The crypto whiz

Cryptography Research President Paul Kocher sees a long, hard slog between the good guys and the bad guys--and how it plays out will surely affect you.

Michael Kanellos
Michael Kanellos Staff Writer, CNET News.com
Michael Kanellos is editor at large at CNET News.com, where he covers hardware, research and development, start-ups and the tech industry overseas.
6 min read
Paul Kocher, president and chief scientist of Cryptography Research, came to prominence in the industry by breaking things.

In 1998, the company cracked security on smart cards by monitoring how much power their internal microprocessors used. Kocher also came up with the software inside Deep Crack, a machine tailored to crack encrypted documents.

Get Up to Speed on...
Enterprise security
Get the latest headlines and
company-specific news in our
expanded GUTS section.

Of course, he also fixes things. In the last few years, Kocher has emerged as one of the key technologists for financial companies and studios that are hoping to protect their intellectual property. He recently sat down with CNET News.com to discuss the ongoing melodramas surrounding privacy, piracy and stolen information.

Q: What is the top agenda issue for cryptography?
A: Let me tell you what it is not. The one thing that is stable--and really, nobody should be spending too much time worrying about--is which algorithms to use and what key sizes to use. Those are simple problems.

The huge challenge, from a technical perspective, is handling complexity, because we are getting systems that are just more and more complicated, and nobody knows how to get the bugs out.

The software side or hardware?

Every legacy feature is a potential exposure.
Software, hardware--everything. You pick it, and it is a lot more complicated today than it used to be, whether it is your network, whether it is your individual PC, whether it is a device of some kind, whether it is your microprocessor. Nobody ever removes features; they only add them--and from a security perspective, every legacy feature is a potential exposure.

If you have one component that you understand really well, it is pretty easy to get your hands around your one simple piece. But then you start having 600 components that all talk to each other. Not only do you have 600 times as many components to worry about, you have to worry about all of the interactions between these things.

So you have now got 360,000 different interactions. This is just horrible, because one person can no longer understand it; one person can no longer even begin to debug it. So, then you try to assign groups of people to individual pieces of the problem. But a lot of people staring at different angles of the elephant often will miss the big picture.

In order to just handle this technical problem, what we often try to do is first simplify things. If you look at some of those things that we design, once you get your mind on what it is doing, it seems simple, compared to a lot of other things. That way, we can be more confident that we have not missed something.

Can you give us an overview of what Cryptography Research does?
Typically, our goal is to bring new technical approaches to solving really hard security problems. When you are dealing with any kind of new technology, if it backfires, there is a substantial risk. The ones that we have had the most success with have been with the security challenges of financial institutions like credit card organizations. Another area we are focusing on increasingly is piracy. We also do a lot of work with infrastructure wireless systems.

Most of our revenues are from technology licensing, but most of our time goes into services.

How bad is the privacy situation getting?
Privacy is going to become a bigger and bigger problem over time, because sensors and data collection capabilities are improving along with Moore's Law. People collect data but do not have any plan of how they are going to get rid of it or what they are going to do with it, and so you end up aggregating vast quantities of data. It is a huge privacy risk.

I can now record as much audio as I will ever experience in my entire life, and video will be there in just a few years. The chips to do location tracking are getting smaller and smaller. There is one in my cell phone. Anybody who knows what they are doing can know where I am. There is this notion that information is bad in aggregate--but good in the cases where you need it. This is something that is very alien to a lot of people, and I am not sure how to solve it.

Piracy continues to be a huge, hot potato, with the studios blaming the device makers and the hardware makers trying to put responsibility on the studios.

How will this get resolved?
The studios are rightly upset that these companies are not spending as much money as they should to solve their security problems. But is it my job to keep your house from getting broken into? The way that I believe that it should work instead is that the studio should put some security code on the disk, and the player should run it.

The technical impediments to piracy that's based on copying and storing the data are going to go away.
The studios have a pretty powerful incentive to protect these materials, so how come this system isn't in place now?
It turns out that there are some very complicated technical problems in making this work. And fixing the problem from an economic perspective is not the way most engineers look at it. Most people look at security as this binary thing: Either it is insecure, or it is secure. If you take that kind of a perspective, this whole notion of apportioning risk does not even really apply.

One of the advantages our research group has is a lot of experience in working with credit card industries. The philosophy you learn there is really valuable, because there is this notion of risk. You can copy your average credit card with a piece of VCR tape and an iron. It is completely insecure technology, and you are always going to have fraud.

But what matters is not whether you have fraud; it is what your fraud rate is. So, Visa's published numbers are 0.07 percent and 0.08 percent. Overall, it is profitable for the different participants. If the fraud rates went up by a factor of 10, it would not be.

I think it has to be applied to other unsolvable problems like spam, like PC security, like piracy. Your goal here is to keep the rate of compromise low but to recognize that you cannot get rid of piracy completely or get rid of spam completely. But if piracy is below 1 percent of your revenues, it is the cost of doing business.

How is the notion of risk sitting with the entertainment field?
We have a guy in Japan who is meeting with CD companies. We've usually got somebody in Japan and somebody in LA. I am spending about half my time with studios right now. Nobody is saying anything publicly, but we have unofficial and strong support from much of the studios for what we are doing.

Some studios have one person whose job is piracy across the entire studio. Others have an actual group of people that are reasonably technical.

Who is more open to this concept--the music or the movie studios?
I think that with the movie industry in particular, there is going to be this sudden and catastrophic point in time, where it becomes more convenient or more economically advantageous for your average person to pirate a movie instead of obtaining it legitimately. The music industry has sort of crossed a threshold already. They are really getting hammered by piracy.

With movies, the only big difference is that you have a lot more data, which takes time to download. Instead of having a couple megabytes, you have a couple gigabytes. But Moore's Law clearly shows that is going to change, and when that changes, piracy rates are going to go up dramatically. Hard disks double about every 12 months. You will be able to put every major Hollywood release ever onto an $80 hard disk in high definition, and if it is not 2013, it will be 2015. The technical impediments to piracy based on copying and storing the data are going to go away.

How did you get into cryptography, anyway?
Well, you grow up in Oregon, and you have no driver's license, and you have a PC in your house--that's part of it. I went to Stanford and studied biology, so I cannot really credit my formal education with anything, but while I was there, I worked part-time for Martin Hellman (co-inventor of Public Key Cryptography). When I graduated, Hellman retired the same year and sent consulting projects my way.

Also, the neat thing about cryptography is that almost any aspect of society you pick has some connection to it. When you look at the government and espionage and military issues, personal liberties to voting, it is very hard to find any issue that does not have some cryptography angle to it.