December 15, 2000
by Bruce Schneier
Founder and CTO
Counterpane Internet Security, Inc.
A free monthly newsletter providing summaries, analyses, insights, and commentaries on computer security and cryptography.
Back issues are available at <http://www.counterpane.com>. To subscribe or unsubscribe, see below.
Copyright (c) 2000 by Counterpane Internet Security, Inc.
In this issue:
- Voting and Technology
- Crypto-Gram Reprints
- Counterpane Internet Security News
- Crypto-Gram News
- IBM’s New Crypto Mode of Operation
- The Doghouse: Blitzkrieg
- Solution in Search of a Problem: Digital Safe-Deposit Boxes
- New Bank Privacy Regulations
- Comments from Readers
In the wake of last November’s election, pundits have called for more accurate voting and vote counting. To most people, this obviously means more technology. But before jumping to conclusions, let’s look at the security and reliability issues surrounding voting technology.
The goal of any voting system is to establish the intent of the voter, and transfer that intent to the vote counter. Amongst a circle of friends, a show of hands can easily decide which movie to attend. The vote is open and everyone can monitor it. But what if Alice wants _Charlie’s Angels_ and Bob wants _102 Dalmatians_? Will Alice vote in front of his friends? Will Bob? What if the circle of friends is two hundred; how long will it take to count the votes? Will the theater still be showing the movie? Because the scale changes, our voting methods have to change.
Anonymity requires a secret ballot. Scaling and speed requirements lead to mechanical and computerized voting systems. The ideal voting technology would have these five attributes: anonymity, scalability, speed, audit, and accuracy—direct mapping from intent to counted vote.
Through the centuries, different technologies have done their best. Stones and pot shards dropped in Greek vases led to paper ballots dropped in sealed boxes. Mechanical voting booths and punch cards replaced paper ballots for faster counting. New computerized voting machines promise even more efficiency, and Internet voting even more convenience.
But in the rush to improve the first four attributes, accuracy has been sacrificed. The way I see it, all of these technologies involve translating the voter’s intent in some way; some of them involve multiple translations. And at each translation step, errors accumulate.
This is an important concept, and one worth restating. Accuracy is not how well the ballots are counted by, for example, the optical scanner; it’s how well the process translates voter intent into properly counted votes.
Most of Florida’s voting irregularities are a direct result of these translation errors. The Palm Beach system had several translation steps: voter to ballot to punch card to card reader to vote tabulator to centralized total. Some voters were confused by the layout of the ballot, and mistakenly voted for someone else. Others didn’t punch their ballots so that the tabulating machines could read them. Ballots were lost and not counted. Machines broke down, and they counted ballots improperly. Subtotals were lost and not counted in the final total.
Certainly Florida’s antiquated voting technology is partially to blame, but newer technology wouldn’t magically make the problems go away. It could even make things worse, by adding more translation layers between the voters and the vote counters and preventing recounts.
That’s my primary concern about computer voting: There is no paper ballot to fall back on. Computerized voting machines, whether they have keyboard and screen or a touch screen ATM-like interface, could easily make things worse. You have to trust the computer to record the votes properly, tabulate the votes properly, and keep accurate records. You can’t go back to the paper ballots and try to figure out what the voter wanted to do. And computers are fallible; some of the computer voting machines in this election failed mysteriously and irrecoverably.
Online voting schemes have even more potential for failure and abuse. We know we can’t protect Internet computers from viruses and worms, and that all the operating systems are vulnerable to attack. What recourse is there if the voting system is hacked, or simply gets overloaded and fails? There would be no means of recovery, no way to do a recount. Imagine if someone hacked the vote in Florida; redoing the election would be the only possible solution. A secure Internet voting system is theoretically possible, but it would be the first secure networked application ever created in the history of computers.
There are other, less serious, problems with online voting. First, the privacy of the voting booth cannot be imitated online. Second, in any system where the voter is not present, the ballot must be delivered tagged in some unique way so that people know it comes from a registered voter who has not voted before. Remote authentication is something we’ve not gotten right yet. (And no, biometrics don’t solve this problem.) These problems also exist in absentee ballots and mail-in elections, and many states have decided that the increased voter participation is more than worth the risks. But because online systems have a central point to attack, the risks are greater.
The ideal voting system would minimize the number of translation steps, and make those remaining as simple as possible. My suggestion is an ATM-style computer voting machine, but one that also prints out a paper ballot. The voter checks the paper ballot for accuracy, and then drops it into a sealed ballot box. The paper ballots are the “official” votes and can be used for recounts, and the computer provides a quick initial tally.
Even this system is not as easy to design and implement as it sounds. The computer would need to be treated like safety- and mission-critical systems: fault tolerant, redundant, carefully analyzed code. Adding the printer adds problems; it’s yet another part to fail. And these machines will only be used once a year, making it even harder to get right.
But in theory, this could work. It would rely on computer software, with all those associated risks, but the paper ballots would provide the ability to recount by hand if necessary.
