Lee Rainie: Good after­noon, every­one. Thank you so much for com­ing. It’s an hon­or for me to be here. I’m Lee Rainie. I direct Internet sci­ence and tech­nol­o­gy research at the Pew Research Center, which is, we call our­selves a ​“fact tank” in Washington, DC. I have a con­nec­tion to Elon because I have a daugh­ter who went here and grad­u­at­ed in 2003. And I stay con­nect­ed with the com­mu­ni­ca­tions school because I love it, because they lis­ten to me when I give them advice, and because I have a spe­cial rela­tion­ship with the fac­ul­ty there, espe­cial­ly pro­fes­sor Janna Anderson. She and I have done work about the future— Yay for Janna!

We are now com­plet­ing our sev­enth round of inter­view­ing of experts. We began in 2004, ask­ing them about the future of the Internet. And we just com­plet­ed the gath­er­ing of our data for the sev­enth time this sum­mer, and we’ll be issu­ing reports lat­er on this year about some of the wor­ries that peo­ple have about the future of the Internet. 

But most­ly it’s my plea­sure here to be with a real hero of mine and a per­son who should be a hero to all of you, Dr. Vint Cerf. [applause]

Vint Cerf: Thank you.

Rainie: He and his friend Bob Kahn invent­ed the trans­fer pro­to­cols that essen­tial­ly made the Internet the Internet. And he’s spe­cial to me because he is the only inven­tor that I can ever think (and I’ve checked this on Google, so it must be true because he’s a Vice President at Google), he’s the only inven­tor of a major tech­nol­o­gy who has through the rest of his life­time been con­cerned about the impact of that tech­nol­o­gy and tried to design or redesign ways to make it bet­ter to serve human­i­ty. I lit­er­al­ly don’t know of any inven­tor who’s cared that much about his cre­ation in that way. They want to make more mon­ey often­times. They want to make a bet­ter light bulb some­times. But he has tried to make the world a bet­ter place con­tin­u­al­ly by think­ing about how the impact of his machin­ery. So we owe him those thanks, too.

And my final thanks to him are, when I went to the Pew Charitable Trust to get the grant that I did, I had to do some fact-finding about what we should do to study the impact of the Internet. And he took my phone call and sat with me for two hours in his office, unlike a lot of peo­ple, because I was a nobody who knew noth­ing in those days. And he sat with me and helped real­ly think about a research design for the ear­li­est years of my project start­ing in 2000.

And five things came out of that con­ver­sa­tion. Four pieces of imme­di­ate research that we did. You prob­a­bly don’t even remem­ber this. You encour­aged me to research the social impact of the Internet. So lit­er­al­ly there were ques­tions at that time about whether we were los­ing our human­i­ty when we were start­ing to com­mu­ni­cate virtually.

The sec­ond thing you encour­aged me to do was to study the impact of com­put­er virus­es on the world, because they were begin­ning to become a per­sis­tent part of the Internet. Nobody was doc­u­ment­ing them, and you said that would be a good thing to do.

Then you cared a lot about pri­va­cy, and what was going on. And you cared a lot about the impact that the Internet in health­care. And those are the first four pieces of research that we did.

You also had the germ of the idea about the Future of the Internet project. You said there’s so much around the hori­zon. I have my thoughts about it, but oth­er peo­ple have their thoughts. Why don’t you study that, too?

So, thank you. Because you’re real­ly the inspi­ra­tion for the com­mu­ni­ca­tion school’s Imagining the Internet pro­gram that sur­vives still, and will have a much big­ger, promi­nent role in the new com­mu­ni­ca­tions program.

Cerf: So, just to place this in time con­text, this would have been around 2000, is that right? 

Rainie: Yeah, 1999. July 1999.

Cerf: So this is just as the Dot Boom is peak­ing, which is one rea­son why this would be an inter­est­ing thing to study, where­as the Dot Bust came in April of 2000. But the Internet con­tin­ued to grow despite that. The bust was all about investors throw­ing mon­ey at Internet, at peo­ple who did­n’t real­ly have a busi­ness mod­el. It was just one of those ​“maybe some­thing will stick.” And so even though that bust hap­pened, the Internet con­tin­ued to grow because of its util­i­ty. And of course today, six­teen years lat­er, you see what you what you see now, which is that 50% of the world’s pop­u­la­tion [is] online.

Of course, as the Chief Internet Evangelist at Google, my job is to get the oth­er three and a half bil­lion peo­ple online. So Eric Schmidt, who’s the Executive Chairman of Google told me I was­n’t allowed to retire because I’m only half done.

Rainie: Nice. Let’s move right into per­haps the thorni­est ques­tion about the Internet’s impact, par­tic­u­lar­ly in the gen­er­a­tion to come. And it relates to the way that dig­i­tal material—robotics, arti­fi­cial intel­li­gence, and oth­er innovations—are going to mess with the cur­rent job sit­u­a­tion. Millions of peo­ple who cur­rent­ly are employed will see tech­nol­o­gy replace the func­tions that they now per­form for pay. There will be inno­va­tions that cre­ate new kinds of jobs. I know you’re think­ing a lot about this and have just pub­lished your thoughts on this. So, could you tell us why we should­n’t be worried?

Cerf: Well, I don’t think that I would tell you should­n’t be wor­ried. I think we should be thought­ful about this. First of all, if you go to a web page called i4j​.info (it’s Innovation For Jobs, effec­tive­ly, .info) there’s a book that David David Nordfors and I pub­lished a few months ago called Disrupting Unemployment. And it’s about this dif­fer­ence between work which is relat­ed to tasks, which is what we typ­i­cal­ly think of as work. But unfor­tu­nate­ly you can use robots to do a lot of tasks. We’re more inter­est­ed in try­ing to cap­i­tal­ize on peo­ple’s capa­bil­i­ties, to find work that they can do that ben­e­fits from their skills and their spe­cial indi­vid­ual prop­er­ties. And that’s a very dif­fer­ent thing than just try­ing to focus on reduc­ing the cost of tasks. So, think­ing about work in a dif­fer­ent sense than just pro­duc­tion or some­thing is an impor­tant thing.

We’ve already been through sev­er­al sit­u­a­tions where new tech­nolo­gies come along. The Industrial Revolution removed a large num­ber of jobs that had been done by hand, replaced them with machines. But the machines had to be built, the machines had to be oper­at­ed, the machines had to be main­tained. And the same is true in this online envi­ron­ment. The machines that we use that enhance not only our mus­cle pow­er but our brain pow­er also have to be invent­ed, main­tained, and operated.

