4 ms·

Now this seems like something that StartSSL would want to be in front of rather than risk everything by having compromised certs out there with their name on th

Now this seems like something that StartSSL would want to be in front of rather than risk everything by having compromised certs out there with their name on them. I would think that kind of behavior would put their CA status at risk...

makomk 13y ago
They are in front of it, they've intentionally decided to continue charging for revoking these certificates.
- sigzero 13y ago
  Which they are 100% in the right to do by the way. StartSSL is a business.
  - MereInterest 13y ago
    And then people are 100% in the right to no longer trust StartSSL certificates.
    - wlesieutre 13y ago
      Or at the very least, StartSSL certs issued before yesterday.
      - aroch 13y ago
        No, any StartSSL cert. There are still unpatched systems and there likely will be for quite some time. Just because there's a CVE doesn't mean everyone updated
        wlesieutre 13y ago
        Sure, and there are also StartSSL certs used on systems that don't use OpenSSL. Do we just nuke every certificate, destroy their business, and force people with secure computers to buy new ones elsewhere even though their servers weren't affected by heartbleed? Because that option sucks too.
        claudius 13y ago
        As a user, I much prefer that my browser and my OS never ever again show a StartSSL-signed cert as valid over even just one compromised cert being displayed with a fancy lock. How StartSSL is going to achieve that, I don’t care, but neither do I care whether or not they go out of business over this.
    - shimon_e 13y ago
      Who is suppose to pay for staff time require to manually maintain the revoke list? BTW, revoking a cert doesn't stop it from being used in the wild as plenty of clients will still see it a valid cert. We are basically screwed until all current certs expire.
      - dpark 13y ago
        Why would the revoke list need manual maintenance? Are they running a root CA on pen and paper?
    - toomuchtodo 13y ago
      Who am I going to trust less? StartSSL certs? Or OpenSSL? I'm not saying I don't appreciate the hard work the two OpenSSL developers put into the code. But its incredulous that we're upset a business would give away a service due to an event completely out of its control.
      - pra 13y ago
        As it was out of the customers' control as well. Or should we have audited OpenSSL?
        toomuchtodo 13y ago
        I would argue that something that is the underlying component of e-commerce, privacy, and security for the majority of Internet activity should probably be audited.
    - latortuga 13y ago
      This is utterly ridiculous. These people signed up for a free SSL certificate knowing that revocations would cost them money. Enforcing a contract that both parties willingly agreed to does not make them untrustworthy, it demonstrates that their customers carry immense senses of entitlement. There are tons of SSL certificate providers out there. Their business model was not a secret when folks signed up.
      - sparkie 13y ago
        The business model an practices of the company are completely irrelevant to the people concerned in this case though - the users of the web browser. They never signed up for shit, and they have no incentive to trust a business which refuses to invoke compromised certs because people are unwilling to pay. Yes, this doesn't make them untrustworthy as a business, since their business model was known - it does make them untrustworthy as a party which is meant to be securing the web though - because they'll be failing at it spectacularly. Again: I never signed up to shit, and their business model means shit to me too - I just want some degree of security when browsing.