Even with a system like this, we need to realize that the risk of errors and fraud cannot be brought down to zero. Cambridge Professor Roger Needham once described automation as replacing what works with something that almost works, but is faster and cheaper. We need to decide what’s more important, and what tradeoffs we’re willing to make.
This is *the* Web site on electronic voting. Rebecca Mercuri wrote her PhD thesis on the topic, and it is well worth reading.
Good balanced essays:
Pro-computer and Internet voting essays:
Problems with New Mexico computerized vote-counting software:
The Fallacy of Cracking Contests:
How to Recognize Plaintext:
“Security is a process, not a product.”
European Digital Cellular Algorithms:
One of the problems facing a network security administrator is that there are simply too many alerts to deal with:
Secret (and unauthorized) CIA chat room:
The world’s first cybercrime treaty is being hastily redrafted after Internet lobby groups assailed it as a threat to human rights that could have “a chilling effect on the free flow of information and ideas.”
A Field Guide for Investigating Computer Crime. Five parts, very interesting:
Interview with Vincent Rijmen (one of the authors of Rijndael) about AES:
Microsoft’s Ten Immutable Laws of Security. A good list, actually.
A new report claims that losses due to shoddy security cost $15B a year. Investments in network security are less than half that. Sounds like lots of people aren’t doing the math.
NESSIE is a European program for cryptographic algorithm standards (kind of like a European AES, only more general). Here’s a list of all the algorithms submitted to the competitions, with links to descriptive documents. Great source for budding cryptanalysts.
More Carnivore information becomes public. Among the information included in the documents was a sentence stating that the PC that is used to sift through e-mail “could reliably capture and archive all unfiltered traffic to the internal hard drive.” Since this directly contradicts the FBI’s earlier public assertions, why should anyone trust them to speak truthfully about Carnivore in the future?
Independent Carnivore review less than stellar:
Carnivore: How it Works
Interesting biometrics reference site:
The People for Internet Responsibility has a position paper on digital signatures. Worth reading.
The Global Internet Project has released a research paper entitled, “Security, Privacy and Reliability of the Next Generation Internet”:
More on the stolen Enigma: When it was returned, some rotors were still missing. And there’s been an arrest in the case.
The pros and cons of making attacks public:
And the question of retaliation: should you strike back against hackers if the police can’t do anything?
Commentary on Microsoft’s public response to their network being hacked.
A review of cybercrime laws:
During WWII, MI5 tested Winston Churchill’s wine for poison by injecting the stuff into rats. This is a photo of a couple of very short typewritten pages detailing the report.
Internet users have filed a lawsuit against online advertiser MatchLogic Inc., alleging that their privacy was violated by the company’s use of devices that track their Web browsing habits.
A Swiss bank, UBS AG, has just issued a warning bulletin to Outlook and Outlook Express users of its Internet banking service. There is a virus out there that, when a customer attempts an Internet banking transaction, will present legitimate-looking HTML menus, prompt the user for his Internet banking passwords and security codes, and send the information to its own server.
Security and usability:
Top 50 Security Tools. A good list, I think.
Social engineering at its finest: The Nov. 27 issue of _The New Yorker_ has a story written by someone who quit his job to write, but discovered he never got anything done at home. So he strolled into the offices of an Internet startup and pretended to work there for 17 days. He chose a desk, got on the phone list, drank free soda and got free massages. He made fake business phone calls and brought his friends in for fake meetings. After 6 PM you’re supposed to swipe a badge to get in, but luckily a security guard held the door for him. He only left when they downsized almost everyone else on his floor—and not because they caught on; he went around saying goodbye to everyone in the office and everyone wished him well. No Web link, unfortunately.
150-year-old Edgar Allan Poe ciphers decrypted:
Very interesting talks on hacking by Richard Thieme (audio versions):
Picture recognition technology that could replace passwords:
Good article on malware:
Not nearly enough is being done to train information security experts, and U.S. companies face a staffing shortfall that will likely grow ever larger.
Luciano Pavarotti could not check in at his Italian hotel because he lacked proper identification. When you can’t even authenticate in the real world, how are you ever going to authenticate in cyberspace?
After receiving a $10M anonymous grant, Johns Hopkins University is opening an information security institute:
Most countries have weak computer crime laws:
Plans for an open source operating system designed to defeat U.K.’s anti-privacy laws:
Microsoft held an invitational security conference: SafeNet 2000. Near as I can tell (I wasn’t there; schedule conflict), there was a lot of posturing but no real meat. Gates made a big deal of new cookie privacy features on Internet Explorer 6.0, but all it means is that Microsoft is finally implementing the P3P protocol…which isn’t all that great anyway. Microsoft made a great show of things, but talk is a lot cheaper than action.
Speaking of action, Microsoft now demands that security mailing lists not republish details of Microsoft security vulnerabilities, citing copyright laws.
More reviews of Secrets and Lies:
<http://www.business2.com/content/research/reviews/…> All reviews:
Crypto-Gram has been nominated for an “Information Security Excellence Award” by Information Security Magazine, in the “On-Line Security Resource” catagory. If you are a subscriber to the magazine—it’s a free subscription—you can vote. You will need a copy of your magazine’s mailing label. Voting is open until 17 January.