The prob­lem may be, though, that for peo­ple whose jobs depend­ed on their per­son­al abil­i­ty to do so some­thing, those jobs may go away. And the ques­tion is what will they do? Some of them will suc­ceed in being retrained. And this is in a sense—in this set­ting here at Elon—this is a very impor­tant obser­va­tion, that edu­ca­tion and work are going to be more entan­gled than they ever have been.

Historically, we went to school, we got a job (maybe get sev­er­al jobs), and then we retired. That was it. And in the­o­ry, the edu­ca­tion that we got at the begin­ning of our lives is sup­posed to serve us for the rest of our lives. Well, there are peo­ple being born today who will prob­a­bly live to a hun­dred years old. They will almost cer­tain­ly have mul­ti­ple work that they can do. And they almost cer­tain­ly will have to learn some­thing dur­ing the course of their adult lives in order to con­tin­ue doing use­ful work.

The impli­ca­tion of that is that this old term, con­tin­u­ous learn­ing and con­tin­u­ous edu­ca­tion, is real. It’s not just a slo­gan. And we have to fig­ure out how to deliv­er that prod­uct. So, we’re sit­ting here on a cam­pus whose pur­pose is to edu­cate. I think that uni­ver­si­ties like this one and oth­ers are going to have to learn how to pack­age their prod­ucts up in order to deliv­er it when peo­ple need it at var­i­ous times in their lives. It’s a very very dif­fer­ent mod­el than the one we’ve had in the past. But part of it has to do longevi­ty. Part of it has to do with the rate of change of tech­nol­o­gy and the need to learn new things.

I can tell you at Google, where I’m sur­round­ed by peo­ple who are a lot younger than I am, I have to learn new things all the time because they keep invent­ing new things to learn. And in some cas­es, I have to rethink what I thought I knew. So a typ­i­cal sce­nario is some­body run­ning up say­ing, ​“Why don’t we do X?” for some val­ue of X. And I’ll think oh, we tried that twenty-five years ago and it did­n’t work. 

Then I have to remind myself there’s a rea­son it did­n’t work twenty-five years ago, and it could be that rea­son is no longer valid. Computers are faster, they’re small­er, they use less pow­er, they’re less cost­ly. And so some of these crazy ideas turn out to be fea­si­ble. I’ve been forced to rethink a lot of things in the course of my almost eleven years at Google, as you will in the course of your life­times. So I am not despair­ing at all. Basically, I’m an opti­mist any­way. For the most part, I think we will fig­ure out how to adapt to the need for new edu­ca­tion as peo­ple find their jobs shift­ing from where it is today to where it will be in the future. So that’s my long answer to a short question.

Rainie: Your col­league, the Chief Economist at Google Hal Varian made a lot of head­lines a cou­ple of years ago, say­ing the sex­i­est job of the future was going to be sta­tis­ti­cian. So there are a lot of stu­dents in this audi­ence who would like career advice from the great Vint Cerf. Are there par­tic­u­lar direc­tions that you would encour­age stu­dents to ori­ent their lives towards, if not spe­cif­ic jobs?

Cerf: Well, the par­tic­u­lar sta­tis­tics that Hal was refer­ring to was almost cer­tain­ly Bayesian math­e­mat­ics, because every­thing we do depends on that lit­tle equa­tion which in the 18^th cen­tu­ry was thought to be cute and use­less. And of course almost every­thing we do now makes heavy use Bayesian analysis.

Well, let me think for a minute. Some young peo­ple come and say, ​“What should I do? What should I study?” And I tell them go into astro­physics. And the rea­son for this is very sim­ple. A hun­dred years ago, we thought we knew every­thing there was to know about the uni­verse and we just had to mea­sure the con­stants more accu­rate­ly to make all of our pre­dic­tions work. Well, then Einstein comes along and blows up Newton. And then the quan­tum the­o­ry guys blow up Einstein. Then the string the­o­ry guys blow up the quan­tum the­o­ry guys. And then we dis­cov­er that 95% of the uni­verse is either dark mat­ter or dark ener­gy. We’ve giv­en labels to these things, we don’t know what it is. 

We know about 5% of the uni­verse. So if you go into astro­physics, the chances of get­ting the Nobel Prize are extreme­ly high because nobody knows any­thing. So any­thing you do is like­ly to win the Nobel Prize. 

If you decide you don’t want to do astro­physics, what to do next? And I think that the hottest top­ics today have to do with the appli­ca­tion of com­put­ing to oth­er dis­ci­plines. I was in Lindau, Germany ear­li­er this year, late June, meet­ing with thirty-five Nobel Prize-winning physi­cists and about 400 grad­u­ate stu­dents. And it was stun­ning to see how much the com­put­er had become a tool for the physi­cists. In fact, the chemists, the Nobel Prize-winning chemists of a cou­ple years ago were not wet chemists. They were mod­el­ing chem­i­cal inter­ac­tions with high-performance com­put­ing in order to pre­dict what the results were going to be. In every dis­ci­pline, in biol­o­gy, in med­i­cine in gen­er­al, we’re see­ing heavy use of com­put­ers in order to mod­el and ana­lyze things. I hate to use this ​“big data” term, but in a sense that’s what’s going on, large quan­ti­ties of data get­ting analyzed.

I hap­pen to have a par­tic­u­lar fond­ness for bio­elec­tron­ics because my wife has two cochlear implants, and so she’s sort of like the Bionic Woman. Her hear­ing was restored after fifty years of deaf­ness by an implant that goes into the cochlea, and the elec­trodes touch the audi­to­ry nerve and are fed by the implant. But sig­nals come through her through her head through a mag­net­ic con­nec­tion from a speech proces­sor that’s about the size of a mobile.

That thing is tak­ing sound in, it does a Fourier trans­form to fig­ure out which fre­quen­cies are present. It then decides which of the elec­trodes inside the cochlea it’s going to stim­u­late arti­fi­cial­ly. The brain inter­prets those sig­nals as sound. So she hears more or less nor­mal­ly. Doesn’t hear music very well because it’s a very crude sig­nal­ing sys­tem, but speech is great. So she uses the phone and lis­ten to books on tape. She car­ries an FM trans­mit­ter with her, so if she were in the audi­ence today I’d be wear­ing a lit­tle FM trans­mit­ter. She could hear me from 150 feet away.

So that’s bio­elec­tron­ics, and it’s get­ting more and more elab­o­rate. We’re see­ing opti­cal implants. We’re see­ing spinal implants. So if you were going to go into a space that would have a huge pos­i­tive impact for peo­ple with var­i­ous dis­abil­i­ties, bio­elec­tron­ics is not a bad place to be.

