>I hope we see a different CA for each ccTLD in the future.
Completely agree with this and IMO it's how the system should've operated from the very start. Unfortunately, I highly doubt that the CA cartel will let go of the power (and associated cash flow) voluntarily.
Browsers are the cartel. CAs have zero power against browsers, but browsers can poof CAs out of existence. And it's not really a cartel - it's a monopoly, it's Google and everyone else had better copy Google.
I have very little reason to believe that NSA is not doing, and had been doing that, more or less for as long as there had been CA.
And on a global scale.
If you don't find this plausible, it is because usually americans believe their predator state to be some kind of a "lion king" (aka superman, spiderman. etc), while it is more of a laughing hyena.
Russians would often fend this off by saying "the CIA major is farther than the FSB one".
But of course there's little reason to doubt that all public-facing separation between world's secret services is but a spectacle, just like the idependence of CAs.
Not only that, but also all encryption running in OSes that run above lower level, battery-powered SoCs with full network stack like Intel ME, AMD PSP and ARM TrustZone.
Yes, the banks can verify it for their own domains - unless Russia is sanctioned out of the CT logs or the government forces Yandex Browser not to check CT.
They can also just load the site from a separate internet connection and see if it has their certificate.
Nope, MitMing will be done by the SORM system [0] using certificates signed by the Ministry of Digital "Development" which will be trusted the Yandex Browser, which will be widely installed out of necessity by ordinary Russians to access banks and subsequently other Web resources.
Surely this will improve user security and trust in the existing Web PKI system in non-Western countries. /s
> Surely this will improve user security and trust in the existing Web PKI system in non-Western countries. /s
I recognize that Russia is making this change for MitM spying, but this particular sarcasm seems incorrect. Given Verisign's willingness to bend the knee to Texas courts, one could reasonably lose faith in US PKI issuers such as Digicert. I certainly hope the EU is studying the problem US-controlled EU-trusted issuers — the EU age verification systems hinge critically on them, much less the entire web. So I empathize with the sarcasm, but best not to offer MitM proponents (whether in Russia or the U.S. or elsewhere!) an argument that could be used against your viewpoint.
>I recognize that Russia is making this change for MitM spying
Nope, they do it primarily out of necessity, because of the mounting pressure on the previously used CAs. The MitM capability is just a nice side bonus.
>So I empathize with the sarcasm, but best not to offer MitM proponents (whether in Russia or the U.S. or elsewhere!) an argument that could be used against your viewpoint.
If browsers truly cared about user security they would've provided reasonable conditions for supporting national CAs:
- Limit its authority only to respective national domain zones.
- Mandate use of Certificate Transparency handled by an independent third party to prevent MitM.
But this debacle only shows that western-controlled (especially financial) systems can be and will be used as a pressure tool, so any large sovereign nation will not trust them as they would in the past. And the taken actions only contribute to further fragmentation of the Internet across national and block borders.
I’d be totally onboard with TLD-locking them to legal jurisdictions they’re comfortable being bound to, except that this would underserve a great deal of the Internet. Don’t really have a great solution yet, either. Perhaps as each TLD operates DNSSEC they could sign authorized issuers by publishing TLD CAA records, which would create some legal zone accountability that’s lacking today (and give the EU a lever by which to cut off U.S. registrars from their zones). But I have no idea how to effect any of that change, and Let’s Encrypt is truly screwed in this model as a worldwide entity. The endgame might actually be “to operate a domain registrar you must be a PKI”, which would ravage the segment and probably permanently kill off Namecheap (one can dream). So, yeah, I agree: I think instead we absolutely will see fragmentation, at both software (PKI) and, eventually, hardline levels, rather than see domain registrars and PKI issuers be forcibly merged by policy.
We should let each country specify trusted CAs the way they specify their DNS signing keys. Or we should just implement DANE already and then the same key serves both purposes and we can delete WebPKI from the world.
>Nope, they do it primarily out of necessity, because of the mounting pressure on the previously used CAs. The MitM capability is just a nice side bonus.
So the West essentially helps Kremlin to control Russian citizens. Why is that? Incompetence or something else?
>I recognize that Russia is making this change for MitM spying
Not really, banks do this. FSB would love to spy on everyone of course, but all was working fine until the CAs started revoking the certificates recently, directly aiding the FSB. From the article:
>The banks first moved to GlobalSign in 2022. This June, GlobalSign began revoking certificates held by sanctioned Russian companies, and they moved on to HARICA, the Greek academic authority.