Thank you for your support.
In November, IBM announced a new block-cipher mode of operation that “simultaneously encrypts and authenticates,” using “about half the time,” and is more suited for parallelization. IBM’s press release made bold predictions of the algorithm’s wide use and fast acceptance. I’d like to offer some cautionary notes.
Basically, the research paper proposes two block cipher modes that provide both encryption and authentication. It’s author Charanjit S. Jutla at the T.J. Watson Research Center. This is really cool research. It’s new work, and proves (and shows how) integrity can be achieved for free on top of symmetric-key encryption.
This has some use, but I don’t see an enormous market demand for this. A factor of two speed improvement is largely irrelevant. Moore’s Law dictates that you double your speed every eighteen months, just by waiting for processors to improve. AES is about three times the speed of DES and eight times the speed of triple-DES. Things are getting faster all the time. Much more interesting is the parallelization; it could be a real boon for hardware crypto accelerators for things like IPsec.
Even so, cryptographic implementations are not generally hampered by the inefficiency of algorithms. Rarely is the cryptography the bottleneck in any communications. Certainly using the same cryptographic primitive for both encryption and authentication is a nice idea, but there are many ways to do that.
Combining encryption with authentication is not new. The literature has had algorithms that do both for years. This research has a lot in common with Phillip Rogaway’s OCB mode. On the public-key side of things, Y. Zheng has been working on “signcryption” since 1998.
Most security protocols prefer separating encryption and authentication. The original implementation of PGP, for example, used the same keys for encryption and authentication. They were separated in later versions of the protocol. This was done for security reasons; encryption and authentication are different. The key management is different, the security requirements are different, and the implementation requirements are different. Combining the two makes engineering harder, not easier. (Think of a car pedal that both accelerates and brakes; I think we can agree that this is not an improvement.)
Unfortunately, IBM is patenting these modes of operation. This makes it even less likely that anyone will implement it, and very unlikely that NIST will make it a standard. We’ve lived under the RSA patents for long enough; no one will willingly submit themselves to another patent regime unless there is a clear and compelling advantage. It’s just not worth it.
IBM has a tendency of turning good cryptographic research into ridiculous press releases. Two years ago (August 1998) IBM announced that the Cramer-Shoup algorithm was going to revolutionize cryptography. It, too, had provable security. A year before that, IBM announced to the press that the Atjai-Dwork algorithm was going to change the world. Today I can think of zero implementations of either algorithm, even pilot implementations. This is all good cryptography, but IBM’s PR department overreaches and tries to turn them into things they are not.
Rogaway’s OCB Mode:
My write-up of Cramer-Shoup:
This is just too bizarre for words. If the Doghouse had a hall of fame, this would be in it.
Digital safe-deposit boxes seem to be popping up like mushrooms, and I can’t figure out why. Something in the water? Group disillusionment? Whatever is happening, it doesn’t make sense to me.
Look at the bank FleetBoston. In October, they announced something called fileTRUST, a digital safe-deposit box. For $11 a month, FleetBoston will store up to 40MB of stuff in their virtual safe deposit box. Their press release reads: “Document storage enables a business owner to expand memory capacity without having to upgrade hardware and guarantees that files will be protected from deadly viruses…” Okay, $11 for 40MB is $0.28 per MB per month. You can go down to any computer superstore and buy a 20 Gig drive for $120; if we assume the drive will last four years, that’s $0.0001 per MB. Is it that difficult to add a new hard drive to a computer? And the “deadly viruses” claim: storing backups offline is just as effective against viruses, and fileTRUST’s feature that allows you to map your data as a network drive makes it just as vulnerable to viruses as any other drive on your computer. Or if you don’t map the fileTRUST archive, isn’t the decryption key vulnerable to viruses?
I dismissed this as a bank having no clue how computers work, but then I started seeing the same idea elsewhere. At least three other companies—DigiVault, Cyber-Ark, and Zephra—are doing essentially the same thing, but touting it as kind of a poor man’s VPN. You can use this virtual safe-deposit box as kind of a secure shared hard drive. Presumably you can give different people access to different parts of this shared space, take access away quickly, reconfigure, etc.
The DigiVault site is the most entertaining of the bunch. There are a lot of words on their pages, but no real information about what the system actually *does* or how it actually *works*. Even the “Technical Specifications” don’t actually specify anything, and instead parrot some security buzzwords.
First off, the safe-deposit box metaphor (Cyber-Ark calls it a “Network Vault: ™) makes no sense. The primary value of a safe-deposit box is reliability. You put something in, and it will remain there until you show up at the bank with your key. That’s why it’s called a “safe-deposit box” and not a “private deposit box,” although privacy is a side benefit. The “digital safe-deposit box” provides privacy (insofar as the system is actually secure), but is just as vulnerable to denial-of-service attacks as any other server on the Internet. And the box is only secure against actual destruction of the data insofar as they back up the data to some kind of media and store it somewhere. These companies presumably make backups, but how often? Where and how are the backups stored? The Web sites don’t bother to say.
