the foundations to get out of the CA dependency rut we’re stuck in. I really hope I’m wrong and they’ll change their mind, but every year that slips by is a lost opportunity for self-sufficient security for the following 5-10years.
I really find it difficult to comprehend the opposition to having crypto that doesn’t depend on 3rd party CAs. Of course those who are involved with selling CA certs will staunchly oppose it. I think browsers are paid by CAs for inclusion too, so that might be a factor. Not to insinuate anything, but do you or your company have any business relationship with CAs at all other than as a user?
Edited 2015-10-26 21:39 UTC
(sorry, didn’t answer all your questions, I didn’t have as much time)
No, they don’t pay browser vendors to be included.
They pay crappy companies like https://en.wikipedia.org/wiki/Big_Four_%28audit_firms%29 to do an audit. Lots of money and they aren’t just crappy companies, they don’t really look at the technical part either.
The browser builders look at the audit documents and add the CA-cert to the browser.
The browsers builders make most of the rules the CA’s have to abide by. The browser builders have all the power in the ‘CA/Browser forum’. They regularly add extra rules. Like no more SHA1 certs.
I work at a provider for businesses to host their websites and other related stuff.
A small part of what we do is we ‘sell’ the SSL-certificates (well, basically the hosting of HTTPS-websites) probably at about 7 times as much as the what it costs us to get them from the CA.
But the cert isn’t the only costs we have: when you get a cert, someone in these companies needs to click a link. This is probably the hardest part. Finding the right person and setting up the right email address and getting the right person to click the confirmation button at the CA to validate they own the domain.
You can’t imagine how often we need to get in contact with the customer to get these things done. And how far in advance you need to start contacting them for the renewals.
Obviously, because of crappy old browsers we also give them a dedicated IP-address and configure the server.
So really, we don’t make any money on these certs.
They are just an annoyance.
When Let’s Encrypt has proven it self, we’ll be using those for sure (any other CA that will adopt the same software might also be useful).
Which probably means: we’ll be payed less. But we won’t have any work with it anymore either. Let’s Encrypt is fully automated.
So, no we are not invested in the CA system. At all.
But if it was just that all you need is to trust one CA it wouldn’t be so bad. The real problem is: you need to trust all the CAs to not make a mistake.
Lennie,
Agreed, this is what we need to escape. Vendors who aren’t adding support for no-CA crypto are holding us back from adopting alternatives.
Not only can DNSSEC allow end users to validate domains at the source without any indirection through CAs, but it could also be used to reject any fraudulent EV-Certificates by parties who don’t control the domain. So, not only would the CA certificate attest that the company is who they say they are, but DNSSEC/DANE would bidirectionally attest that a CA certificate is approved by the domain owner. This way rogue CAs could sign fraudulent certifiates all day long, but it wouldn’t matter because they don’t control the domain. Now that would be real crypto security!
Anyways, since we’re at a basic level of agreement, does this sound like a happy place to end?
I don’t know why you mention DNSSEC/DANE again.
At least the developers of Chrome and Firefox agree that at this time it can’t be deployed. Because if you want to do checks and actually act on it, DNSSEC has to be always available and it isn’t on all networks.
And no browser vendor, not even for their own browser wants to run a DNSSEC-fallback service (like DNSSEC over HTTP when normal DNSSEC over DNS fails) and enable it by default: https://www.imperialviolet.org/2011/06/16/dnssecchrome.html
You seem to under estimate that impact of that ?
Let me be more clear: the imperialviolet blog is of Adam Langley, he is basically (one of ?) the head(s) of browser and webserver crypto at Google. That code for DNSSEC-stapled certificates was in Chrome, enabled by default. It was part of Chrome, it was removed because you couldn’t make it a working standard which was backward compatible (so both cert chains in the same HTTPS-connection). And that was the best bet we had for getting DNSSEC/DANE working in the browser. Because it doesn’t depend on the network giving you non-broken DNSSEC (as I explained before, you don’t need DNSSEC-enabled DNS servers in the network, just DNS-servers and firewall that don’t BLOCK DNSSEC or large reponses, etc.).
DNSSEC/DANE is something for the long run, very long run. Not any time soon.
Trust me, they really wanted to use and deploy this they just haven’t found a way that works.
Have a look at what the experts wanted to do:
https://www.youtube.com/watch?v=kXvZZ-t_WYA#t=45m20s
I don’t think you understand how they, really, really, wanted to do deploy this. Listen to all the reasons why you’d want it. He makes it very clear in the talk. They tried and failed. 🙁
Anyway, as I explained before I think the browser vendors are actually do some very useful work these days.
