Oral-History:John Cioffi

From ETHW

About John Cioffi[edit | edit source]

Cioffi.jpg

Considered by many in the field as the “father of DSL,” Dr. John Cioffi participated significantly and tirelessly in inventing, supporting and commercializing the DSL technology used throughout the world. He developed the first asymmetric DSL (ADSL) and very high bit rate DSL (VDSL) modems, whose designs account for approximately 98% of the over 300 million DSL connections in use today.

Dr. Cioffi began his mission of creating DSL technology, which uses the copper wires already in telephone lines, during the 1980s at a time when industry thought optical fiber should be the focus. Dr. Cioffi and his students at Stanford University developed discrete multitone modulation (DMT), which enables ADSL technology to operate near the theoretical channel capacity of the telephone line. Dr. Cioffi then founded Amati Communications to commercialize his technology. Behind his leadership at Amati, the American National Standards Institute (ANSI) chose DMT technology as the U.S. standard for DSL in 1993. Now, all worldwide DSL standards are exclusively based on DMT technology.

Dr. Cioffi continues to support DSL development through research at Stanford University and at ASSIA Inc., a company he founded in 2003 and in which many major DSL service providers have invested and/or purchased ASSIA products. His focus is on dynamic spectrum management (DSM) to improve performance in multiuser DSL and wireless transmission channels. An IEEE Fellow, Dr. Cioffi is the Hitachi America Professor Emeritus of Electrical Engineering at Stanford University, Calif., and also the chairman and chief executive officer at ASSIA Inc., Redwood City, Calif. Cioffi was the recipient of the 2010 IEEE Alexander Graham Bell Medal and the 2014 IEEE Leon K. Kirchmayer Graduate Teaching Award.

About the Interview[edit | edit source]

JOHN CIOFFI: An interview Conducted by Michael Geselowitz, IEEE History Center, 24 June 2021

Interview #858 for the IEEE History Center, The Institute of Electrical and Electronics Engineers, Inc.

Copyright Statement[edit | edit source]

This manuscript is being made available for research purposes only. All literary rights in the manuscript, including the right to publish, are reserved to the IEEE History Center. No part of the manuscript may be quoted for publication without the written permission of the Director of IEEE History Center.

Request for permission to quote for publication should be addressed to the IEEE History Center Oral History Program, IEEE History Center, 445 Hoes Lane, Piscataway, NJ 08854 USA or ieee-history@ieee.org. It should include identification of the specific passages to be quoted, anticipated use of the passages, and identification of the user.

It is recommended that this oral history be cited as follows:

John Cioffi, an oral history conducted in 2021 by Michael Geselowitz, IEEE History Center, Piscataway, NJ USA.

Interview[edit | edit source]

INTERVIEWEE: John Cioffi

INTERVIEWER: Michael Geselowitz

DATE: 24 June 2021

PLACE: Virtual

Geselowitz:

This is Mike Geselowitz with the IEEE History Center. And I am here doing an oral history interview with John Cioffi of Stanford University. This is a remote interview. I am sitting in New York City, and John, you are in…

Cioffi:

Atherton, California.

Geselowitz:

You are in Atherton.

Cioffi:

California.

Geselowitz:

Excellent. So John, I'd like you to start, tell me about your early years, your family background, and how you got interested in technology.

Cioffi:

Well as a kid, I loved mathematics and technology and science, like a lot of engineers do. And so I was naturally driven towards that. And some of the areas that were interesting to me were just people communicating in those days by the old landline phones…I’m dating myself a bit. But I recall going to a World's Fair in New York; I remember as a very young boy at that time. There I saw AT&T's video phone demonstration, which was of course terrible quality; there were a lot of excuses and comments about why it didn't work so well, particularly that phone lines were limited. It intrigued me, even though I was I think six or seven years old. That video shouldn't really look that way. Of course, the phone line was not really limiting, as today we've done much better. That kind of got me thinking.