The problem with this metaphor becomes apparent when you read the “No Time for a VPN?” article (second DigiVault URL). The author says it’s like a safe deposit box because you need two keys to open it: the bank uses your public key and you use your private key. But the point of having two keys to a real safe deposit box is that the bank only provides its key after you prove your identity; that way, someone stealing your key can’t automatically get into your box. It works because the bank’s key is *not* public. With DigiVault, the bank uses the same public key that you give to others so they can send stuff to your box. In that case, what’s the point of the bank’s key?
Second, I don’t understand the business model. Yes, VPNs are hard to configure, but they’re now embedded in firewalls and operating systems, and are getting easier to use. Yes, it’s nice to have a shared piece of digital storage, but 1) I generally use my webserver, and 2) there are companies like X-Drive giving this stuff away. Once you combine encrypted mail with free Web storage space, you have the same functionality that a virtual safe-deposit box offers, for free. Now you’re competing solely on user interface.
A digital safe-deposit box (or whatever it should be called) might be the ideal product for someone. But I just don’t see enough of those someones to make a viable market.
There are some new (proposed) interagency guidelines for protecting customer information. Near as I can tell, “interagency” includes the Office of the Comptroller of the Currency (Treasury), Board of Governors of the Federal Reserve System, and Office of Thrift Supervision (also Treasury). If you’re a bank, this is a big deal. Ensuring the privacy of your customers will now be required.
Here are some highlights of the proposals:
The Board of Directors is responsible for protection of customer information and data.
The Board of Directors must receive reports on the overall status of the information security program, including materials related to attempted or actual security breaches or violations and responsive actions taken by management.
Monitoring systems must be in place to detect actual and attempted attacks on or intrusions into customer information systems.
Management must develop response programs that specify actions to be taken when unauthorized access to customer information systems is suspected or detected.
Staff must be trained to recognize, respond to, and where appropriate, report to regulatory and law enforcement agencies, any unauthorized or fraudulent attempts to obtain customer information
Management must monitor, evaluate, and adjust, as appropriate, the information security program in light of any relevant changes in technology, the sensitivity of its customer information, and internal or external threats to information security.
These rules are an addition to something called Regulation H. Regulation H is an existing section of legal code that covers a variety of stuff, including the infamous “Know Your Customer” program.
Comments on the proposed rules:
Some other privacy regulations that went into effect on 13 November, with optional compliance until 1 July 2001:
You didn’t hear this from me, but:
– The attackers didn’t get in using QAZ. As of last week, Microsoft still didn’t know how they entered the network. The media invented the QAZ story, and Microsoft decided not to correct them.
– The damage is much worse than anyone has speculated.
I was involved with Microsoft’s interaction with the press over “the event.” What actually got told to the press was a completely *separate* incident than the one that really caused the problems. The reason that none of the stories agreed was that they were all fiction.
From: Julian Cogdell <Juliancrfs.co.uk>
Subject: Microsoft “set hacker trap” theory
Not quite the “penetration test by a Microsoft tiger team” you predicted in
the latest Crypto-Gram, but it’s almost there….
From: “Ogle Ron (Rennes)” <OgleRthmulti.com>
Subject: Implications of the Microsoft Hack
I agree with you about this being an unprofessional job, but I wonder what will happen when this becomes a professional job with long-term objectives. I keep thinking that the computerized world is going to have its Black Plague.
If someone wanted to devastate the computerized world, one way would be to plant code into a future release of an operating system that would be widely disseminated and remotely triggerable. If an attacker were to have a long-term objective, she could steal the code, create 30 or 40 vulnerabilities in several different parts of the software, and return the code. Then, say in three years, the attacker could determine which vulnerabilities remained in the “released” software.
She would then devise ways to find the quickest and deadliest attacks while waiting for an additional two years for the software to become entrenched in the world. At this time, she would deploy one vulnerability to show the world what power she could wield. Because the vulnerabilities would be in several different parts of the operating system, it would be very difficult (i.e., near impossible) to remove these other surviving vulnerabilities or even defend against them. The one exception for a defense would be to unplug yourself from the Internet. I’m thinking in five years, to unplug yourself from the Internet for any prolonged period of time would be tantamount to going out of business.
The hacker would wait for two days and put out a demand for whatever she wanted and of course immunity from prosecution, or she would unleash the other vulnerabilities to the other computers that still remained on the Internet. Either way, corporations are losing billions a day, and these corporations would put such pressure on their governments to do whatever was required. Remember, if you’re off the Internet to protect yourself, then you can’t support commerce. The nice part for the operating system company is that they are covered because all of these corporations are using “AS IS” software with no guarantees or warranties.
How is this possible? Technically, all of the pieces are there to accomplish such an attack. I believe that motive is still missing with the people who are technically capable. I believe this is a reasonable possibility because of the following:
1. With a little more knack, Microsoft could have been hacked without being detected and the attacker could have downloaded the software for a future release. The attacker would also steal a few passwords to be able to get back in as an authorized user for the future.