Studying phar­ma­col­o­gy, for exam­ple. Inventing new drug treat­ments based on mol­e­cules that you invent as opposed to dis­cov­er. All that stuff is there. It’s just… It’s a whole new world, wait­ing for peo­ple to explore.

Rainie: Let’s talk a lit­tle bit about the future of your baby, the Internet. Tomorrow night at mid­night, the for­mal con­nec­tion between the US gov­ern­ment and the ICANN address-based sys­tem of the Internet will be dis­con­nect­ed. Potentially, depend­ing on what American courts do. To whom are those func­tions being giv­en, and should we be wor­ried about that or take com­fort from that?

Cerf: Well, I think you should take com­fort. First of all, ICANN is the Internet Corporation for Assigned Names and Numbers. Its task is pri­mar­i­ly to man­age the IP address space (those are the numer­ic address­es of the Internet) and the Domain Name System, which you use all the time when you’re using the World Wide Web. It was set up in 1998 at the request of the White House. This was under pres­i­dent Clinton. Ira Magaziner pub­lished a green paper propos­ing the cre­ation of this orga­ni­za­tion. And even­tu­al­ly a white paper came out, and ICANN was established.

Why would they do that? Well, this busi­ness of man­ag­ing the Domain Name System and the IP address allo­ca­tion used to be done by one guy. He was a grad­u­ate stu­dent, and lat­er a mem­ber of the sci­en­tif­ic staff at USC Information Sciences Institute, Jon Postel. Way back in the late 1960s, he had this task. We called him the Numbers Czar. And he was just allo­cat­ing address space for the pre­de­ces­sor to the Internet, the ARPANET. And lat­er when the Internet was estab­lished, he did the same thing for the Internet.

ICANN’s job was basi­cal­ly to take the great pieces of the Internet address space and allo­cate them to what are called Regional Internet Registries, which in turn allo­cate IP address space to the Internet Service Providers. They also man­age the root zone of the Domain Name System. So, .net, .com, .org, and all those oth­er top-level domains exist in a file, and the file says if you’re look­ing for foo​.com, here’s where all the .com names are, this IP address and that computer. 

And the oth­er thing they do is to main­tain tables of para­me­ters that are need­ed for the Internet pro­to­cols gen­er­at­ed by the Internet Engineering Task Force, which does stan­dards for the Internet, joined now by the World Wide Web Consortium. So, their job was essen­tial­ly admin­is­tra­tive. They did­n’t tell peo­ple what to do with those address­es. They only made sure that IP address­es were allo­cat­ed unique­ly, so two peo­ple did­n’t get the same address. The same is true for the domain names. They made sure that the domain names were allo­cat­ed unique­ly to a par­ty that would man­age the next lev­el down.

The struc­ture of gov­er­nance of ICANN was mul­ti­stake­hold­er from the begin­ning. The stake­hold­ers includ­ed the the tech­ni­cal com­mu­ni­ty, typ­i­cal­ly the IETF and the World Wide web con­sor­tium. Which, IETF is part of the Internet Society’s oper­a­tion. So, they were the tech­ni­cal com­mu­ni­ty. We had the busi­ness com­mu­ni­ty. The pri­vate sec­tor was anoth­er stake­hold­er group. A third one was the intel­lec­tu­al prop­er­ty group. A fourth one was the gov­ern­ments of the world were invit­ed to par­tic­i­pate in what was called the Governmental Advisory Committee. They did­n’t make deci­sions but they influ­enced deci­sions because they fed advice to the board of ICANN.

So, there’s this mul­ti­stake­hold­er process which has been refined over the last eigh­teen years and has been refined even more over the last two years in antic­i­pa­tion of this ter­mi­na­tion of the con­tract between the Department of Commerce—specifically the National Telecommunications and Information Agency—and ICANN. 

What does that con­tract oblig­ate ICANN to do? Well, it oblig­ates it to man­age the top-level domain space, man­age the IP address space, man­age the tables of para­me­ters. The only thing that NTIA does is ask that ICANN tell them what changes they’re mak­ing to the root zone. Like, is .com point­ing to this IP address or that one? This has noth­ing to do with the pro­lif­er­a­tion of domain names that you and I use. This is only the thing that points to the serv­er that says this is where you find foo​.com or bar​.com and so on. have. It’s just what serv­er is that and what’s its address. That’s all that’s in the root zone. And the only thing and NTIA ever did was say, ​“Please con­firm to us that you used the agreed process­es to process changes that were request­ed by the oper­a­tors of the top-level domains.” That’s it. 

And the only thing they could say was ​“no.” They could­n’t say ​“take this out.” They could­n’t say ​“put this in.” The only thing they could say is, ​“We don’t agree that you fol­lowed the right process to make this change.” They nev­er, ever in the course of eigh­teen years said no. 

So their func­tion was to make sure that the process was fol­lowed. Well, now after two years of work design­ing a process in the absence of the NTIA com­po­nent, the same sort of thing will hap­pen. The same sort of mech­a­nisms will hap­pen. But the mul­ti­stake­hold­er com­mu­ni­ty will be the par­ty respon­si­ble for assur­ing that the process­es are prop­er­ly fol­lowed. And there are a whole bunch of mech­a­nisms that can be invoked if some­body thinks the process was not prop­er­ly executed.

That’s it. You won’t see a change at all. Nothing will change. And so the peo­ple who are object­ing to this mis­un­der­stand and mis­rep­re­sent the role of the US gov­ern­ment today in the oper­a­tion of the Internet. And so from my point of view, this is essen­tial­ly a NOOP. And it’s impor­tant to do it because there’s a lot of unhap­pi­ness around the world that the US has this spe­cial posi­tion which does­n’t actu­al­ly amount to much. And so it’s sym­bol­ic. And for coun­tries that feel like they should have an equal say, they find this spe­cial posi­tion offen­sive. When we remove that one spe­cial thing, every­body will be in the same boat. We’ll all be part of the mul­ti­stake­hold­er community.

And so from my point of view it’s the right polit­i­cal thing to do, and it’s the right tech­ni­cal thing to do. So, that’s a long answer again to a short ques­tion, but that’s where things stand. 