Later, I went to college at the University of Illinois; and I'm from the Chicago area originally. If you were good in math and science, that's probably where you wound up in that area, at U of I as they called it. It happened to be a very good public engineering school, and was in my home state. There, I heard about the old Bell Telephone Laboratories. I was reading in my classes about all the famous scientists and their contributions. These achievements were very interesting to me, and I was very fortunate to work there upon graduation. That was my first job. I entered some of the then-Bell Systems programs, which were kind of work-study programs. You could get a master's degree and even a Ph.D., which eventually I did, under those programs. You had to work some, you had to go to school fulltime but as an employee, then come back and work fulltime, then again go to school fulltime. And so that happened over a period of about six years. I learned how to design communication systems at Bell Labs and as well had the opportunity to go to Stanford as part of those programs. And it worked pretty well for me, and it really set my career interest, which has been in the communications area. I love that area, making things work better, whether it's wired or wireless, particularly the physical connectivity, and I have taught the courses at Stanford for decades in that area. I have worked in industry as well beyond Bell Labs. I worked at IBM, I started a couple of companies one of which I'm still active in, and I've served on many boards of directors.

Geselowitz:

I'd like to get back to that part of the story, John. But first, two quick questions. First, when you got to Bell Labs, did you make the connection that that was the place when you were a small child that had designed the video phone that AT&T tried to show off at the World's Fair when you were a kid? Did you make that connection, like, wow, here I am…?

Cioffi:

I did. In fact, I specifically asked to be in that group that did what were called voiceband modems. And actually in the course of the educational programs at Stanford, I met some of the professors who also had common interests. These professors liked me, and they knew the people who worked in that group. They called that area and said "Hey, you should move this guy over to that group, which is what happened." And so I learned from some of their best transmission people, over time; their perspectives were a little bit more industrial, more practical than the university side. But I had the best of both worlds really in the early part of my career.

Geselowitz:

Great. And my second question was, who decided on Stanford? Was that you, or did Bell Labs recommend that that's where you go in this work study program?

Cioffi:

Bell Labs recommend that you switch schools for broadening of perspective. And there was a U of I professor who kind of took me under wing. I was kind out of cash to finish school. He gave me a job designing filters, Prof Mac Van Valkenburg; he's dead now. Back then, he eventually became U of I 's Dean of Engineering. He just kind of found me, and he had gone to Stanford. And so I had this almost paternalistic relationship where he was the father, I was more the kid, and he loved me and gave me special jobs to do and other things as an undergraduate student. And so I think that, plus it was 40 degrees below zero the day I decided between MIT and Stanford, on this program. I heard there was no snow in California near Stanford, and I was ready to go.

Geselowitz:

So who did you study with at Stanford?

Cioffi:

My Ph.D. advisor was Professor Thomas Kailath, who's been another great influence on me throughout my career. I also got a lot of help from Prof . Bernie Widrow over the years, and Prof Marty Hellman is a close personal friend who helped guide my faculty career. Marty wasn't my advisor, but I probably have as much to do with Marty these days as anyone on the Stanford faculty.

Geselowitz:

And what was your dissertation on?

Cioffi:

It was on some fast-learning algorithms, basically, and their potential application to communications and transmission, a subject that's very hot today. But that was the title of my thesis actually back in the mid-1980s.

Geselowitz:

You were ahead of your time, okay. So now…

Cioffi:

Well I think the things I had in my thesis probably wouldn't work too well anyway. But I was learning and thinking. I think I probably have a better way to do that today. But nonetheless, that was the subject area.

Geselowitz:

And then so when you got your Ph.D., you then decided to leave Bell Labs and go to IBM. So how did that work out?

Cioffi:

That was the divestiture of the old Bell system. And it was an interesting time. And they were sending the group where I worked into something called American Bell, which nobody remembers, but that company was going to do a personal computer; a PC that competed with the IBM PC. And I wasn't sure that that was really a company was going to survive. I did seek a job in the wireless area; and believe it or not, I proposed MIMO (multi-input multi-output) to Bell Labs in 1984. As a function of my dissertation at Stanford, I was watching some other students who were working on lots of antennas for defense applications. And so I proposed the multiple antenna concept to the Bell Labs people. And the wireless people there loved it. This is like ten years before they really got serious about it. But what happened was as they broke up the Bell system, AT&T was deemphasizing wireless, believe it or not. And they were making all those people switch into either fiber projects or other types of projects because they didn't believe there was a future in wireless communications. And so, the job didn't exist. The guy I was going to work for, it was hard on him. I won't say his name, because it was a difficult time. He lost his job. He'd got divorced. He moved to Europe. But he was instrumental in bringing the GSM effort together in Europe, which eventually led to drive wireless everywhere. So I would've been working in that group on multiple antennas, but it didn't exist. So, I instead joined IBM for a couple of years and worked on disk drives, which was kind of a different area, or so you would think. But a disk is kind of like a communication system in disguise where transmit and receive are the same as write and read. And that disk, the spinning disk, has got all kinds of challenges as a communications channel. So did that for a few years. And then a Stanford faculty position opened, and been on the Stanford faculty since then.

