Comments

AlanSFebruary 14, 2012 7:42 AM

Trust requires that there are consequences when the trust is violated. It requires accountability.

This is a point that's well made in Moxie Marlinspike's talk at last year's Blackhat: "SSL And The Future Of Authenticity".
http://www.youtube.com/watch?v=Z7Wl2FW2TcA

Go to 24:30. What happened to Comodo? Nothing! Why...?

Kouett'February 14, 2012 9:09 AM

@AlanS
Because they cooperated. They weren't penetrated, one of their RAs were. They said they changed their procedures and access controls methods.
On the opposite, DigiNotar said nothing for more than a month, then lied, and weren't able to determine what had been done on their own platform. What happened to them?

AlanSFebruary 14, 2012 9:36 AM

@Kouett

Thanks. I wasn't asking a question or makng a comment about Comodo or even making a point about CAs. (The question was posed and then answered by MM. Watch the video.)

I was just making a general point about trust. Transparency is required but there also need to be consequences when trust is abused otherwise what's the point of transparency?

MM argues that we need "trust agility" which requires that:
1. A trust decision can be easily revised at any time
2. Individual users can decide where to anchor trust.

Kouett'February 14, 2012 10:03 AM

@AlanS
Sorry, I should have viewed the video again (saw it a few months ago).
I don't agree with MM on several aspects, and that's probably why I reacted that way, but that's not the point of the post.

I agree, trust needs transparency. And in Mr Schneier post, "VeriSign, Inc." clearly lacked transparency, and in fact lack of trust. It's not clear (to me, at least) what activities are conducted under the name "VeriSign, Inc.", and so how much trust is necessary to run these activities. Who, between "VeriSign, Inc." or "VeriSign by Symantec", operate the root DNSSEC key? The A and J root DNS servers? The .COM and other TLDs?

Disclaimer: I work for a CA and CA operator (not Comodo), I worked with the PKI activity of VeriSign, the part that was bought by Symantec and isn't covered by "VeriSign, Inc.".

AlanSFebruary 14, 2012 10:12 AM

@Kouett

I should probably have put that last line of my original e-mail in quotes. Not sure it was an exact quote but there were several slides that went something like that.

PersonFebruary 14, 2012 11:35 AM

I can't recommend "The Language of Trust: Selling Ideas in a World of Skeptics" highly enough. After I read this book my eyes were opened to all kinds of things taking place around me. I read a PR statement from Exxon-Mobil and recognized huge mistakes in their communication and how the more they tried to defend themselves, no matter how truthful, the worse their image became. I realized a lot of issues like global warming are trust issues. Very few people possess the knowledge and understanding to evaluate for themselves whether global warming is taking place and whether it is or is not caused by man. Skeptics simply do not trust the people ringing the alarm. The harder alarmists push the lower the trust. The alarmists then resort to name calling or other attempts to portray the skeptics as stupid. Animosity towards MS that bends to the absurd is another example. What's really intriguing is why this happens. If the CA paradigm loses trust it cannot be replaced. Nothing they say or do will get it back in time. They do not know this nor do they care.

Clive RobinsonFebruary 14, 2012 12:42 PM

In these days of "ever persistent" but invisable personal data "trust" is a huge problem.

In the old human world if trust was breached, in time most people forget about it and it did not usually form part of the public record, and even when it did it was usually very localised and in a way that would be difficult to search. Thus with time the wounds caused by a breach of trust healed.

However in our modern world "ever persistent" data means that a breach of trust is not forgoton ever. The data released is now usually in the public record (the Internet) in an easily searchable way from just about anywhere. Thus even a very very minor transgression can and in some cases definatly will come back to haunt you at a later date.

As an example of what can happen, I'm aware of a compleate and utter break down in a family, where one couple decided that they had the right to add the entire familes details to their web site. The couple did not ask they just did it and then got upset when other members of the family asked the couple to remove their details from the site. The couple dug their heals in and refused to remove any details until a little over a year later another member of the family suffered identity theft and fraudsters ran up huge debts using information that could have been obtained from the couples web site. Even though the couple finally saw the light of day and took the details down over ten years ago, the details can still be found cached in various places including several geneology sites who none of the family claim to have had contact with... For one member of that family they had a broken credit history they could not fix, debt collectors and courts to face and a reputation sullied beyond repair that haunts them today. Thus they refuse to have anything what soever to do with the couple and the rest of the family has taken and still maintain sides with an ever widening gap between them.