Unfortunately, on the stroke of mid­night on September 30^th, that tran­si­tion may still not hap­pen because four attor­neys gen­er­al in four states have sued ICANN, or I guess they’ve sued NTIA, to pro­hib­it the trans­fer from hap­pen­ing. The Congress was going to put a bill or a rid­er on the con­tin­u­ing res­o­lu­tion— Senator Cruz in par­tic­u­lar was in favor of that. And it failed. The Congress did not vote to incor­po­rate that block­age into their con­tin­u­ing res­o­lu­tion. So in this I guess last-ditch effort four of these attor­neys gen­er­al have decid­ed that they have stand­ing to object to the trans­fer. In my view, they don’t have stand­ing. There is no trans­fer of prop­er­ty or any­thing involved. But the judge in Texas will have to decide whether a restrain­ing order is jus­ti­fied or not. So we won’t know, frankly, until some time at mid­night whether or not the tran­si­tion actu­al­ly takes place.

Rainie: One of the last times you were in this com­mu­ni­ty, you gave a won­der­ful talk in Raleigh, a part of which you out­lined a dozen or more cen­tral ques­tions about the impact of the Internet that were at that point not yet well-understood, and which researchers like the schol­ars in this room might be well-served to study. What would be at the top of your list now for things that we don’t know about the Internet and its impact that you would like to see resolved?

Cerf: Well, I don’t know whether these are resolv­able, but they cer­tain­ly deserve atten­tion and study. So let me start out by observ­ing that the more Internet is pen­e­trant in our dai­ly lives, the more we depend on it and the more its brit­tle­ness affects us. So when it does­n’t work, lots of bad things hap­pen. This is true of almost all infra­struc­ture. Nobody pays any atten­tion to infra­struc­ture until it does­n’t work right. So, when the elec­tric­i­ty goes out, sud­den­ly you wor­ry about your ice cream melt­ing. But dur­ing the nor­mal course of events you don’t lie awake at three in the morn­ing wor­ried that the elec­tric­i­ty is going to go out. Unless you hap­pen to live in a place where it’s not reli­able. Or the road sys­tem, for exam­ple. You take the roads for grant­ed until they con­gest. Then you’re unhap­py about that.

So if the Internet does­n’t work right and we’re depend­ing on it, that’s a risk fac­tor. Think for a minute about this avalanche of things, this ​“Internet of Things” which is com­ing. You can see lit­tle bits and pieces of it have been around for a while, like you know, Internet-enabled pic­ture frames. But when you start design­ing your house around the notion of devices that have pro­grams in them that com­mu­ni­cate with each oth­er and some­how use the Internet in order to inter­act with you remote­ly from your mobile. You don’t want your house to stop work­ing just because the Internet isn’t acces­si­ble. That would not be a good design. 

So I wor­ry about the peo­ple who are invent­ing these var­i­ous devices and have this assump­tion that the Internet will be there so they don’t have to wor­ry about that part. Can you imag­ine going into your house with your mobile and about a hun­dred dif­fer­ent apps, because some­body was so stu­pid as to make each light bulb a dif­fer­ent app. And you’re try­ing to fig­ure out how to turn the lights off and on. But the mobile isn’t work­ing because its bat­tery is dead, or you don’t have a good sig­nal, or the WiFi is down. This is not sane. And so we real­ly have to think hard about build­ing robust sys­tems that take advan­tage of the pow­er of the net­work and the pow­er of pro­grammed devices, but at the same time pro­tect us from the down­side, which is when stuff does­n’t work right.

A third thing you men­tioned already, and that’s mal­ware, which is all over the place. Part of the rea­son that mal­ware is a prob­lem is not the under­ly­ing net­work itself, except for the fact that it trans­ports the mal­ware from one place to anoth­er. The rea­son mal­ware is so dam­ag­ing is that the oper­at­ing sys­tems that it attacks, the edge devices—your mobiles and lap­tops and servers—are the ones that are vul­ner­a­ble. So it’s not the Net that’s vul­ner­a­ble as much as it is the devices that use it for communication. 

And here we have to put the blame on the peo­ple who write the soft­ware. And the fin­ger points at me, too, because for years I made a liv­ing writ­ing soft­ware. I admit I don’t make a liv­ing writ­ing soft­ware any­more but I know carp about peo­ple who do. And in par­tic­u­lar, the prob­lem is that the soft­ware is not writ­ten with­out bugs. Frankly, for the last sev­en­ty years we’ve been try­ing to write soft­ware with no bugs, and we haven’t fig­ured out how to do that. 

So the result is we cre­ate soft­ware that has bugs. And we don’t notice it until too late. Somebody exploits the bug, some­thing bad hap­pens, then we try to fix the bug. And some­times that means down­load­ing new soft­ware into your lap­top or your mobile or your toast­er or some­thing else. Then you get into this real­ly inter­est­ing prob­lem. Okay, can I devel­op soft­ware sys­tems that will pre­vent me from writ­ing mis­takes in the code? And I have this… Imagine this arti­fi­cial­ly intel­li­gent thing kin­da sit­ting on my shoul­der look­ing at the code that I’m writ­ing. And it’s say­ing, ​“You you have a problem.”

“What do you mean I have a problem?”

It says, ​“Well, you have a buffer over­flow at line 27.”

“Are you sure?”

“Yeah, let me show you.”

And it puts up a piece of the code. I need this lit­tle thing sit­ting on my shoul­der to call my atten­tion to mis­takes I’ve made. And I don’t think it’s impos­si­ble to design some­thing kind of like that, that is a piece of soft­ware that’s pay­ing atten­tion to the code I’m writ­ing and is look­ing for mis­takes. For exam­ple, here’s a vari­able in the code, and you read it, you use the val­ue in that vari­able, except that nobody set it to any­thing. So you’re get­ting a ran­dom val­ue. Well that could branch you off into cloud-cuckoo land. 

We need to devel­op a much bet­ter abil­i­ty to write soft­ware that is large­ly bug-free. We need also to make sure that if there are devices that we’re rely­ing on that run with soft­ware, that they can be updat­ed at need. 

But now we get into anoth­er prob­lem. How does the device, let’s say it’s your refrig­er­a­tor, how does it know whether the soft­ware that’s being down­loaded is com­ing from a legit­i­mate source as opposed to some hack­er some­where? So now we get into this strong authen­ti­ca­tion question. 

This just keeps going. I mean, you can a pull on this thread and you will dis­cov­er a vari­ety of dif­fer­ent prob­lem areas that need atten­tion. Strong authen­ti­ca­tion implies good-quality cryp­to. Then you start wor­ry­ing about quan­tum com­put­ing. And you won­der whether the cur­rent public-key cryp­tosys­tems are going to stand up against quan­tum computation. 