Geselowitz:

So you mentioned earlier that when you got to Stanford, you taught a lot of communications courses, I guess at the graduate and undergraduate level?

Cioffi:

The graduate program in electrical engineering at Stanford is huge. The undergraduate program is very small. So I did teach the basic digital signal processing class a few times to undergraduates. But almost all the classes at Stanford in electrical engineering, unlike most universities, are in the graduate program. So I had a series of four classes that would start with kind of the introductory master students and go all the way to the very advanced Ph.D. students in the area of digital communications at Stanford I was the key developer and instructor for those classes over the decades. And one of them I just taught last quarter. I don’t teach them as often, as emeritus now, but we still do have them occasionally.

Geselowitz:

Oh, wow. So how did you get specifically interested in DSL (Digital Subscriber Lines)?

Cioffi:

It was actually a guy named Joe Lechileider at Bellcore, who got me pumped about DSL. I had heard about DSK when I was at Bell Labs. They didn't call it DSL then; they called it another name back when I first joined Bell Laboratories. And I was in some early DSK meetings. Indeed, I then suggested that to do video, Bell Labs needed to run faster than what became ISDN (early DSL). All these targeted going digital in the last segment all the way to the home. That started as the ISDN concept. I suggested Bell Labs should attempt to go ten times faster. Why don't we try that? And I was kind of young, so they told me to shut up. But Joe believed in that. He wasn't at that particular meeting where they suggested I listen and not speak. For some reason, he missed that one. However, I meet him a few years later at an IEEE conference in Japan. He spoke about this higher-speed DSL subject, indeed right after me in one of the sessions. So I introduced myself, and we wound up going to dinner. Many call him the inventor of the DSL concept, and I think that's fair. I get credit for being Father of DSL nonetheless (I did not create that name). So Joe is the Uncle of DSL. Joe and I had a joke between the two of us until he passed: "I'm father, he's uncle on DSL." I'd take a call from him and say "Hello, Uncle Joe," and he'd retort "Hello Dad" or something like that (he was older than me). He died a few years ago. His kids contacted me and I provided some text for his epitaph; I really liked Joe. He was enthusiastic and encouraged me to do DSL, and even gave some money to my Stanford group. I had been named a Presidential Young Investigator, which was a big deal back in those days. Ronald Reagan was president. I still got the certificate in my office at Stanford signed by Ronald Reagan. And they gave you significant amount of money, and you could go match it and get more. And Joe was one of the contributors to that matching program through Bellcore, and we basically started looking at the phone lines and how could we really do DSL well. It turned out to be a good project, and it was kind of funny because the NSF people announced this as fiber research. Everybody at that time had to call things fiber, right. So they put it under the fiber project, but nonetheless my Stanford group was funded to do this. And we came up with the designs that are now used everywhere on all the DSLs in the world really, from the slowest ones in the early days to the fastest ones today. They still use the same basic concepts that were developed at that time to try to squeeze the last bits per second out of a telephone line, which is pretty difficult digital channel. So there's a lot of artificial intelligence in these DSLs, even the earliest ADSL(Asymmetric DSL) modems have an extreme amount of artificial intelligence in them. They're learning constantly from the phone line as it changes and adapting. There's some pretty complex concepts that go on inside there, so the DSL system can keep running at the highest possible speed no matter what the phone line looks like or how it changes with time. So it was a challenging project, but it worked. Prototypes were developed, and they did exceptionally well in the tests that were run, so the world standardized on that design, basically.

Geselowitz:

As I recall in true Stanford Silicon Valley style, you then took a leave to do a startup around the concept. Is that correct?