>A month ago, HARICA refused to revoke: its issuance is self-service and domain-validated, so its certificates identify a domain and nothing else; it was not, it argued, “the competent authority to make these legal attributions.” However, on July 27, Greece’s eIDAS supervisory body appeared to confirm the disputed certificates had been revoked and referred the case to the national financial sanctions unit.
This is probably the future. Who is better able to verify an identity than a state? State run registrars regulate companies. States issue individuals ID documents. If you have trusted central parries issue encryption certificates it will gravitate to fewer and more centralised issuers.
Decentralised systems such as ssh are different, but I find it hard to imagine trusted central authorities ending up as anything other than states, or so heavily regulated as to be effectively state be controlled. This wax always the big flaw in SSL/TLS. In DNS too.
Downvotes but no counter arguments? DO people think I am wrong but cannot be bothered to explain why, or are people shooting the messenger, or have poor reading skills and interpret prediction as advocacy? Genuinely curious.
People on HN are resistant to government power, but fail to realize the US government having power over Russia (via CAs) is worse than the Russian government having power over Russia.
Even that goes beyond what I was saying, and its a matter of PoV - Americans would presumably prefer their government to have more power over other countries in at least some cases, and very likely with regard to Russia.
I am not saying its a good thing and I strongly implied that it is a bad thing (I referred to it as a "flaw"). Surely the anti-government power people on HN should agree with me? I am warning of an expansion of government power. They might not like it, but no one likes warnings.
Of course. Who else would be their CA? Some USA state-run CA? That's far too much political risk.
I hope we see a different CA for each ccTLD in the future.
>I hope we see a different CA for each ccTLD in the future.
Completely agree with this and IMO it's how the system should've operated from the very start. Unfortunately, I highly doubt that the CA cartel will let go of the power (and associated cash flow) voluntarily.
Don't think the cartel would go against a state
Browsers have to kowtow too ...
Browsers are the cartel. CAs have zero power against browsers, but browsers can poof CAs out of existence. And it's not really a cartel - it's a monopoly, it's Google and everyone else had better copy Google.
> The banks came back in disguise, repackaging their apps as coupon trackers
That’s a wild move by a bank. How is the place not overrun with scams?
If you scam Russian citizens, you are defenestrated. If you scam foreign citizens, they don't care.
Based on the timing, I assume this is a direct response to the US state directed revocation of certificates of Iran's Fars News Agency.
Might as well merge them all and call FSBank, for they will be sure as hell MITMing all the communication between clients and backends.
I have very little reason to believe that NSA is not doing, and had been doing that, more or less for as long as there had been CA. And on a global scale. If you don't find this plausible, it is because usually americans believe their predator state to be some kind of a "lion king" (aka superman, spiderman. etc), while it is more of a laughing hyena.
Russians would often fend this off by saying "the CIA major is farther than the FSB one".
But of course there's little reason to doubt that all public-facing separation between world's secret services is but a spectacle, just like the idependence of CAs.
Not only that, but also all encryption running in OSes that run above lower level, battery-powered SoCs with full network stack like Intel ME, AMD PSP and ARM TrustZone.
Oh, really? That's interesting.
But as they say, the chain is as strong as its weakest link.
You can verify this for your own domains by using certificate transparency.
Wouldn't that be true also for the Russian CA?
Yes, the banks can verify it for their own domains - unless Russia is sanctioned out of the CT logs or the government forces Yandex Browser not to check CT.
They can also just load the site from a separate internet connection and see if it has their certificate.
They don't need to. These are politically connected entities.
Nope, MitMing will be done by the SORM system [0] using certificates signed by the Ministry of Digital "Development" which will be trusted the Yandex Browser, which will be widely installed out of necessity by ordinary Russians to access banks and subsequently other Web resources.
Surely this will improve user security and trust in the existing Web PKI system in non-Western countries. /s
[0]: https://en.wikipedia.org/wiki/SORM
> Surely this will improve user security and trust in the existing Web PKI system in non-Western countries. /s
I recognize that Russia is making this change for MitM spying, but this particular sarcasm seems incorrect. Given Verisign's willingness to bend the knee to Texas courts, one could reasonably lose faith in US PKI issuers such as Digicert. I certainly hope the EU is studying the problem US-controlled EU-trusted issuers — the EU age verification systems hinge critically on them, much less the entire web. So I empathize with the sarcasm, but best not to offer MitM proponents (whether in Russia or the U.S. or elsewhere!) an argument that could be used against your viewpoint.