2. We have seen that people can write some very dangerous code, usually through viruses. Given the source code, a person could devise very very dangerous code and could disguise it. Remember that Microsoft programmers often embed “Easter eggs” and self-promoting code that makes it through their quality assurance checks. Now to make sure that enough vulnerabilities survive (5 to 10) into the released version, the attacker would need to create 30 to 40 such vulnerabilities.
3. Based upon the openness of the software sharing, the attacker could come back in with one or more of the authorized user accounts. The attacker then uploads the “new” software into the code base. Some of this code will be lost through normal evolution of the code base, but enough of the exploits should survive.
4. We know that security is not really looked at from a quality assurance or testing perspective because of the sure number of vulnerabilities that are uncovered that should have never been there in the first place. Programmers/testers are basically not very knowledgeable in good software engineering practices, so “bad” code doesn’t affect them much. Therefore, if the code works, they are likely to say good enough.
5. Most companies support a computer infrastructure made up of mostly a MS Windows environment. Because companies have this homogeneous solution, all of their systems would be very vulnerable to this or other types of attacks which would devastate their business. This of course was seen during the Love Bug virus when companies with mostly MS Windows systems were brought to their knees. Just as in nature, diversity is rewarded, but the computer world is reversed (that is, until the Black Plague!).
I think that the above scenario is definitely possible. With the dependence upon MS Windows and the growing dependence upon the Internet to conduct business, the above attack would cause huge devastation. The piece lacking to make this scenario real is a professional group or person with a motive that is willing to invest the time and effort for a long term pay-off. Terrorist groups could have a field day with this. The nice part about this is that if in the next five years the attacker decides not to go through with the attack, then they can just leave the vulnerabilities intact and nobody will be the wiser.
From: “Louis A. Mamakos” <louieTransSys.COM>
Subject: Digital Signatures
I found your essay on “Why Digital Signatures Are Not Signatures” very interesting. There’s an analogue in the Real World which might help explain the situation.
It’s the check signing machine. It contains the “signature” of a Real Person, and is used to save the Real Person the drudgery of actually signing 5000 paychecks every couple of weeks. Did the Real Person every actually see each of these documents? Nope, but there’s an expectation that the check signing machine is used only for authorized purposes by authorized personnel. Much the same as software which computes the RSA algorithm to “sign” documents.
It’s interesting that the use of the check signing machine probably wouldn’t be allowed for, e.g., signing contracts. I suppose it’s all about expectations.
From: Douglas Davidson <drdprimenet.com>
Subject: Digital Signatures
We can perhaps gain some historical perspective on this issue by considering a predecessor of signatures, namely seals. Seals are an ancient human invention, probably antedating writing; they have been used by cultures around the world for purposes similar to those that might be served by digital signatures: providing evidence of the origin and authenticity of a document, indicating agreement, and so forth. Seals also have a similar drawback: they do not really provide evidence of the intent of a particular person, only of the presence of a certain object, which could equally well have been used by anyone who came into possession of it. A common theme in the literature of cultures that use seals is their misuse by unauthorized persons, often close associates or family members of the rightful owner. Even if you have a trusted signing computer, for which you can maintain complete hardware and software security, can you be certain that your children can’t get to it?
From: Ben Wright <Ben_Wrightcompuserve.com>
Subject: Digital Signatures
You are correct about the problems with digital signatures; they do not prove intent. They do not perform the legal wonders claimed by their most zealous proponents.
But you are wrong about the new E-Sign law. First, the law does not say that digital (or electronic) signatures are equivalent to handwritten signatures. Laymen summarize the law that way, but strictly speaking, that is not what the law says.
Second, the E-Sign law does not say that digital signatures prove intent or anything else. The new E-Sign law is very different from the (misguided) “digital signature” laws in states like Utah.
The E-Sign law is good. It simply says that an electronic signature (whether based on PKI, biometrics, passwords or whatever) shall not be denied legal effect solely because it is electronic. That is all it says. It does not address proof of intent, proof of action or proof of anything else. It does not specify technology. It does not even mention digital signatures, asymmetric algorithms, public/private key systems or PKI.
From: “Herbert Neugebauer” <herbert_neugebauerhp.com>
Subject: Digital Signatures
Your article shoots into the wrong direction. You discredit the principle of digital signatures. Your explanation, however, does not convince me. The examples of why digital signatures will never be 100% safe are correct, but the same thing applies to real ink-on-paper signatures. You partly even acknowledge in your article that there are cases in court where these real signatures are denied. And these cases are sometimes won, sometimes lost.
Is the risk of digital signatures higher than the risk on the ink-on-paper signatures? We don’t know. There are hundreds of ways to fake “ink-on-paper” signatures. There are similar “ways of attacks,” like technical attacks (fake signatures) or social attacks, attacks to lead people to sign a paper that contains different things than they believe they sign.
Some are good, others are weak and people can prove easily that they didn’t sign or didn’t want to sign or thought they actually wanted to sign something else. Where’s the difference to digital signatures?
I personally think the future will have to show how strong or weak the digital signatures actually are compared to “real” signatures. In the meantime I think your article is counter-productive. It generates distrust. I think you intended to warn people that blind trust in technology is wrong and that just by implementing PKI and using digital signatures things are not automatically completely secure. That’s correct. That’s good. That’s important.