Cioffi:

That's correct. For the Stanford people in those days, it was still early with startups spinning out. They actually had a program that was funded by some venture capitalists, but it was part of the Stanford infrastructure where you could apply and get some money. And the idea was to try to do tech transfer out. And so that company (Amati) started as a Stanford contract with this venture capital group. But it morphed into an independent company over a period of a couple of years. Then, I took the leave that you're talking about and was there for a couple of years, three years maybe fulltime. I stretched out all the leave I could use. Special thanks here go to the students who joined that company and helped enormously, IEEE Fellow Dr. Peter Chow, Dr. Jacky Chow, Dr. Jim Aslanis, and Dr. Krista Jacobsen -- along with some of the fantastic engineers who worked there, including IEEE Fellow John Bingham, Mark Flowers, and Mark Mallory. That company did the designs in the early chips for DSL. We also spent a lot of time on the standards - - trying to help define the best way we knew possible to make it work as well as possible. And those designs are pretty solid. They're still in use today everywhere. So the basics, I think we probably got right at that time. Amati was named after an Italian violin maker. I didn't name the company; somebody else did. But they thought, "John has an Italian surname, and I was walking around how we need to teach everybody to do this well. Amati was the violin maker who taught Stradivarius, the most famous violin maker. So teacher ... Amati." Amati empowered most of the companies that did well in DSL. The largest DSL supplier in the world today is Nokia, but it was then the Alcatel Group. Alcatel sent many employees to Amati in California. I trained them for a period of months basically. Some of their key managers were part of that program a long time ago. Some went to other companies as well (like Broadcom). Texas Instruments was also one of them. And Amati worked with them as it went public. Amati was not profitable. But Amati was acquired for a large amount by Texas Instruments at that time, and then I kind of slid back into Stanford fulltime.

Geselowitz:

So another pioneering effort to go public and sell it at great cost, a company that's not making a profit, because that's very in today.

Cioffi:

Yeah, that's what happened. Basically my company today that I'm in charge of is profitable. It works in DSLs and Wi-Fi. It doesn't get the same attention that Amati got, at least not yet. It probably should get a whole lot more because it's done better. But at in Amati's time, there was a need for that type of thing. Amati was the number-two growth stock on NASDAQ one year. It's price was going up, and there's interest, and there's bidders, and Texas Instruments put the best deal on the table and bought the company.

Geselowitz:

Wow. So I want to get back to your current work that you just alluded to in your current company in a minute. But this is a really excellent time, I think, for me to segue to one of my prepared questions, which is when did you first become aware of and involved with IEEE, in that arc that we've just covered, up to founding of your first company?

Cioffi:

As a student at the University of Illinois, I think it was maybe sophomore year, I heard about the IEEE. Somebody teaching a class encouraged the students to become members. And so I became a student member and have been a member ever since.

Geselowitz:

And were you active in any way? I know you’ve published publications in IEEE transactions…

Cioffi:

Yes.

Geselowitz:

But were you involved in any other way?

Cioffi:

I have had a number of positions in IEEE. I was an editor for one of the very technical magazines for a five-year period. I did a couple of special issues. My secretary at Stanford told me if I ever agreed to be an editor for a journal again, she was quitting. So that was the last IEEE journal for which I served as an editor, because she was doing all the work. But I did that. I've been a part of various conference committees and other things over the years. I can't really remember them all. And I've been a keynote speaker for, I don't know, more conferences that I can count for the IEEE over the years as well. So it's a good organization. I encourage students at Stanford to join it. I encourage employees who work for me to be members of it. And I try to partake as much as I can in their activities. There are so many of them that it's hard to even keep track of them all, much less participate, but do try to attend conferences when I can and participate in all. So, I think it's a great organization, it's grown a lot since I first joined it. And you know, it is basically the global standard for electrical engineers at least, at this point. And perhaps only the ACM is, for computer scientists, is a bigger, more formidable technical organization than IEEE.

Geselowitz:

So when you started a company that was pioneering a new area of communications, obviously standards is going to be an issue.

Cioffi:

Yes.

Geselowitz:

Did you get involved at all with the IEEE standards?

Cioffi:

A little bit with the IEEE standards. The IEEE standards are more for data communications, and they are in the area of the 802 committee as the local area network standard. Ethernet (802.3) and Wi-Fi (802.11) are the two big ones. Those two IEEE standards are an enormous success over the years, because those two standards are probably the most heavily used ways to get onto the Internet. Between Ethernet and Wi-Fi - you got it there. And very ubiquitous, very well-adopted and supported on a global scale, at least in terms of the inner-connectivity of the devices. Some other standard levels above that may be a little more challenging.

