Slashdot Mirror
Self-Regulating SSL Certificate Authority?
bcg asks: "It has come that time again to renew some of my SSL certificates and part with substantial amounts of cash. This has got me thinking - why should we pay large amounts of cash for authorized certs when so little is done by the companies issuing them? Sure they get you to send them a copy of a business certificate but how does this prove the character of those running the SSL server? What ideas can we come up with for a self-regulating certification authority? Could we set something up along the lines of the many free DNS servers around but use it to authenticate SSL certs?"
We last touched on this subject in October, when someone was searching for cheap
SSL certs. We've also discussed why certs are so expensive. Why not take it one step further and discuss ways of making and authenticating our own certs for free...or as close to free as possible?
Just self-sign a certificate. Truly, if it's not signed by some big name registrar, most internet users (IE of course) will get messages notifying them that it's not a "trusted" certificate anyways.
Hell, even Microsoft says that on their windows update site for the active X download it throws onto your computer during your first visit!
Someone should do a study on this, sounds like a great high school science fair project! I can see the display in the gym now, pasted on the cardboard display case "Are people idiots?" and have nice pie charts and tabular data from your research. It beats boiling something in a test tube to see how long it takes at different temperatures or testing the growth rates of different molds...
My standard rant about why I use my own certs:
Digital certificates are available, for a fee, from a commercial certificate authority (aka CA) such as Verisign. For about $15 a year Verisign will claim to know who you are though you provide no proof other than the grand American Dollar. If your credit card clears, then Verisign says email from you is from you. Why is this worth $15? If I send a signed email to someone and they verify that signature based on the cert I send them, then the only reason to trust that the cert is based on the trusting the signing CA. Verisign says that if I have a credit card with a name on it, then I am the person with that name. Unfortunately due to identy fraud, this is often not the case. In our family we have been victims both of simple credit card fraud (where are card number was stolen and the card duplicated) and full on identity fraud where our social security number was used to open credit accounts by people other than us. So merely the possession of a credit card number does not imply identity. By trusting Verisign you are trusting the US credit industry, which is corrupt and insecure.
Assume that you do trust that credit cards are valid identifications. Why would you trust the CA who took that as ID? How do you know who the CA is? CA's are identified by certificates just as users are. How did you get a certificate for the CA? Usually it is because Microsoft and Netscape include a set of certificates from trusted CA's in their products. If the cert comes from one of those CA's then Microsoft and Netscape say it's valid. Therefore you must trust that Microsoft and Netscape included authentic certs, and you assume that those certs have not been compromised since you installed the software. Maybe you think I'm paranoid. Really I just object to paying money for something I can do better myself.
I have created the Greenbaum.Org Certificate Authority to create digital certificates which are free and trusted. If you get an email from me, signed by a certificate issued by me, verified by the CA certificate you download from this site, then the email was from me. If you get an email from me, signed by a Verisign certificate, then it could have come from the gangsters who stole my credit card to buy Nikes and chinese food.
I see several difficulties with a free SSL-CA (as I see with free DNS/TLDs/whatever):
:-)
It's a great idea, but... who will use them? To be more specific: Verisigns capital is that it's root-certificate is in every browser on this planet. I don't want to know how much cash they had to throw at M$ to get their cert. into IE, but I doubt that a free CA can come up with that amount. Sure, we can probably get the certs into mozilla etc. and joe-schmoe IE-user can add the root-cert to his known certificates, but question is: what impression will your trustworthy buissiness give him, if he gets lots of warnings when on accessing your gimme-your-visa page. 'It's the value of trust(tm)'
just my two cents...
You could perhaps add the idea of a threshold -- once a cert is signed by enough well-trusted individuals, the cert becomes "good enough" to go public.
Of course, there might be an issue of startup time -- a requestor of a new cert wouldn't get one until it has had time to make the rounds and get signed by many trusted individuals.
There is also a bit of a seeding problem. How do you establish a large enough trusted community in the beginning, so that sufficient signings can be made on new certs.
Also, I would guess that one of the things that current commercial cert corporations provide is a source of culpability, should something go wrong with the cert they issued. With a public signing group, you might not have this same level of responsibiliy. This could be good or bad, depending on your perspective.
The only reason the big companies charge so much (their claim, not mine) is the insurance they provide, and the fact that they are "trusted" by the various vendors.
Any new group wanting to be a trusted CA will face the liability issue -- if one of your customers sues you, even if you try to disclaim all liability up front, you will still face massive court fees. Even if you won in court, you would lose financially if not insured.
There is no technical or logistical problem with setting up a Free (and free) common-geek's CA, the problems are entirely legal ones. I know because I looked into it right after SSL came out. It looks like a good business plan, right up until someone takes you to court.
frob.