In fact, just yes­ter­day I was at Stanford University, and we had a dis­cus­sion with about twenty-five dif­fer­ent cryp­tog­ra­phers about what comes after the quan­tum com­put­er that breaks all the cur­rent codes that are rely­ing on fac­tor­ing as the work fac­to­ry. And for­tu­nate­ly there’s good news. There are some math­e­mat­ics that don’t rely on this par­tic­u­lar prop­er­ty, which is how our cur­rent sys­tems work.

So, I won’t go on and on. But there is just a fab­u­lous array of con­cep­tu­al prob­lems that need atten­tion, that have real-world con­se­quences. And so I think that we all have an oppor­tu­ni­ty to tack­le those if we have the inter­est in it.

Rainie: Pretty relat­ed­ly, there’s… Any num­ber of your quite promi­nent col­leagues have begun to express very deep con­cerns about the future of arti­fi­cial intel­li­gence and whether it becomes essen­tial­ly a run­away or break­away process. I don’t know that you’ve offi­cial­ly endorsed any of that. Could you tell us what to wor­ry about, and where you think they might be over­step­ping log­i­cal concerns?

Cerf: Yeah. So, I’m going to talk a lit­tle bit about arti­fi­cial intel­li­gence lat­er today. But first of all when I hear peo­ple whose opin­ions I usu­al­ly respect, like Elon Musk for exam­ple, or Stephen Hawking, or Bill Gates, who are say­ing, ​“Ahhh! The arti­fi­cial intel­li­gence, the robots, are going to take over,” I’m a lit­tle sur­prised at this. 

The the­o­ry that’s being put for­ward is rough­ly that once a com­put­er learns how to do some­thing that a per­son can do, then it will learn how to do it bet­ter and bet­ter and bet­ter until final­ly it’s bet­ter than any­body. And the things that get cit­ed along those lines are for exam­ple win­ning Jeopardy!, or win­ning Go, which is a pret­ty impres­sive feat, as many of you may remem­ber back in February. I think we played five Go games using the AlphaGo arti­fi­cial intel­li­gence sys­tem that the Google com­pa­ny DeepMind put togeth­er and trained. And it won four out of five games against an inter­na­tion­al Go play­er. And then of course there was the chess games that were played back around 1997 or so with Kasparov, and he was defeat­ed as well.

The thing that you should appre­ci­ate is that the machines that do this have been very very nar­row­ly pro­grammed. This is not general-purpose intel­li­gence. This is not under­stand­ing that if you let go of a glass it’ll shat­ter on the floor because there’s this thing called grav­i­ty. This is about very very nar­row train­ing to get a machine to do a par­tic­u­lar thing. And often what it’s doing is either doing a rapid search through all pos­si­bil­i­ties faster than a human can, or it has been trained to build a neur­al net­work which after tens of mil­lions of games has adjust­ed its para­me­ters so it rec­og­nizes what to do. In some ways that’s what inter­na­tion­al grand­mas­ters do, too. They see pat­terns in the chess game and they rec­og­nize those pat­terns and they know rough­ly what they should be doing about defend­ing against the oppo­nen­t’s moves. 

This is not the same as the kind of intel­li­gence that you have. You have the abil­i­ty to induct from obser­va­tion. You have the abil­i­ty to rea­son, and you have the abil­i­ty to pre­dict con­se­quences. Most com­put­er pro­grams don’t have any­thing close to that except in very nar­row cas­es like the Bayesian analy­sis that I men­tioned earlier.

So I don’t see the threat here. If there’s any­thing in the future for us, it’s using these tools to aug­ment our own abil­i­ties. And I think this col­lab­o­ra­tive inter­ac­tion, which you use every sin­gle day. When you do a Google search, when you use Google Translate, you are using arti­fi­cial intel­li­gence. But it’s under your con­trol. And even with some of the more advanced things like Google Now and Google Assistant and so on, what’s going on is the machine try­ing to cap­ture infor­ma­tion that it things will be use­ful to you and then pro­vide it to you in a time­ly way.

But that’s not the same as some­thing tak­ing over your life. It’s just pro­vid­ing you with advice. So I don’t think you should wor­ry as much about that as you should wor­ry about bugs in the soft­ware that cause all kinds of trou­bles. Or mis­takes that peo­ple make. 

Rainie: Let’s look back­wards for a lit­tle bit at your baby and how it’s been used, all of the appli­ca­tions and dif­fer­ent lay­ers of the Internet. I won­der if you could just do a lit­tle quick march about your own dis­cov­er­ies of the way that peo­ple have used the Internet. Your first expe­ri­ences on the Web, your first emails, your first Skype calls, your first use of apps. Walk through the sort of Vint Cerf hall of delights as you’ve watched your baby grow up.

Cerf: Except that you know, I did­n’t write down… So, first of all elec­tron­ic mail was invent­ed before the Internet. It came as part of the ARPANET exper­i­ment, in 1971. It was an engi­neer named Ray Tomlinson who passed away a cou­ple of months ago who real­ized that you could move files from one com­put­er to anoth­er on the ARPANET. We designed it to do that. He said well, why—email was some­thing that peo­ple shared on com­mon time-sharing machines. If you had an account, and some­body else had an account on the same machine, you could send files back and forth. 

But he had the idea that he could use the ARPANET to send files from one machine to anoth­er. And so he thought, ​“Well, if I could just leave this file in a direc­to­ry that the recip­i­ent could find, then I could com­mu­ni­cate.” So he wrote a lit­tle pro­gram that took a file from one machine, sent it to anoth­er. Then he said, ​“Okay, how do I fig­ure out— I have to say which machine it’s sup­posed to go to, and then I have to say which direc­to­ry it’s sup­posed to end up in.” Well, the direc­to­ries were asso­ci­at­ed with users’ names, with their login names.

So he decid­ed, ​“Well, okay. I write that some­how. How do I sep­a­rate the user name from the com­put­er name?” And the only char­ac­ter he could find on the key­board which was­n’t already used by some oth­er oper­at­ing sys­tem was the @-sign, and some­how that made sense—user@host, right. So that’s why you have user@host as your email addresses.

So he invents that in 1971. We all go nuts because sud­den­ly we real­ized that we have a computer-mediated com­mu­ni­ca­tion sys­tem. Then we real­ized that means that you don’t have to be con­cur­rent­ly inter­act­ing with each oth­er like you do on a phone call. That means you that could talk to each oth­er even if you were in dif­fer­ent time zones. And so we got all excit­ed about that, real­iz­ing that meant we could man­age projects over mul­ti­ple time zones more eas­i­ly than we had before.