      - MereInterest 13y ago
        I am not their customer. At most, I am their customers' customer. Why do I have any obligation toward StartSSL? If StartSSL is free to choose a business model, then others are free to make judgment based on that business model. Should the customers of StartSSL pay the revocation fee? Yes. Should StartSSL be required to give free revocations? No. Are either of those at all relevant to whether or not I should trust StartSSL certificates, given that they are less likely to be revoked when they need to be revoked?
    - lern_too_spel 13y ago
      The other responders have no idea what you're talking about. You're right, of course, but I suspect most other CA's certificates should also not be trusted even though they don't have this revocation hurdle because a lot of these certificates were given to unsophisticated users who aren't going to revoke compromised certificates anyway.
    - bsder 13y ago
      By that logic, we should remove every CA. At least StartSSL will have good certificates for everybody in their free tier in 12 months. Can any of the other CA's make that claim? So, are you going to lead the charge at the rest of the CA's? And, I suspect that StartSSL couldn't possibly handle every single person calling them up to revoke a certificate right now. And that is NOT their fault--they didn't cause this bug.
    - epsylon 13y ago
      People would better be advised not to trust admins who can't shed 20 bucks for their users' security. Most likely these are the same admins to whom you certainly wouldn't want to give an email, password or any sort of confidential information.
revelation 13y ago
Their CA status is not at risk, was never at risk and will not be at risk in the future. Mozilla has made their stance clear that once you are in, you will never be removed (unless you were compromised, at which point they'll happily add your new CA). I'll certainly not claim that we can magically make TLS work by starting to enforce the requirements we put on CAs. But we should call Mozilla out for their strongly worded letters and other nonsense to get some momentum for better solutions. The first step is admitting you have a problem.
- klapinat0r 13y ago
  Incorrect. DigiNotar[1] was permanently removed. Quoting Mozilla[2]: "This is not a temporary suspension, it is a complete removal from our trusted root program." [1]: https://en.wikipedia.org/wiki/DigiNotar#Issuance_of_fraudulent_certificates https://en.wikipedia.org/wiki/DigiNotar#Issuance_of_fraudule... [2]: https://blog.mozilla.org/security/2011/09/02/diginotar-removal-follow-up/ https://blog.mozilla.org/security/2011/09/02/diginotar-remov...
  - revelation 13y ago
    They were removed for being compromised and went bankrupt before trying again. I don't see how that invalidates what I said. Meanwhile, CAs that have issued CA=yes certs remain, but you know, Mozilla wrote them a letter reminding them on security best practices.
    - klapinat0r 13y ago
      True that they went bankrupt, but they filed on Sep 20th 2011, and the Mozilla decision to permanently remove them was a week prior or at least on Sep 2nd 2011; thus "unless you were compromised, at which point they'll happily add your new CA" is not correct. EDIT: To clearify, this is the phrase that we seem to be arguing about: Complete revocation of trust is a decision we treat with careful consideration, and employ as a last resort. (from the Mozilla Security Blog link above). They were effectively out in the cold, not "in once you're in" after the incident.