However the statement “These laws are a mistake. Digital signatures are not signatures, and they can’t fulfill their promise,” is in my view plain wrong. We can only judge this 10 years down the road once we really used the technology and can really compare how it works in comparison with “real” signatures. Today digital signatures are virtually non-existent—not used at all.
We should start adopting. We have to constantly review, check, test, warn, revise and newly invent both technology and laws. We should be careful, not be blind, but we should not dig a big hole and hide in fear of the “end of the world.”
From: Peter Marks <macramediahome.com>
Subject: Trusting Computers
In the latest Crypto-Gram you wrote in one context:
> Because the computer is not trusted, I cannot rely on
> it to show me what it is doing or do what I tell it to.
And in another:
> “… the computer refused to believe that the power had
> gone off in the first place.”
There’s an ironic symmetry here. Perhaps computers feel hampered by a lack of trusted humans. 🙂
From: jfunchionanswerthink.com (Jack Funchion)
Subject: Semantic Attacks
I have been following the discussion on semantic attacks in the Crypto-Gram the last two months, particularly the idea of changing old news stories in archives and the like. In a previous job I worked for a company that among other things provided a technical analysis system for evaluating stocks. It was based on a database of pricing history, and I can remember dreaming up an idea of how to make a killing in the stock market. The idea is simply to go back and change the stock pricing data in small increments in the databases so that the various technical analysis equations used by quantitative traders will be wrong, and predictably so. You then take positions opposite those predicted by the now known (by you) to be incorrect analyses. I even came up with a name for this kind of attack—the Saramago Subterfuge. The name comes from Jose Saramago, Portuguese novelist and Nobel Literature prize winner. He wrote a book a few years back called _The History of the Siege of Lisbon_ which revolves around a proofreader who changes a single word in a historical archive and thus changes the history of his country. I recommend it for your readers.
From: Xcott Craver <sacraverEE.Princeton.EDU>
I’m one of the Princeton researchers who participated in the successful hack of SDMI’s technologies. I read your column about SDMI with interest, but have a few small comments and possible objections:
> Near as I can tell, the SDMI break does not conform to
> the challenge rules, and the music industry might claim
> that the SDMI schemes survived the technology.
Indeed, we have many reasons to believe the contest rules were overstrict. Watermark detectors were not directly available to us, but kept by SDMI, who would test our music _for us_ with an overhead of at least a few hours, sometimes half a day. Not only did this prevent oracle attacks (which real attackers will perform, almost surely,) but the oracle response did not tell us whether failure was due to the watermark surviving, or due to a decision that the music was too distorted.
Also, as you suspect, our submissions were not considered valid in the second round because we did not provide information about how the attacks worked by their deadline.
We also had reason to believe that at least one of the oracles did not behave as documented. That’s perhaps the least extreme way to say it. The two “authentication” technologies (the other four were watermarking technologies) were inherently untestable; when SDMI claims that three technologies survived, chances are they are counting those two.
> Even if the contest was meaningful and the technology
> survived it, watermarking does not work. It is
> impossible to design a music watermarking technology
> that cannot be removed.
Ahem. Watermarking works just fine in other application domains, just not this one. By changing the application, one can move the goal posts so that attacks are no longer worth anything.
Consider as an example the (digital) watermarking of currency, so that scanners and photocopy machines will recognize a bill and refuse to scan it. This can be attacked in the usual way, but if the watermark was made visible rather than invisible, the standard attack of removing the mark becomes worthless; for without the mark, the bill appears clearly counterfeit to a human observer.
> Here’s a brute-force attack: play the music and re-
> record it. Do it multiple times and use DSP technology
> to combine the recordings and eliminate noise.
It is not clear that this will always work for all watermarking techniques. On the other hand, if you have the capability of playing and re-recording music, you have already foiled the watermark.
My colleague Min Wu developed a similar technique for video, which involves simulating transmission error by leaving out MPEG blocks, then correcting for those missing blocks using DSP techniques. After enough “playing and re-recording” a good deal of the original data is long gone.
> Even if watermarking works, it does not solve the
> content-protection problem. If a media player only
> plays watermarked files, then copies of a file will
> play. If a media player refuses to play watermarked
> files, then analog-to-digital copies will still work.
Watermarking schemes are designed to survive digital-analog-digital conversion. Very robust image watermarking schemes exist which appear to survive printing, xeroxing, then rescanning a watermarked image.
> Digital files intrinsically undermine the scarcity
> model of business: replicate many copies and sell each
> one. Companies that find alternate ways to do
> business, whether they be advertising funded, or
> patronage funded, or membership funded, or whatever,
> are likely to survive the digital economy. The media
> companies figured this out quickly when radio was
> invented—and then television—so why are they so
> slow to realize it this time around?
It is indeed surprising, given media companies’ previous history. Until the Internet (or maybe until Digital Audio Tape,) the recording industry seemed to view new technology as a new business opportunity. They went digital over a decade ago. Now they seem to want to sue the landscape itself into not changing anymore.