But what was sur­pris­ing to me is that that was­n’t the most inter­est­ing use of email. Within a few weeks of email’s arrival, which we all took up and and made use of, there were two mail­ing list cre­at­ed. The first one was called Sci-fi Lovers [SF-Lovers] because we were all engi­neers, we all read sci­ence fic­tion, and we want­ed to argue over who was the best author and what were the best sto­ries. So that was a mail­ing list which had social con­se­quences. The next one that comes along is Yum-Yum. It’s a restau­rant review thing that Stanford put togeth­er, which even­tu­al­ly expand­ed over a larg­er ter­ri­to­ry than Palo Alto.

But the point is that the two mail­ing lists that got start­ed were clear­ly social net­works. So right away you could see that this net­work was becom­ing a social­ly inter­est­ing phe­nom­e­non. And that’s still in the ARPANET days. By the time the Internet comes along, of course, now mul­ti­ple net­works can hap­pen, it’s spread­ing more broad­ly, more and more peo­ple can participate. 

When the World Wide Web shows up, it’s a very inter­est­ing obser­va­tion. Tim start­ed that work, Tim Berners-Lee—sorry, Sir Tim Berners-Lee—in 1989 while he was at CERN. And he released his first ver­sion of the World Wide Web around December of 1991. Nobody noticed. 

Then Marc Andreessen and Eric Bina of the National Center for Supercomputer Applications in Champaign-Urbana at the University of Illinois do the Mosaic brows­er, which is a graph­i­cal brows­er. Suddenly the Internet looked like a mag­a­zine with for­mat­ted text and imagery and even­tu­al­ly stream­ing audio and video.

Everybody goes nuts when they see that, includ­ing a guy named Jim Clark, who grabs Eric and Marc Andreessen, drags them out to the West Coast and starts Netscape Communications, which goes pub­lic in 1995, and the stock goes through the roof. It starts the Dot Boom. But what impressed me was not the Dot Boom. What impressed me was the avalanche of con­tent that peo­ple poured into the net once it was easy to cre­ate that con­tent and share it.

And the thing that was the most amaz­ing is not just the quan­ti­ty, it’s the fact that peo­ple want­ed to share what they knew. They weren’t ask­ing for remu­ner­a­tion, they just want­ed the sat­is­fac­tion of dis­cov­er­ing that some­thing they knew was help­ful to some­body else. And that phe­nom­e­non per­sists today. And so that prob­a­bly is the biggest sur­prise, is how much we want to share what we know. And of course the social net­works that we see today are just evi­dence of this increas­ing amount of desire to share what we know. So that’s prob­a­bly— My intro­duc­tion to the social impact of the Internet comes twen­ty years ago plus, twenty-five years ago, as the World Wide Web starts to show up.

Rainie: Unlike a lot of oth­ers who were involved in the cre­ation of the Internet, you haven’t yet writ­ten a book.

Cerf: That’s true.

Rainie: And I won­der if there is any­thing hold­ing you back, or if there are things that you would say in a book if you ever had the oppor­tu­ni­ty. Maybe a book is itself isn’t the schol­ar­ly con­tain­er to know what you know, and that you are think­ing dif­fer­ent thoughts about pulling it all togeth­er. But I’d be inter­est­ed in hear­ing you talk about that.

Cerf: There are actu­al­ly five books that I would like to write. One’s a book of poet­ry, anoth­er one is… Well, let me get them in the right order.

The first one I want to write is my wife’s biog­ra­phy, because that’s the most inter­est­ing sto­ry, much more inter­est­ing than mine. I mean, this trans­for­ma­tion from being deaf to hear­ing after fifty years is just unbe­liev­able. There are all kinds of great sto­ries to tell. So that’s first—

Rainie: You have to tell the title sto­ry of that book.

Cerf: When Sigrid got her cochlear implant in 1996, the first thing that she would say when we took the dog out for a walk was, ​“I heard that!” Like the birds were tweet­ing, and things like that. And so the book, I was going to title it I Heard That!. And in fact my good friend Gene Gabbard who’s sit­ting here in the front row— She was here vis­it­ing in North Carolina and bor­rowed a mobile phone to see whether or not she could use the mobile phone to call some­one, and it worked. She took his car and the mobile phone, drove around the cor­ner and called Gene, and it worked. So this was spec­tac­u­lar. It’s been one spec­tac­u­lar thing after another 

So the thing that I thought I would do is call it I Heard That, but then one day we were going to Washington on the Metro, and there was a per­son on a mobile phone behind her. She has this lit­tle patch cord with a micro­phone on the end of it that goes into the aux­il­iary input of her speech proces­sor. So she plugs it in, and she toss­es the micro­phone over her back so she can snoop on the guy who’s talk­ing on the phone. And I was kind of embar­rassed about that. I kind of moved away. And lat­er I told her, ​“Okay, I am going to write your biog­ra­phy, but it’s not going to be called I Heard That, it’s going to be called Snoopy.”

What was real­ly amus­ing, though, was after her first implant was done, they turned it on. I was­n’t there. At the time I was in a board meet­ing. But she called me twen­ty min­utes after they got the speech proces­sor orga­nized, and we had a con­ver­sa­tion over the phone for the first time, after being mar­ried for thir­ty years. That was amazing.

By the time I got home, I dis­cov­ered I had a fifty-year-old teenag­er at home. I could­n’t get her off the phone. Anybody would call— I was Senior VP at MCI at the time, right. So AT&T would call to try to get her to switch to AT&T. So she’d call them, say, ​“Oh, hel­lo. Where are you? Oh, you’re in Bangalore. Well, tell me about that.” So half an hour would go by and this poor per­son on the oth­er end of the phone would say, ​“Well, you’re going to switch now, aren’t you?” She said, ​“No, my hus­band works at MCI, but thanks for calling.”

So then she called the library and said she want­ed to to get record­ed books for the blind, because she want­ed to lis­ten to words that she had­n’t heard. She did­n’t say that part. She just called the library and said I’d like to sign up for record­ed books for the blind. They said, ​“Oh fine, no prob­lem; name, address, phone num­ber. Now, you’re blind, aren’t you?” And she said, ​“No, I’m deaf.” And there’s this long pause, and they’re try­ing to fig­ure out how’s that going to work? So she lis­tened to five hun­dred books on tape, or record­ed books. Now she can tell accents and mis­tak­en pro­nun­ci­a­tion. Very impressive.

