Few people in the world know more about the future of computing than Paul M. Horn, senior vice president and director of IBM Research.
As the top manager of IBM’s research throughout the world, Horn is focusing the efforts of more than 2,900 IBM researchers, including five Nobel laureates, on defining technologies for the early 21st Century. He directs eight labs that have helped IBM produce the most U.S. patents six years in a row.
Horn has overseen the IBM labs during the development of numerous pivotal technologies. The most important developments in recent years have included the creation of the first computer chip to use copper (instead of aluminum) to conduct current; the first encryption system mathematically proven to be unbreakable; and huge advances in magnetic storage density.
Horn, 52, was a physics professor at the University of Chicago before joining IBM in 1979.
Here is an edited transcript of Horn’s conversation with the Mercury News:
Q–IBM Research has two basic thrusts, one on what you call “pervasive computing,” where you move computing power off the desktop and into the day-to-day world, and the other on what you call “deep computing.” What’s that?
A–Imagine you have this data coming in from all over — people are doing transactions, they’re getting information, they’re creating information — all of which is data. It’s huge amounts of data on-line. It’s so massive that it can saturate a system. Certainly the human brain can’t just take in all the data and use it sensibly. You need a computer to filter through all the information and get the really key important nuggets of gold that are in there.
That’s part of a concept that we call deep computing. It’s doing stuff with computers that you were never able to do before. It’s all the way from large-scale problems like simulating nuclear bombs without blowing them off to things like culling information out of a sea of random data that’s important and significant.
In deep computing the computer is doing things that you would have traditionally had to do physically. Think of digital things replacing physical things.
Q–Can you give me some examples?
A–The Boeing 777 is a good example. It was designed and virtually built on a computer before it went into production. This extends to various other things, such as weather forecasting.
We include very complicated business problems. How do you optimize a whole company? You could simulate a whole company and everything it does in a virtual world and then optimize the processes that are part of that company: the manufacturing processes, the whole fulfillment process, everything that’s got to do with how the company operates, you could optimize the whole thing on a computer.
Q–Tell me more about pervasive computing and how we’re going to see this in our lives going forward as much as two decades.
A–Two decades from now the local devices will probably store anywhere from 10 to 100 terabytes (10,000 to 100,000 gigabytes) of data.
Here are the numbers: In 1954, it cost $10,000 a megabyte to store data. It’s gone down now to 10 cents or less a megabyte. As things continue, we’ll be on the order of $1 a terabyte. I could be off by maybe $10 a terabyte, but it isn’t a lot.
Q–How will this affect hand-held devices?
A–When you go into a store, both your device and the store’s device will know the kinds of things that you are really interested in. Your system will know all about you and what your preferences are and your interests.
And it will communicate to other devices whatever you think is appropriate to communicate. It’ll know when you’re low on groceries. Of course, your refrigerator will know that. When I say the refrigerator will know when it’s low on milk, it sounds science fiction-y, but it’s not far away.
Q–People worry about the Big Brother effect of all of this. Is that a concern?
More Top Picks Best 75 Inch Tvs
A–To get the benefits out of the technology, the systems have got to know you as well as any person in your life would know you. And that means that you’ve got to trust that they won’t be revealing information about you that you don’t want them to reveal. It’s sort of like confiding in another person. You’ve got to trust the person when you confide in them that whatever it is you’re telling them is confidential. We’ll have cryptographic algorithms which are provable (unbreakable), standards that will let you screen what comes in and what goes out. But in the end you’ve got to believe that the right people are using them in the right ways. That’s a question of standards, and it’s a question of trust.