They already have a mechanism in place for making sure CA’s can’t arbitrarily issue certificates for domains they shouldn’t. Certificate Transparency. It’s deployed right now for EV certs. All EV-cert issuing CAs are participating and Chrome and soon Firefox support the protocols. I believe it’s even required by the CA/Browser forum. Even if the rule did not exists, I’m sure they’ll add it soon for EV because Chrome requires it.
Second, for DV (domain validated) certs, Let’s Encrypt will be available, they still think: this year.
The certs are valid for 3 months, it will include features like:
– OCSP-stapling (fast certificate revoke and fast website loading)
– probably with SNI (so no more IE on Windows XP), just like CloudFlare already does for a large number of sites for free. Maybe non-SNI is still an option.
– OCSP must-staple is enabled
All the major browsers have a mechanism in place to blacklist subCAs (comprised CAs). And OCSP-stapeling allows for quick revoke.
When Let’s Encrypt has proven itself, I think the other CA’s will adopt the ACME protocol (the automated system Let’s Encrypt developed) in at least some way, even if to just bootstrap their normal process. The software used is already open source and available.
Lots of changes like HTTP/2 will ‘force’ more websites to adopt HTTPS. W3C is making certain browser features like webcam to be HTTPS only.
Let’s be very clear, the goal really is: move all HTTP traffic to HTTPS.
IETF and W3C have said they want all their protocols to be encrypted for privacy and security and against mass surveillance. For example the DNS working group is working on adding transport encryption and privacy to DNS. Don’t know if they’ll succeed, but they have all kinds of proposals. I think they’ll wait for the TLS/1.3 specification is ready.
Edited 2015-10-28 09:27 UTC
Lennie,
My agenda for self sufficient encryption is pretty straightforward, but to be honest I still haven’t figured out your agenda against it. You may say you care but it’s contradictory when you reject the need to deploy the very frameworks that need to be in place to get there.
I’ll make this easy. If you can demonstrate any way to enable encryption in the major browsers today without reliance on a third party CAs and without forcing users to take manual action like importing certificates or installing plugins, then I’ll take back my criticism against the vendors. However if you can’t demonstrate that, then you acknowledge that my points are valid. Fair enough?
The only thing we have now is: you can bootstrap trust using the CA system which after that does not rely on the CA system anymore. That is the HTTP Public Key Pinning (HPKP): https://developer.mozilla.org/en-US/docs/Web/Security/Public_Key_Pin…
It works for ‘frequent visitors’ on your site (like ones every couple of months).
It’s a ‘trust on first use’, similar to SSH. But instead of asking the user: do you want to accept this key (and asking again when the key changes). The CA is clamming the first key is the correct one. A key change after that is fully in control of the site owner.
All the other changes are: improving the crappy CA system to be less crappy. And more accessible for the common (wo)man. So we can have more HTTPS. Aka good thing, it really does help against those passive and active attackers.
Also if you use Firefox and you like self-signed, you can do HTTPS with opportunistic encryption to fight passive attackers.
I’m happy we got more/better. You want a lot more, I’m saying in the short term it’s not so realistic.
___
Some of the people in the DNSSEC-root-keys-in-the-US-is-bad camp might like that things are changing:
http://www.internetsociety.org/what-we-do/internet-issues/internet-…
Some think it might be a bad thing.
___
There are lots of efforts like .bit and namecoin and so on trying to build a decentralized system.
But no browser vendor has gotten behind that, again because it doesn’t look like, at this time, it would fit the general public.
It’s like Bitcoin, you can go around all the laws and try to build anonymity, but does that create a system the general public wants ?
Edited 2015-10-28 16:28 UTC
Lennie,
This merely detects when the keys change. It does absolutely nothing to tell us we have the right key to begin with, it’s still dependent on CAs. This is the same problem exposed by SSH warnings that ask you to manually verify the server’s key fingerprints because the client doesn’t know if the key is legitimate. HPKP does not solve the bootstrapping problem whereas DNSSEC does!
I extend the challenge again, if you can demonstrate that websites can use this to encrypt traffic in mainstream browsers without user intervention and without third party certificates, then I’ll take back my criticism of vendors, but otherwise concede that my points are valid.
I don’t know why you set these challenges. 🙂
Maybe I should make it short and simple:
I would love to see something better adopted by browsers, I’ve actually given it a lot of thought how it could be done. But none of the ideas I had or many other people actually worked.
So realistically, I don’t see it happening right now.
It could still happen in the future.
Ah, I didn’t read this before posting. While there are some difficulties, I think they are being greatly exaggerated and the biggest obstacle is actually in the politics of adding support. Anyways, while it’s for a different reason, I don’t disagree with your conclusion – it is probably not going to happen.