I was a little bit involved in those activities. The main standards groups in which I participated were the more nationalistic bodies like American National Standards Institute, and then later the global International Telecommunications Union, ITU, standards. The ITU is where everything becomes a big global standard. It is a very powerful organization. Much more politics than IEEE in terms of standards, both national politics as well as huge company politics. So those are the ones I participated in the most, I also participated somewhat in ETSI, which is the European Telecommunications Standards Institute. I'm still a member of the Broadband Forum and their activities, which actually spun out of the company Amati a long time ago. Actually we had a CEO who was departing Amati (Kim Maxwell). He had helped us the first couple of years, and we were growing. He was a marketing person, so that matched what was then called the ADSL Forum (but now Broadband Forum). He got that off the ground, just as he did with us, and then retired. The BBF is also a good standards organization that provides enormous benefit to the world.

However on the IEEE standards, Ethernet and Wi-Fi, those are, they just affect billions of people every day, and those are IEEE standards.

Geselowitz:

I heard a little rumble in the background, but the recording should be fine. But I was going to say about that, obviously now Wi-Fi much more than Ethernet. The other day I was on a Zoom call, and someone was having some trouble with the Wi-Fi or something. Someone said well, why don't you just try to plug the Ethernet directly into the router? An Ethernet cable. The person said I don't know if I have one. So Ethernet was huge, huge when it started, but really Wi-Fi is really driving the world right now.

Cioffi:

It is, and here in my home, my Wi-Fi is off. I am using the Ethernet. We do have a fiber connection to my home, and it runs fast, I think we're the only ones on the fiber, maybe one other home at present. So the Ethernet connection is I get is 900 megabits per second. My wife and daughters, anyone else in the house is using Wi-Fi, and they often will complain, right. And it's not the Internet, it's the Wi-Fi, right, because that is limiting certainly the bandwidth. And I knew enough to run an Ethernet cable to all the rooms in the house. I had a monster cable installed a long time ago. It cost us to do that. But so I take advantage of it, but they all want the mobility so use Wi-Fi. They're on their smartphones or tablets. They're walking around and talking to people, and I'm sitting at my desk most of the time, you know, working. So I'm okay on Ethernet to do it. And it is a better connection, obviously. And this is a big, one of the things my present company, ASSIA, works on is trying to make Wi-Fi better, because after the fiber and the copper getting to your home, the Wi-Fi is definitely the bottleneck. And that's getting worse. And so it's the next area of improving connectivity to everyone is to get Wi-Fi to be much better. And that's not an IEEE standards problem. That is a layer above that, where the management of the Wi-Fi systems and coordination and so forth is very fragmented. And so we're trying to address that area and make it more uniform. And there's a lot of challenges, many of them political. Vested interests of companies or countries and so forth also there also get in the way of that. But we try and advance incrementally.

Geselowitz:

So let's actually pick up that story. So you went back to Stanford. Ultimately Amati was sold to TI. And then you went back to Stanford. So then your research shifted how? What was that…?

Cioffi:

I still had a little bit of a foot out the door. I was on the boards of directors, several companies. The most successful was Marvell Semiconductor that I joined in 1999. I was also a Marvell investor, and essentially was the lead director, when I retired from that four or five years after they were public in 2006. I talked the Marvell CEO into entering first ethernet and then Wi-Fi. So I continued to kind of help with the chip areas. And Marvell still today is basically in Ethernet and Wi-Fi today. I served some other companies that didn't do as well as Marvell. And then I was teaching at Stanford the basic communications classes, and the graduate school had a lot of excellent students in those classes over the years. And we really built Stanford's program in digital com. A lot of people, even the MIT guys, would tell you it was the best in the world, for a while. It's probably not there anymore. We haven't put as much effort into it in recent years. So Stanford and I had a run there where we were taking the adaptive and artificial intelligence concepts to a new level, wireless and wire line systems.

The research tried to understand the interaction of all these things that occupy the same transmission medium at the same time in the same frequencies, maybe in the same place and space. And how do you put that together, and what do you do to make that work better, which of course is the big challenge today, Wi-Fi being important because about 80 to 90% of Internet traffic that's wireless is on Wi-Fi. (The rest is on mobile direct to your cellular device.) Wireless channels vary a little more often, because there's mobility, but also there's just this ubiquity concept that goes beyond the wireline or wireless subcategories. And wireless Wi-Fi is the natural extension once you get to the end of the wires. We were talking earlier about Wi-Fi; everybody's on it.