It is difficult to suppress the image of the crazy old miser, driven paranoid by fabulous wealth. Perhaps the flimsy compact “disc,” enclosed in a flimsy jewel box yet wrapped in so much anti-theft plastic that a local TV news show aired a segment on how annoying they were, reinforces this view.
Interestingly, a great deal of the rise of MP3s is due to the recording industry’s shoddy technology and unsuccessful distribution of music. People want music that won’t skip, while we still use a 15-year-old medium that requires moving parts to read. People want to find specific albums, that just can’t find room in a physical record store. The recording industry is not merely a victim of shifting landscape, but a major cause of it, through their own failure to act.
From: Andrew Odlyzko <amoresearch.att.com>
I agree with all your points about the SDMI hacking challenge, and would like to add another, which, surprisingly, I don’t hear people mention. (I just came back from a conference in Germany on Digital Rights Management, and although many speakers dealt with watermarking, not one mentioned this problem.) What exactly is the threat model that watermarking is supposed to address? Even if you do have an iron-tight technical solution, all that will allow the content producer to determine is who bought the goods from a legitimate merchant. If I am an honest citizen who abides by the rules, and my laptop loaded with honestly purchased movies is stolen, Hollywood might be able to tell that the pirated copies came from my hard drive, but are they going to hold me responsible for their losses?
> The media companies figured this out quickly when
> radio was invented—and then television—so why
> are they so slow to realize it this time around?
I agree with you completely about the need for new business model. (My talk in Germany was on “Stronger copyright protection for cyberspace: Desirable, inevitable, and irrelevant,” and I discussed how the industry really needs to think more creatively about their business instead of threshing around hoping for secure protection schemes.) However, the claim that “[t]he media companies figured this out quickly when radio was invented” is definitely not correct. It took about a decade for this process. You can read about it in Susan Smulyan’s book, “Selling Radio: The Commercialization of American Broadcasting, 1920-1934,” Smithsonian Institution Press, 1994.
From: “Marcus J. Ranum” <mjrnfr.com>
Subject: Window of Exposure
I finally got a chance to re-re-read your article on reducing the window of exposure for a vulnerability, and I’d like to make a few comments. First off, I think that you’ve hit on a few very important ideas. I don’t know of a way to tie your “exposure window” charts to a real, measurable, metric, but if we could, that would provide invaluable information to help people decide on their course of action in dealing with a vulnerability. There’s a subtle point, which you note, that the important goal is to minimize the space under the curve: the number of users that are vulnerable at any given time.
So, you’ve given us a model whereby we can point and say “you are here” during the course of any given security flaw/response cycle. If you look at Figure 2 (limit knowledge of vulnerability) the area below the curve is dramatically less than the area below the curve in Figure 1 (announce the vulnerability). That’s very significant.
Your model of how the threat of the vulnerability “decays” is also thought-provoking. For an example in many of my talks I refer to the “rlogin -froot” bug, which was a vulnerability in AIX 3 (if I recall correctly) in the early ’90s. Just about a month ago, I had a system administrator in my class ask me for details on how to fix that particular problem; he’s still running AIX without patches. So, there is, indeed, a “tail off” factor to the curve, like you predicted. I’ve seen it.
You’ve also missed a very important special case scenario: the one in which a vulnerability is found, quietly diagnosed, quietly fixed, and never brought to the attention of the hacking community-at-large. In that case, the area under the curve is _zero_. Nobody is threatened or hurt at all. This points to a couple of things:
1) Vendors need to take security-critical quality assurance to a much higher level than they do. Finding and fixing your own bugs quickly and quietly is the only 100% victory solution.
2) Vendors need to be able to ensure that users actually install their patches.
The latter point is critical. I believe that within the next five years software will become self-updating for many applications. Antivirus software and streaming media/browsers do this today. The former does it to update its rules, the latter to install new bugs on your system faster and more easily. But security critical products need to do the same thing. Imagine installing a piece of security critical software and having it, at install time, ask you:
“This software has the ability to self-update in the event of critical functionality or security patches. In such an event, should I:
A) Cease functioning and notify an administrator to manually install an upgrade before resuming processing B) Continue functioning in a reduced capacity
C) Automatically install the update and continue to function.”
Providing a good automatic update service has some daunting technical requirements (signed code, secure distribution servers, etc.), but those problems are not significantly worse than the problems we face today in getting our users to update all their software manually. Perhaps savvy vendors will realize that such service provides an opportunity to “touch” their customers in a positive way (good marketing) on a regular basis, as well as to justify software maintenance fees. Ironically, Microsoft, who many hold as the great Satan of computer security, is leading the way here: recently the Microsoft IIS team fielded a program called HFCheck that automatically checks for IIS server security updates and alerts the user. The first vendor that can make a believable claim to have licked this problem will reap potentially huge rewards.
In such an environment, a vendor could easily base their judgements on progress along your exposure charts. As soon as there is a certain number of users at risk, it’s time to push out an upgrade. Indeed, I predict that in such an environment, it’ll become an interesting race between the hacker and the vendor to see if the hacker can issue an alert before the vendor can draw their fangs and make them look redundant by already having released a patch. I look forward to seeing this happen, since it’s a necessary step in triggering a change in the current economy of vulnerability disclosure. Under the current economy, the hackers reap real benefits (ego and marketing) in spite of the users they are placing at risk. If they no longer reap those benefits, a significant component of their motivation will be gone.