makomk 13y ago

They are in front of it, they've intentionally decided to continue charging for revoking these certificates.

sigzero 13y ago
Which they are 100% in the right to do by the way. StartSSL is a business.
- MereInterest 13y ago
  And then people are 100% in the right to no longer trust StartSSL certificates.
  - wlesieutre 13y ago
    Or at the very least, StartSSL certs issued before yesterday.
    - aroch 13y ago
      No, any StartSSL cert. There are still unpatched systems and there likely will be for quite some time. Just because there's a CVE doesn't mean everyone updated
      - wlesieutre 13y ago
        Sure, and there are also StartSSL certs used on systems that don't use OpenSSL. Do we just nuke every certificate, destroy their business, and force people with secure computers to buy new ones elsewhere even though their servers weren't affected by heartbleed? Because that option sucks too.
        claudius 13y ago
        As a user, I much prefer that my browser and my OS never ever again show a StartSSL-signed cert as valid over even just one compromised cert being displayed with a fancy lock. How StartSSL is going to achieve that, I don’t care, but neither do I care whether or not they go out of business over this.
  - shimon_e 13y ago
    Who is suppose to pay for staff time require to manually maintain the revoke list? BTW, revoking a cert doesn't stop it from being used in the wild as plenty of clients will still see it a valid cert. We are basically screwed until all current certs expire.
    - dpark 13y ago
      Why would the revoke list need manual maintenance? Are they running a root CA on pen and paper?
  - toomuchtodo 13y ago
    Who am I going to trust less? StartSSL certs? Or OpenSSL? I'm not saying I don't appreciate the hard work the two OpenSSL developers put into the code. But its incredulous that we're upset a business would give away a service due to an event completely out of its control.
    - pra 13y ago
      As it was out of the customers' control as well. Or should we have audited OpenSSL?
      - toomuchtodo 13y ago
        I would argue that something that is the underlying component of e-commerce, privacy, and security for the majority of Internet activity should probably be audited.
  - latortuga 13y ago
    This is utterly ridiculous. These people signed up for a free SSL certificate knowing that revocations would cost them money. Enforcing a contract that both parties willingly agreed to does not make them untrustworthy, it demonstrates that their customers carry immense senses of entitlement. There are tons of SSL certificate providers out there. Their business model was not a secret when folks signed up.
    - sparkie 13y ago
      The business model an practices of the company are completely irrelevant to the people concerned in this case though - the users of the web browser. They never signed up for shit, and they have no incentive to trust a business which refuses to invoke compromised certs because people are unwilling to pay. Yes, this doesn't make them untrustworthy as a business, since their business model was known - it does make them untrustworthy as a party which is meant to be securing the web though - because they'll be failing at it spectacularly. Again: I never signed up to shit, and their business model means shit to me too - I just want some degree of security when browsing.
    - MereInterest 13y ago
      I am not their customer. At most, I am their customers' customer. Why do I have any obligation toward StartSSL? If StartSSL is free to choose a business model, then others are free to make judgment based on that business model. Should the customers of StartSSL pay the revocation fee? Yes. Should StartSSL be required to give free revocations? No. Are either of those at all relevant to whether or not I should trust StartSSL certificates, given that they are less likely to be revoked when they need to be revoked?
  - lern_too_spel 13y ago
    The other responders have no idea what you're talking about. You're right, of course, but I suspect most other CA's certificates should also not be trusted even though they don't have this revocation hurdle because a lot of these certificates were given to unsophisticated users who aren't going to revoke compromised certificates anyway.
  - bsder 13y ago
    By that logic, we should remove every CA. At least StartSSL will have good certificates for everybody in their free tier in 12 months. Can any of the other CA's make that claim? So, are you going to lead the charge at the rest of the CA's? And, I suspect that StartSSL couldn't possibly handle every single person calling them up to revoke a certificate right now. And that is NOT their fault--they didn't cause this bug.
  - epsylon 13y ago
    People would better be advised not to trust admins who can't shed 20 bucks for their users' security. Most likely these are the same admins to whom you certainly wouldn't want to give an email, password or any sort of confidential information.

revelation 13y ago

Their CA status is not at risk, was never at risk and will not be at risk in the future. Mozilla has made their stance clear that once you are in, you will never be removed (unless you were compromised, at which point they'll happily add your new CA). I'll certainly not claim that we can magically make TLS work by starting to enforce the requirements we put on CAs. But we should call Mozilla out for their strongly worded letters and other nonsense to get some momentum for better solutions. The first step is admitting you have a problem.

klapinat0r 13y ago
Incorrect. DigiNotar[1] was permanently removed. Quoting Mozilla[2]: "This is not a temporary suspension, it is a complete removal from our trusted root program." [1]: https://en.wikipedia.org/wiki/DigiNotar#Issuance_of_fraudulent_certificates https://en.wikipedia.org/wiki/DigiNotar#Issuance_of_fraudule... [2]: https://blog.mozilla.org/security/2011/09/02/diginotar-removal-follow-up/ https://blog.mozilla.org/security/2011/09/02/diginotar-remov...
- revelation 13y ago
  They were removed for being compromised and went bankrupt before trying again. I don't see how that invalidates what I said. Meanwhile, CAs that have issued CA=yes certs remain, but you know, Mozilla wrote them a letter reminding them on security best practices.
  - klapinat0r 13y ago
    True that they went bankrupt, but they filed on Sep 20th 2011, and the Mozilla decision to permanently remove them was a week prior or at least on Sep 2nd 2011; thus "unless you were compromised, at which point they'll happily add your new CA" is not correct. EDIT: To clearify, this is the phrase that we seem to be arguing about: Complete revocation of trust is a decision we treat with careful consideration, and employ as a last resort. (from the Mozilla Security Blog link above). They were effectively out in the cold, not "in once you're in" after the incident.