Making Wi-Fi better and making those systems more collaborative or respectful of one another is a big machine-learning problem. So how do you do that? Well a lot of research and concepts were developed in my Stanford research group then in that area, and in classes updated on it for about another good 10, 15 years, indeed to this spring when I taught the latest version of one of those classes. And during that time, my second founded company, ASSIA… I'm CEO/chairman of ASSIA today…spun out of Stanford to look into that area and try to help it. My wife is cofounder. Her first name is Assia, and the company name ASSIA is a "doing business as" acronym, which spells out Adaptive Spectrum and Signal Alignment.

AT&T (then SBC) people were very good and helpful also on ASSIA. They were an early investor, almost a cofounder, more than just a contract with us. I should call out Steve Sposato and Raj Savoor specifically for this. Today, AT&T owns some shares in the company as well. And we co-developed some products that were first used on their network and started taking them to the rest of the world. And that continues in a wired and wireless media today. And continue to try to make connectivity better. As we all know, it doesn't work all the time.

Geselowitz:

Can you put a label on? Is there a name for that concept?

Cioffi:

I have called it Dynamic Spectrum Management (DSM) from the beginning. You will hear it sometimes called dynamic spectrum allocation, dynamic line management, digital quality management. But they're all in this same area of, of trying to learn what's going on and make adjustments to the basic parameters used by the transmission systems so that they work better. Such systems are more compatible with one another; they're more respectful of one another in different ways, so that everyone can use our communications airwaves better than they could previously. The example I like to give, Michael, is today you and I are sitting different places, but we probably have 10 gigabits per second or more of wireless data going through our bodies right now. But yet we don't have access to it all. And we often still have problems, right? So how do we get to that 10 Gbps already there (or more) in the future? Are there ways to make policy more dynamic, more adjusting to the situation? How do we measure that? How do we know where our money should be spent for infrastructure? I can guarantee you that any politician is out there running around saying well, we're going to spend a lot of money on fiber, we're going to run it to everybody's home.

Well, first off, that doesn't solve the problem. And secondly, it costs five times more than your most expensive estimate to do that, and it'll take five times longer. And it's been 40 years been coming, and it's still not here. So good telecom infrastructure investment requires more intelligence and more, and better use of whatever funds are available, if there are infrastructure funds that are available from the government, such as for the new programs when they get through Congress. (Eventually they probably will.) How do you best spend that money so that you really do improve the experience? And just throwing a lot of money at the wrong thing does not necessarily fix a problem. It may look good from an appearance standpoint, but doesn't necessarily solve the problem. So how do you do that? And I'm very confident however that gets solved in the next 20 to 30 years, there's going to be an enormous amount of artificial intelligence that's used to try to regulate and control in randomly distributed manner of all of those, but help everything work better. And there's orders of magnitude improvement possible with that, with respect to what's being done today.

Geselowitz:

And so who are sort of the players? You mentioned national governments; is ITU still a big player in that?

Cioffi:

Well the large standards player in the cellular area, the ITU blesses the standards. They call it, I believe they call them IMT, that group. But it's really the 3GPP group that does cellular standards. For Wi-Fi, of course it's still the IEEE 802.11 groups. There are various forums that support those things, do certification testing. There's a Wi-Fi alliance, the Wireless Broadband Alliance. ORAN, ONF, there's a lot of them out there. And part of that today is, that's part of the problem is this huge, a number of competing different standards organizations. And the confluence of interests are enormous. There are a lot of service providers around the world, the big Internet service providers have kind of degraded their businesses by maybe not making the best decisions the last 10, 15 years; consequently, they have high debt, and they pass difficulty that onto their vendors. The result is an ecosystem that doesn't necessarily best support that future growth. On the other hand, there are also the Internet companies who are building data centers and gradually moving into the telecom space, but are feared in many countries as too powerful. Indeed these companies today are much bigger and more powerful than the ISPs, and there's contention there between those groups in terms of who has control of this massive, quadrillion-dollar-value Internet and how it's used. What's done in the future for that. So it's a big challenge, for sure. And you've got these lethargic, aging telecom companies on one end of it, but they kind of own infrastructure, although they're selling it off and leasing it back to reduce their debt. As a side comment, that asset is really the one thing that distinguishes them so it may be long-term questionable to sell it to reduce debt. Once they lose it, it's hard to see them competing well with all the innovative application companies, large or small. IEEE fits into that pretty well because Ethernet and Wi-Fi, they're for everybody. The ITU wireline standards and the 3GPP standards that presume licensed spectrum. Those require infrastructure investment. And it's more restrictive in terms of who can enter and what they can do.