So any­way she has this great sto­ry, so I want to do that biog­ra­phy. I should either try to do a defin­i­tive his­to­ry of the net­work, which would be very hard because there’s a lot of it now. I might be able to go from ​’73 to ​’93 large­ly on my own with a lot of mate­r­i­al that I’ve accu­mu­lat­ed. But to go anoth­er twen­ty years would real­ly be tough.

Or I’ll just do a mem­oir, you know, because I know where all the skele­tons are any­way. So I might do that. And then I have a few oth­ers I’d like to write. The trou­ble is that trav­el seems to get in the way, and I trav­el about 80% of my time. So that’s my excuse so far. But if peo­ple like Lee keep beat­ing up on me and say­ing, ​“You’re going to for­get all this stuff. You need to write it down,” I will be listening. 

Rainie: Well, I add my vote to the ​“yes” crowd that would like to hear from you in that way. I’d like to invite any­one in the audi­ence who has a ques­tion for Dr. Cerf to come to the micro­phones and ask it. As peo­ple are doing that, I’m going to go back to a sort of geeky part of the tech­nol­o­gy con­ver­sa­tion. Have you used Bitcoin?

Cerf: No, I haven’t used it. [crosstalk] I run away from it.

Rainie: But part of the hype cycle. Oh, okay. Because one of the things now that is cap­tur­ing the imag­i­na­tion of the tech­nol­o­gy com­mu­ni­ty in a large way—it’s prob­a­bly crest­ing at the top of the hype cycle—is blockchain, [crosstalk] the under­ly­ing ledger system.

Cerf: But let us now make an important—

Rainie: He’s going to cor­rect me folks. 

Cerf: Yeah, no no no no, no no. Blockchain is a cor­rect state­ment. I want to dis­tin­guish, how­ev­er, between Bitcoin and blockchain. Blockchain is the under­ly­ing mech­a­nism, Bitcoin puts on top of that, slathers on top of that, a whole bunch of oth­er stuff, includ­ing this bit­coin min­ing thing which has led to a kind of missile/anti-missile you know, build a big­ger com­put­er to do the min­ing and win the race to win the bit­coins. So that first of all is not a sustainable—

Rainie: My real ques­tion was about blockchain.

Cerf: Oh, okay.

Rainie: There are a lot of peo­ple who have very high hopes for its capac­i­ty to to fix some of the prob­lems of the Internet, and…

Cerf: Well… I don’t think—

Rainie: The trust and authentication— 

Cerf: Blockchain is not fix­ing a prob­lem in the Internet. What blockchain is doing is allow­ing you to build what’s called a dis­trib­uted ledger. And that means that mul­ti­ple par­ties can store infor­ma­tion about for exam­ple a trans­ac­tion. And because of the way the trans­ac­tions are con­sti­tut­ed, you can’t alter the trans­ac­tion with­out it being vis­i­ble. So this is an integri­ty issue more than any­thing else.

The prob­lem with the blockchain is that in one form, none of the par­ties are known to each oth­er. And so it’s anony­mous. And I don’t believe that the anonymi­ty is nec­es­sar­i­ly very valu­able. I would much rather know who is main­tain­ing the blockchain and have some con­fi­dence in the insti­tu­tion that’s doing that. So those are two dif­fer­ent kinds of uses of blockchain.

But the thing I want to warn against is that the fun­da­men­tal blockchain mech­a­nisms, the cryp­to­graph­ic mech­a­nisms, may be cod­ed exact­ly right. And they prob­a­bly are for the most part. But there is a big sur­round­ing chunk of soft­ware that has to be writ­ten in order to make use of the blockchain tech­nol­o­gy itself. And if that soft­ware has bugs, then you wind up los­ing the val­ue of the blockchain. And the case in point is Bitcoin, where com­pa­nies like Mt. Gox lost $400 mil­lion worth of bit­coins because some­body pen­e­trat­ed the soft­ware. Not the blockchain part of it, but the rest of it.

And so this gets back to my ear­li­er per­ora­tion about the prob­lem of bugs in soft­ware. So we’re back to blockchain may be won­der­ful, but we had bet­ter learn how to write the soft­ware that sur­rounds it more reli­ably or it will fail to be as use­ful as peo­ple would like it to be.

Rainie: Thank you.

Cerf: Shall we get some questions?

Rainie: Yeah.

Vint Cerf: Now, I’m going to test some­thing here. Don’t pan­ic, I may jump up and walk to the edge to the stage in order to lip read if nec­es­sary, because I’m hear­ing impaired. But go ahead and have a shot.

Audience 1: So my ques­tion is, looks like from what we’re talk­ing about, AI and the advances in Internet, it’s real­ly restruc­tur­ing all of soci­ety and kind of chang­ing the way human­i­ty thinks about what its role is. So, being one of the most impor­tant peo­ple in found­ing the Internet, what do you think human­i­ty’s new role is, or what are some of the new roles that human­i­ty needs to con­sid­er as we go through all these mas­sive struc­tur­al changes?

Cerf: So let me sum­ma­rize, and you tell me if I got this. The short sto­ry is is arti­fi­cial intel­li­gence going to change our soci­ety in ways that we don’t like and what should we do about it?

Audience 1: No, more so what do you think—

Cerf: Alright, so that’s why I’m going to get up and— Well, the rea­son is that there’s this lit­tle speak­er here. Go ahead.

Audience 1: So what do you think our goals should be as we go about dras­ti­cal­ly chang­ing all of human­i­ty with the Internet?

Cerf: All of the…?

Lee Rainie: Humanity.

Audience 1: Internet, AI, VR, AR…

Cerf: Well, look. I think there is an eth­i­cal issue here. And the eth­i­cal issue has to do with being able to if not pre­dict at least try to ana­lyze the con­se­quences of the tech­nol­o­gy that gets field­ed. And so I’m very con­cerned about the soft­ware side of things, which is why I’ve been harp­ing on it this morn­ing, or this after­noon I guess. That’s my biggest wor­ry, is that peo­ple will not under­stand that they are cre­at­ing an envi­ron­ment which may be brit­tle and may be exploit­ed, and not pay atten­tion to that or not even care about that. And I don’t con­sid­er that to be prop­er either indi­vid­u­al­ly and pro­fes­sion­al­ly, or cor­po­rate­ly. So that’s my big wor­ry, is this pro­lif­er­a­tion of soft­ware which fails to take into account what hap­pens when it does­n’t work right. And that’s my biggest worry.