>I recognize that Russia is making this change for MitM spying
Nope, they do it primarily out of necessity, because of the mounting pressure on the previously used CAs. The MitM capability is just a nice side bonus.
>So I empathize with the sarcasm, but best not to offer MitM proponents (whether in Russia or the U.S. or elsewhere!) an argument that could be used against your viewpoint.
If browsers truly cared about user security they would've provided reasonable conditions for supporting national CAs:
- Limit its authority only to respective national domain zones.
- Mandate use of Certificate Transparency handled by an independent third party to prevent MitM.
But this debacle only shows that western-controlled (especially financial) systems can be and will be used as a pressure tool, so any large sovereign nation will not trust them as they would in the past. And the taken actions only contribute to further fragmentation of the Internet across national and block borders.
I’d be totally onboard with TLD-locking them to legal jurisdictions they’re comfortable being bound to, except that this would underserve a great deal of the Internet. Don’t really have a great solution yet, either. Perhaps as each TLD operates DNSSEC they could sign authorized issuers by publishing TLD CAA records, which would create some legal zone accountability that’s lacking today (and give the EU a lever by which to cut off U.S. registrars from their zones). But I have no idea how to effect any of that change, and Let’s Encrypt is truly screwed in this model as a worldwide entity. The endgame might actually be “to operate a domain registrar you must be a PKI”, which would ravage the segment and probably permanently kill off Namecheap (one can dream). So, yeah, I agree: I think instead we absolutely will see fragmentation, at both software (PKI) and, eventually, hardline levels, rather than see domain registrars and PKI issuers be forcibly merged by policy.
We should let each country specify trusted CAs the way they specify their DNS signing keys. Or we should just implement DANE already and then the same key serves both purposes and we can delete WebPKI from the world.
>Nope, they do it primarily out of necessity, because of the mounting pressure on the previously used CAs. The MitM capability is just a nice side bonus.
So the West essentially helps Kremlin to control Russian citizens. Why is that? Incompetence or something else?
Just some Nanay Boys wrestling.
https://www.youtube.com/watch?v=ztstTo3dVp4
Allowing Russia to have a TLD is also helping the Kremlin control Russian citizens. Do you recommend that IANA should delete the .ru domain?
How so?
>I recognize that Russia is making this change for MitM spying
Not really, banks do this. FSB would love to spy on everyone of course, but all was working fine until the CAs started revoking the certificates recently, directly aiding the FSB. From the article:
>The banks first moved to GlobalSign in 2022. This June, GlobalSign began revoking certificates held by sanctioned Russian companies, and they moved on to HARICA, the Greek academic authority.
>A month ago, HARICA refused to revoke: its issuance is self-service and domain-validated, so its certificates identify a domain and nothing else; it was not, it argued, “the competent authority to make these legal attributions.” However, on July 27, Greece’s eIDAS supervisory body appeared to confirm the disputed certificates had been revoked and referred the case to the national financial sanctions unit.
This is probably the future. Who is better able to verify an identity than a state? State run registrars regulate companies. States issue individuals ID documents. If you have trusted central parries issue encryption certificates it will gravitate to fewer and more centralised issuers.
Decentralised systems such as ssh are different, but I find it hard to imagine trusted central authorities ending up as anything other than states, or so heavily regulated as to be effectively state be controlled. This wax always the big flaw in SSL/TLS. In DNS too.
Downvotes but no counter arguments? DO people think I am wrong but cannot be bothered to explain why, or are people shooting the messenger, or have poor reading skills and interpret prediction as advocacy? Genuinely curious.
People on HN are resistant to government power, but fail to realize the US government having power over Russia (via CAs) is worse than the Russian government having power over Russia.
Even that goes beyond what I was saying, and its a matter of PoV - Americans would presumably prefer their government to have more power over other countries in at least some cases, and very likely with regard to Russia.
I am not saying its a good thing and I strongly implied that it is a bad thing (I referred to it as a "flaw"). Surely the anti-government power people on HN should agree with me? I am warning of an expansion of government power. They might not like it, but no one likes warnings.
Move related with wildberries attacks probably