Geselowitz:

So you mentioned that you became emeritus at some point. What does that do?

Cioffi:

Yes. Well I reached the age, I was fairly young as a professor at Stanford, and they had a formula of age plus service-type of thing where you become emeritus. I had so many outside interests, including ASSIA at the time. and One of the associate deans came to me and said you know John, you're not supposed to do all this stuff. And I said yes, I know, but what am I going to do? And he said well you know, we have this emeritus status, and maybe that's what you want to do. And so I've been emeritus, I'm 25% recalled, which gives me ability to teach, but I'm unpaid at Stanford, so I have no financial conflict of interest. And so it's a special status that allows me to do what I do. But so I still have an office at Stanford. I have a small group of researchers, smaller than it was in the past, that I work with, weekly basis. Plus I have the company which has over 100 employees and business in five continents, servicing about I don't know, 40 big ISPs, 100 million customers. ASSIA looking at every 15 minutes at each connection, and tries to help with the AI techniques I mentioned. We make the Internet work better. It is a difficult industry because of that aging of the service providers that's been occurring along the way. It doesn't have quite the sex appeal of the internet application companies, but its needed. In a way, we pave the road on which all the good things come. Everyone likes the road, but no one cares who built it - just that it works.

Geselowitz:

So anything you're working on right now that you want to share that's interesting.

Cioffi:

Yes, well I think that the next wave, my company of course, we're doing this practical-type of thing every day, and that's at a certain level. But if we look at really software-defined everything, virtualization of the networks globally, and making it more like a data center, so putting the computation at the edge, I really believe that that is best accomplished when you take what's going in software today all the way down to almost the analog level, just the D-to-A converters and maybe the RF modulator to the antenna remain. And everything else is done at the edge, all the signal processing, computing, you can share the compute power, you can do a lot more, you can coordinate better. You can do that with Wi-Fi, or you can do that with 5G cellular. And so what they are calling "bare metal switches" today, which means that the software is above that level in the separation of control and data planes, I think the data point should go down lower. More software above below - down to the converters. And that's something that excites me. I think there's huge opportunities. If that goes software in the future, it will open up flexibility. There is going to be all kinds of groups fighting against that or trying to stop it or trying to control it or take control. There will be a lot of politics and issues with it. But fundamentally and technically, there are some great things that we can do there.

Geselowitz:

Great. So that's fascinating, from your early beginnings at the World's Fair in New York to this. I should've asked at the time, did your family drive from Illinois? How did you get to New York from Illinois?

Cioffi:

Yeah, I didn't get on a plane, basically until Bell Labs interviewed me. There's no way we could afford a plane trip anywhere. So we drove. And I had grandparents who lived in New York, and they said hey, we have this World's Fair thing, it's not too far from where we live. Why don't we go over there one day? And we did. So that was when I saw the first videophone.

Geselowitz:

Great. So all the way from the infamous video phone at the World's Fair to DSM in the 21st century. It's been a really great story. Is there anything that you would like to add that we didn't cover, for the record, so to speak?

Cioffi:

Oh, there's always going to be something that you're mad at yourself later that you forgot to say, whatever that is but.

Geselowitz:

No. You do get, like I said, you get one chance. I'll send you the transcript, and you can, I said, cut out derogatory things, but you can also add a paragraph if you realize that you forgot to say something.

Cioffi:

Right. Well probably the best, the best thing is, the good Lord tells us to thank everyone. If you don't have anything more to say or if they've put you in front of a crowd and you don't have a script and you're supposed to talking, provide the entertainment, always start with thanking people. So thanks to the IEEE for helping my career along and along with the millions of others that I'm sure they've helped. Also all the students at Stanford who have done the dirty work on my crazy ideas over time and helped me out and contributed a lot of great ideas themselves to that. And the people in the universities and the companies who are more mentors or investors or worked with me. And, in particular my wife Assia, my six children (lucky number 7 with stepson) and 4 grandchildren and counting. So nobody has a career by themselves. We all get a lot of help from other people. So thank you to everyone.

Geselowitz:

Message that many people need to hear, thank you. So John, thank you very much for a great interview. I’m going to turn off the recording now.