Edited 2015-10-28 19:09 UTC
Anyway, a completely different discussion, but my worries are actually at higher layers of the stack.
It’s people being tied to platforms and going back from more standards to more platform building to lock people in. Without proper standards there might be adoptions, but also lock in on a grand scale.
It’s the IoT data, it’s the cloud computing, the US laws, the TPP, TTIP and so on:
http://techcrunch.com/2015/10/24/why-iot-security-is-so-critical/
https://media.ccc.de/v/31c3_-_6195_-_en_-_saal_g_-_201412272145_-_th…
https://media.ccc.de/v/27c3-4263-en-resisting_excessive_government_s…
https://www.youtube.com/watch?v=QOEMv0S8AcA
What I also think is interesting from a purely technical side is what cryptocurrencies like Bitcoin are doing with cryptography and distributed systems, but I don’t think we can build a society on that. Unless we can build something else based on that which isn’t pure anarchy. Because that is what some people seem to want.
FSF says this is easy to fix if DoJ wins the argument: just sell copies, not licenses.
A customer wants to buy your software? Produce a download copy (technically straightforward) and sell that copy. No license is sold. To hold water, the copy must be produced on demand instead of just a download link (which I find puzzling, since the act of downloading produces a copy, but legal stuff is always puzzling to me)
Clever.
Simplest solution: Sell a “single copy” of the software. Actually sell it, but don’t allow customers to duplicate it because it is copyrighted. Just like the design for a pair of scissors might be.
Edit: ninja’d.
Edited 2015-10-24 21:32 UTC
The U.S. government is simply throwing a tantrum because it can’t get at data it has no right to have. This legal case isn’t going to go very far because Apple has already made it clear that it literally cannot decrypt devices with iOS 8, 9 or greater and it has only limited access to data on devices running iOS 7 and below. That pretty leaves few devices for the government to try and compel Apple to decrypt. Meanwhile, newer versions of Android are decrypting data by default too which means the number of Android devices that the government could access is shrinking as people upgrade their handsets.
People are delusional if they don’t believe that the U.S. and other governments are combing through their data in the so called name of national security. With back doors built into Windows and OSX this is like taking candy from a baby. If people are smart they will immediately switch to open source software such as Linux where Linus Torvalds has already come on record as refusing to build a back door into the Linux operating system. Unfortunately most people today don’t care about their own privacy. Most people I know simply state that they have nothing to hide so who cares who sees their data. This is dangerous thinking on many, many levels.
Edited 2015-10-25 17:59 UTC
cmost,
Sometimes there’s a disconnect between what companies will claim, and what is actually true. Apple has told authorities it could not wiretap imessage to comply with court orders, even though security researchers ultimately proved that they could if they chose to.
http://www.nytimes.com/2015/09/08/us/politics/apple-and-other-tech-…
http://www.osnews.com/comments/28837
http://www.imore.com/imessage-secure-good-bad-and-complicated
While imessage may not be exactly the same as what we’re talking about in the article, it is nevertheless very important to have independent confirmation of claims. Apple has lied (or at least been mistaken) about it’s capabilities before. Also, if the NSA were to “assist” apple, then suddenly it might become much more capable of complying with court orders. When it comes to security, things are rarely as simple as we like to make them out to be.
Edited 2015-10-25 20:09 UTC
I don’t care what key sizes you use or prime number you select.
Commercial, closed source products are required to have back doors in them by all western governments that I can see.
If you think not, try this little experiment once:
1) Grab a static IP off the internet.
2) Setup a TOR node.
3) Allow it for public use.
If you are still alive after 30 days of operation, I doubt you will want to continue to be should you live past that time line.
All of this discussion about key sizes, libraries is a waste of time.
So the DoJ tries to convince some court, that Apple owns the content created with an iOS device, just like Adobe owns all content created with Photoshop? Everyone believing this reasoning to be legit better turn down their license to practice law. This reasoning has be rejected in court decades ago.
Edited 2015-10-25 22:47 UTC
Thinking more about this, I am not a lawyer, but the iOS is not “thwarting the execution of a search warrant” because for it to be actively doing so, it implies that there is a way for that “thwarting” to be stopped. Since that is not the case, and the data is scrambled already, it is a moot point.
Except for that nasty Supreme Court ruling that stated that the consumer does in fact own the product, but that the license only restricts certain activities, namely infringement.
This is just good PR for Apple, it appears as the good guy against bad cop. It seems to me like a fabricated argument to let people say around the world “see it? american companies thought to pass secrets to NSA went to the good side”
It’s not, the only guarantees they can’t really crack into an iPhone is to open source (not necessarily give it for free) their code, so independent analysis can be done.