//TODO: Think of witty sig statement
Why not have a self-regulating authority? Well, let me submit a request to sign my certificate saying I'm Amazon.com, hijack the domain and steal credit cards. The point of CA's is to do some background checking to verify you are who you say you are. Debatable, agreed, but is you're average script kiddie, cracker, etc. gonna shell out bucks to get a fake cert? Probably not. Not to mention once money is involved, there is an audit trail of some sort.
As for whether the prices are gouged a bit, I won't argue with you there. Seems that it shouldn't cost as much as it does, but at the same time I'd think most companies rack it up as a cost of doing business (just like rent, equipment leases, etc)
-- A computer without COBOL and Fortran is like a piece of chocolate cake without ketchup and mustard
Most of us just want the encryption features of SSL; most of us don't want it for authentication.
If you are a bank or something, then by all means authenticate your identity. If you just want to keep packet sniffing from being effective, self sign it.
GPG/PGP keys are always self-signed, yet no one complains about authentication of identity. Maybe we should all carry a compact flash card of our SSL keys!
You can't judge a book by the way it wears its hair.
The best way I can think of to do this is setup an infrastructure similar in principle to Google's PageRank. So, anyone can be granted a certificate, but the strength of that cert is based upon an index of reputation. Which to me personally, is somewhat more meaningful than any given company(TM) buying a certificate. What method you'd use to create such an index would require more investigation, with considerations for security and spoofing prevention.
At it's base though, I like the concept. And would like to hear some ideas on what we could use as "karma" *cough*... Realistically though, (and this is where I need help from those more familiar with SSL certificates than I...) is there a facility in the signing process which allows for extra certificate information at the time of request? To my memory, I think there is. For instance:
With the infrastructure already there, methinks the implementation is somewhat trivial. Can anyone help me refine the method?
Remember: umount it before you fsck it.
Have a ranking system that would base trust off the number of certificates, the age of the certificates and complaints from users.
So basically a centralized authority that gives out free or cheap (as in as cheap as domains) certificates.
You sign up with them as a reseller. All of your customers buy certs from you.
I'm thinking of this in terms of being a hosting provider as I am.
So I sign up with this centralized authority and purchase certificates for my customers.
Browsers could have a blacklist check on certs. So you try to hit one of my sites, it validates against your list of blacklisted sites that you updated last month and either:
A. Shows up with a good rating.
B. Doesn't show up because it's too new.
The user could then set a threshhold of trust and if the cert passed that threshhold it wouldn't warn them.
This idea isn't very thought out, just an idea I threw together. Run with it.
The man who trades freedom for security does not deserve nor will he ever receive either. - Benjamin Franklin
I used to work on Microsoft's Public Key Cryptography QA team. We worked with Verisign to create fake certificates to test IE's SSL and Authenticode signed downloads. When we were done testing, someone on our QA team called Verisign customer service and said, "hi, I work on Microsoft's QA team. We are done using those fake certificates for our tests. Can you please revoke (cancel) them?"
Without any further verification, the Verisign customer service agent pushed a button and canceled the real Microsoft certificate, the one used to sign all of Microsoft's downloads, device drivers, and CDs. oops. Luckily, no one pays attention to Verisign's CRL (Certificate Revocation Lists) because certificate revocation is off by default in IE. Since no one really used the CRL, Verisign was able to the remove Microsoft from the CRL and reinstate the Microsoft certificate after a couple days.
So when you "trust" Verisign, think hard about what that really means..
cpeterso
My problem occured when trying to get a cert for a small group of alumni. We've got about 50 people in it. We're just trying to make it possible for us to discuss things on our bulletin board with passwords protected with SSL.
We payed our money to Entrust. We still have not gotten a certificate or a refund. They first required that we prove we have a relationship with the school. We aren't an official organization, don't pretend to, and don't use their domain at all. It's completely separate.
So next they required we show articles of our encorporation. Is this what's required to have a certificate? Why can't joe-random-webmaster have a valid certificate from the "big guys"? Sure, you can go with smaller outfits, but their certs aren't in older browsers.
IMHO, a cert should simply say "This cert was given to the folks who run www.this_domain.com." They can check and verify whois data and your ability to receive email. Any other requirements are just stupid. Just because you want SSL doesn't mean you want to be an e-commerce site.
TLS (SSL) does not need the ugly PKI technology to operate. SSL/TLS could very well use PGP keys. The difference is that PGP technology is more well designed and lends better to help building a web of trust.
Some people might say that newbies can't handle the complexity. Well it's the responsibilty of software developers to help them overcome this. Example: As the same PGP keys would be used for mail, the web of trust could be linked to the addressbook handling.
Besides, the current model gives a sense of security which is not real. Do we really trust CA's? When you go to an "internet cafe", do people check that the list of trusted CA's haven't been altered. In this way, PGP would bring the real sense of security/insecurity which is currently "masked".