Then we’ll be left to deal with the individuals whose motives are purely malicious.
Subject: Anecdote about “open” WaveLAN networks.
I found my first “open” WaveLAN (IEEE 802.11) network by accident. I had a WaveLAN card in my laptop when I visited the California office of the company I work for. My first reaction to getting a working dhcp lease was “Great, I won’t have to fiddle with cables. But I think I need to talk to the local sysadmin if he has thought about security.” My happiness quickly changed into annoyance when I felt how slow the network was and the annoyance changed into surprise when while debugging the network I realized that XXX.com wasn’t the domain name of the company I work for (as a side note: XXX sells crypto hardware). I reported the incident to the local sysadmin and forgot about it.
When I got back to Sweden, I told about the stupidity of XXX to a few friends at a restaurant in downtown Stockholm. Some time before the food arrived we started to discuss WaveLAN and somehow a laptop showed up on the table and voila! We were inside YYYinternal.com. We knew a guy working at YYY, told him about this, he told his sysadmin, the sysadmin responded “I’ll have to talk to the firewall guy.” (I didn’t know that firewalls had TEMPEST protection in their default configuration.) AFAIK the network has been shut off.
Another month or two passed. I was riding the bus around downtown Stockholm to get home after a pretty late evening and I was too tired to read. I fired up my laptop and started to detect networks. I found six or seven (one could have been a duplicate) during 30 minutes.
A week later a friend from Canada visited us. He stayed at a hotel in central Stockholm. He had a working network in some spots in his room. Apparently it belonged to a law firm. On the square outside the hotel the networks didn’t work, simply because there were three of them fighting with each other. When we walked around 10 blocks in central Stockholm we found 5 to 15 networks.
And so on…
Many of the networks we found gave us DHCP leases and good routing out to the internet. Most of them were behind a firewall, but the firewall was “aimed” in the wrong direction; the WaveLAN was a part of the internal network. We were inside private networks of telcos, law firms, investment companies, consulting companies, you name it.
From: “David Gamey/Markham/IBM” <dgameyca.ibm.com>
Subject: SSL Caching device?
I recently came across a device that appears to cache SSL! It appears that it can cache pages containing personalized data. I haven’t got the full story, but I suspect that the HTTP request didn’t contain distinguishing data other than an authentication cookie.
The press release:
It appears that the device works with a layer 3/4 switch and can transparently grab SSL connections (by port or packet content?). The marketing piece tries to position it as (or like) an SSL accelerator. It talks about graphics in SSL, being deployed on the network boundary and being transparent to the end-user. It’s setting itself up as man-in-the-middle.
From: Greg Guerin <glguerinamug.org>
Subject: DCMA Anti-Circumvention
The Digital Millennium Copyright Act (DMCA) prohibits certain acts of circumvention, among other things. In particular, section 1201(a)(1) begins: “No person shall circumvent a technological measure that effectively controls access to a work protected under this title.”
Look at the word “effectively.” Does it mean that the technological measure must be effective in order to qualify under section 1201(a)(1)? That is, if the measure is shown to be ineffective in controlling access, a mere tissue-paper lock, does that measure then cease to be a protected technological measure? But that means that any defeatable measure will lose its legal protection just by being defeated. And if that’s not an incentive to circumvent, I don’t know what is.
Or perhaps “effectively” means “is intended to”, and only the INTENT of protecting the work matters, not the demonstrated strength or quality of the measure itself. In short, well-intentioned incompetence is a sufficient defense. But then, arguably, Java byte-codes in original unobfuscated form might qualify as an access control measure, since they are not easily readable by humans and require an “anti-circumvention technology” known as a disassembler or decompiler in order to be perceived by humans.
So what does “effectively” really mean under section 1201(a)(1)? Upon such fine points do great lawsuits hinge.
CRYPTO-GRAM is a free monthly newsletter providing summaries, analyses, insights, and commentaries on computer security and cryptography.
To subscribe, visit <http://www.schneier.com/crypto-gram.html> or send a blank message to firstname.lastname@example.org. To unsubscribe, visit <http://www.schneier.com/crypto-gram-faq.html>. Back issues are available on <http://www.counterpane.com>.
Please feel free to forward CRYPTO-GRAM to colleagues and friends who will find it valuable. Permission is granted to reprint CRYPTO-GRAM, as long as it is reprinted in its entirety.
CRYPTO-GRAM is written by Bruce Schneier. Schneier is founder and CTO of Counterpane Internet Security Inc., the author of “Applied Cryptography,” and an inventor of the Blowfish, Twofish, and Yarrow algorithms. He served on the board of the International Association for Cryptologic Research, EPIC, and VTW. He is a frequent writer and lecturer on computer security and cryptography.
Counterpane Internet Security, Inc. is a venture-funded company bringing innovative managed security solutions to the enterprise.