The moral being anybody can release your personal details without asking and not even say sorry, and others can and will gather up that information for their own benifit or profit as the law in other countries either alows them to or does not prohibit them from doing so. Large organisations will actually site parts of their operations in such jurisdictions just to evade problems in their home jurisdictions.

Untill such time as individuals have the right to say "no" cheaply and easily and insist that the copies of data are destroyed wherever they are in the world then people would be unwise to show any level of trust in others.

I personaly don't partake in the (mad?) craze of "social networking" nor do I alow my photograph to be taken, nor do I partake of store loyalty or other credit/debit cards.

I was taught during my early years as part of "work" to not trust people as the work was sufficiently confidential that others lives depended on it. So as a mater of habit I simply don't "trust people" in the way other more "socialy connected" people do. And in fact I actualy feel very uncomfortable around people who "spray and pray" with their personal details and daily activities as the chances are they will just as readily spray mine and other peoples personal details around along with their own with no consideration at all.

I have been told a number of times that trust is important in relationships, and that I should reveal all... I've even been told it's "all or nothing". To be honest I've found the exact opposite to be true as I suspect have quite a few others when it's to late. Because when relationships break down as they inevitably will people are hurt and some will get angry and throw what ever they have to hand. Thus you can look on their definition of trust as "leaving amunition for the enemy".

One thing many people don't understand is that trust can be partial or segmented. That is you can trust an individual in some ways but not in all. They have several saying in the UK army that begin "I'd trust him with my life..." and end with "but not to look after my drink" or "but not my wallet" or "but not alone with my girlfriend" etc etc.

In many ways I suspect life would be way way better for most people if they firstly didn't trust so easily, secondly segmented their trust with individuals and thirdly segmented ther lives.

However the trend driven by the likes of Facebook, Twitter and most other social media is the exact opposite to spray everything to the four corners of the earth for ever. So you have to ask "to who's benifit?" and "why?" along with "what am I realy getting out of this?".

The idea of Privacy Enhancing Technology (PET) has been around for atleast 15years, yet it appears to be for ever stymied by legislation or the lack of it. Usually the battle cry for removing privacy or preventing it is "think of the children" or "if you have nothing to hide" or it's necessary to "help fight" terrorists / drugs / crime etc etc. Yet every time we get this sort of legislation invariably the only thing it is used for is to remove the privacy of ordinary individuals never for what it was "advertised for". And why? usually for no better reason than somebody things it might be of benift to them financialy...

karrdeFebruary 14, 2012 12:56 PM

@Clive,

the whole concept of attempting to delete all copies of anything posted about me does not appear to be in the realm of the possible.

I agree with your general framing of the problem of trust. I also agree with your advice about segmented life and limited trust.

But being able to claim ownership of deletion privileges over data spread onto hundreds of servers by third-parties? I don't see that happening, not even if the entire user population of the Internet suddenly changes their mind to agree with you about it.

AlanSFebruary 14, 2012 2:34 PM

@Person

Another example of poor knowledge (or just not caring) and misplaced trust.

The Wall Street Journal today: "Chinese Hackers Suspected In Long-Term Nortel Breach". The Electronic version is behind a paywall here:
http://online.wsj.com/article/SB10001424052970203363504577187502201577054.html

Some of the main details in the WSJ article are also covered here:
http://business.financialpost.com/2012/02/14/nortel-computers-breached-by-hackers-for-over-a-decade-wsj/


PerseidsFebruary 14, 2012 5:12 PM

@Randy
"""Silly Question: What is your opinion of Convergence? (http://convergence.io/)"""

I think it's a nice Addon for us technically inclined users that boosts our security. But like NoScript I don't believe it's usability scales so well to the average user. Even more importantly while it fixes some problems it introduces new ones:

  • Privacy issues: You're telling every notary which websites you visit. Should TLS connections actually gain more widespread use (which would be good) this becomes a problem. You could counter this with having many, many notaries, yet this makes managing trust more difficult.
  • Design issues: Convergence relies on the assumptions that multiple network perspectives on a server raises your confidence in the certificate. Let's assume I want to communicate with a Chinese website. Notaries outside China are useless because of the single standard MITM access point the government and probably some private companies have access to. How do I (or Mozilla in the case of most users) know which notaries inside of China are to be trusted and won't subdue to court order? And even if that could be sorted out, relying on the network topology to provide security feels like abusing yet another infrastructure that wasn't built with security in mind.
  • Scalability: In the case TLS connection gain more widespread use, these notaries would become as basic a necessity to use as DNS. Unfortunately by design they cannot be as topologically close to you as DNS server are now thus the overhead would be substantially higher.

What I would like to see are more aggressive and thus potentially more effective ideas to reform the TLS trust model.
For example traditionally the problem with CA lies in the fact that identities are hard to verify and names mostly ambiguous. I guess we all know only one Bruce Schneier. But it is reasonably to assume that there might be others out there. Therefore a CA cannot simply reject a request by someone claiming to be a Bruce Schneier but has to check this potentially rogue request.
Fortunately in the DNS names are unambiguous. A new trust model could built on the ideas that CA no longer certify identities but one on one pairings between domains and keys, basically asserting "I've seen this key first and won't accept a second one". The key would then recursively implement a root CA for the specified domain. As an owner of a domain you would only launch your website, if all the relevant CAs have your domain root key stored. The operation of a top level domain CA would become very easy and enable the trust agility that is one of the key features of Convergence. This idea would require a coordinated effort to redesign the X.509 standards (or better yet built a modern one from scratch) and new policies regarding the CAs

PerseidsFebruary 14, 2012 5:46 PM

[Dammit, I accidentally clicked on "Post". Continued from last post:]
This idea would require a coordinated effort to redesign the X.509 standards (or better yet built a modern one from scratch) and new policies regarding the CAs, though, that nobody want's to tackle. I fear we will be stuck with yet another patch to a broken design.
In my eyes the best proposition so far is the Sovereign Keys approach of the EFF. Unfortunately their mistrust of authorities goes a little bit to far. If they had built their concept around more "official" institutions, key revocations even if you lost your revocation certificate would have been possibly. Their drastic approach also allows an attacker to block a domain forever, should there ever be a security problem with the authentication of the initial key insertion, or if there is a domain exchange ordered by a judge. Both of these issues will probably stop it from being adopted in the major browsers. And without their support the idea is basically dead already.

Clive RobinsonFebruary 14, 2012 7:10 PM

@ karrde,

But being able to claim ownership of deletion privileges over data spread onto hundreds of servers by third-parties? I don't see that happening, not even if the entire user population of the Internet suddenly changes their mind to agree with you about it.

True, the real problem of Personal Identifing Information, is the same as Pandora and her box, life is wonderful providing you keep the lid on it, but the minute the contents take flight like the proverbial genie it just won't go back again :-(

JohnstonFebruary 14, 2012 8:11 PM

@Perseids

"Privacy issues: You're telling every notary which websites you visit."

Moxie Marlinspike addresses this issue in his talk. Go to about 40 minutes in where he talks about:

1. local cert caching (contact notaries only upon seeing a new or different cert)

2. notary bouncing (sort of a single-hop Tor tunnel) which tunnels your cert lookup over SSL through an intermediary, so that the remote notary sees the intermediary and not you. So you keep your privacy.

Nick PFebruary 14, 2012 8:11 PM

I need to be sued in the european courts. This will encourage transparency if it happens enough.

VlesFebruary 15, 2012 12:25 PM

As disclosed in an SEC filing in October 2011, parts of Verisign’s non-production corporate network were penetrated. After a thorough analysis of the attacks, Verisign stated in 2011, and reaffirms, that we do not believe that the operational integrity of the Domain Name System (DNS) was compromised.

They believe instead of know for sure...

Leave a comment

Allowed HTML: <a href="URL"> • <em> <cite> <i> • <strong> <b> • <sub> <sup> • <ul> <ol> <li> • <blockquote> <pre>

Photo of Bruce Schneier by Per Ervland.

Schneier on Security is a personal website. Opinions expressed are not necessarily those of Co3 Systems, Inc..