And I don’t have a lot of good answers for how to pre­vent that from hap­pen­ing, oth­er than to put some pres­sure on peo­ple who pro­duce that soft­ware. Eventually I think there will come a time when pro­gram­mers will not be able to get away with ​“it’s just a bug.” It will be, ​“That was your soft­ware, you’re respon­si­ble for it.” Now, what the con­se­quences are I don’t know. We’ll have to see what mea­sures can be tak­en. But I think we have to face that.

Let me go over here and I’ll alter­nate back and forth. So that means the guys that’re over here, if you want to get there faster might run around the door here. You’ll reduce the average—no, that’s alright, go ahead.

Audience 2: Good after­noon, Dr. Cerf. My name’s Kanon[sp?], and at the begin­ning you men­tioned try­ing to achieve Internet access for every­one, of the whole plan­et. So, do you think the respon­si­bil­i­ty of get­ting Internet access to every­one lies more on the pub­lic sec­tor, so get­ting the gov­ern­ments to try to pro­vide WiFi or just Internet access in gen­er­al? Or the pri­vate sec­tor, because I know Mark Zuckerberg is very famous­ly going after it very aggres­sive­ly and try­ing to get Internet access in Africa and places like that.

Cerf: So, the hon­est answer is that this is an oppor­tu­ni­ty for all kinds of dif­fer­ent par­ties to engage. And I’m very con­cerned, for exam­ple, that the American Indian pop­u­la­tion does­n’t have access to the Internet on the reser­va­tions here in the United States. It’s just nonex­is­tent, for the most part. And we’ve failed mis­er­ably to do any­thing about that and that’s just wrong.

But the same argu­ment could be made in oth­er parts of the world. Now, there are tech­no­log­i­cal respons­es that we can make, and there are pol­i­cy respons­es we can make. On the tech­nol­o­gy side, Google is doing some fair­ly crazy things. We have this Loon project. These are bal­loons at 60,000 feet. They cir­cle the Earth at that alti­tude. We steer them based on the jet stream and alti­tude, so that depend­ing on which way the jet stream is blow­ing we’re able to guide the bal­loons over their next site to deliv­er Internet ser­vice by WiFi or LTE. 

So that’s one kind of thing. Others are look­ing at the pos­si­bil­i­ty of putting in low-flying satel­lites, for exam­ple the O3b sys­tem is a dozen satel­lites in equa­to­r­i­al orbit at 8,000 kilo­me­ters. Which means it’s a fifty mil­lisec­ond up and down, which is more like a con­ti­nen­tal delay than the usu­al syn­chro­nous satel­lite delay. Again, cov­er­ing 40 degrees North and South. So there are a bunch of tech­no­log­i­cal things that we can do and are being pur­sued as real busi­ness­es in the pri­vate sector.

But, it also can hap­pen that gov­ern­ments can choose poli­cies where, like in the case of Australia there was an inter­est in build­ing a nation­al broad­band net­work. That was a gov­ern­ment invest­ment in infra­struc­ture, just like the roads, on top of which the pri­vate sec­tor could then build. We have some expe­ri­ence with this at Google. In Kampala, Uganda we built in a giga­bit fiber net­work which we oper­ate as a whole­sale facil­i­ty. We sell to retail providers, and they resell to the pub­lic. And so it was an invest­ment in infra­struc­ture which is intend­ed to help jump­start the use of Internet in that part of the world.

So I can see com­bi­na­tions of gov­ern­ment pol­i­cy that encour­ages invest­ment, pol­i­cy that encour­ages com­pe­ti­tion, pol­i­cy that encour­ages com­pa­nies to invest. The mobile has helped very great­ly in get­ting access to Internet out there. Because before the mobiles, you had to pull wires. WiFi did­n’t reach far enough. But now we have mobiles and a lot of peo­ple are using smart­phones as their first way of get­ting access to the Internet. And as time goes on that may be the pri­ma­ry ini­tial expe­ri­ence that peo­ple have, which is why a lot of com­pa­nies like Google and oth­ers are try­ing to make mobile as use­ful as possible.

So, my sense right now is that there are wide-open oppor­tu­ni­ties to cre­ate more access to the Internet all around the world. And depend­ing on where you, what phys­i­cal con­di­tions are, and what the poli­cies are, you might use choose mul­ti­ple dif­fer­ent paths to get there. So, I spend a lot of my time around the world talk­ing to gov­ern­ments about poli­cies that would pro­mote that. And any­body else who wants to help, you know, you should do that.

Audience 2: Thank you.

Cerf: Okay I’m going to switch over here now.

Audience 3: Good after­noon. My name’s Ben. My ques­tion is, based off what you’ve already spo­ken about and oth­er expe­ri­ences, what was the hard­est chal­lenge to over­come in your pro­fes­sion­al career?

Cerf: The hard­est chal­lenge? [long pause] Besides rais­ing two sons. Which my wife gets most of the cred­it for. You know, the most inter­est­ing and hard­est prob­lem is get­ting peo­ple to want to do what you want to do. And so I teach my engi­neers that you have to learn how to sell your ideas if you want to do some­thing big. And so I spent a sig­nif­i­cant amount of my time con­vinc­ing a whole lot of oth­er peo­ple they want­ed to make the Internet hap­pen. And that’s been fair­ly suc­cess­ful. And part of the rea­son is that a lot of peo­ple decid­ed they want­ed this to hap­pen, they want­ed to par­tic­i­pate in it. 

And so some of the engi­neers, for exam­ple, think that mar­ket­ing and sales is this thing over here which is not worth their atten­tion at all. I have to remind them that if the sales teams don’t suc­ceed they won’t get paid. And their atten­tion sud­den­ly goes up again. Selling your ideas to oth­er peo­ple is prob­a­bly one of the most valu­able skills you can have, and it’s cer­tain­ly the biggest chal­lenge for me.

The sec­ond biggest chal­lenge in the case of the Internet turned out to be a com­pe­ti­tion between the TCP/IP pro­to­cols and the OSI (Open Systems Interconnection) pro­to­cols. That com­pe­ti­tion went on from 1978 to 1993. And although I was­n’t the only one fight­ing this bat­tle, it was gov­ern­ments that had to be per­suad­ed that the inter­na­tion­al stan­dards were not as func­tion­al and not as ready for imple­men­ta­tion as the ones that the American Defense Department had spon­sored. And you can imag­ine the polit­i­cal ten­sion here. ​“Why would we want to adopt a US defense depart­ment pro­to­col when we have our own standards-making activ­i­ty in the International Telecommunications Union and the International Standards Organization?” 

So that bat­tle took a long time and even­tu­al­ly was over­come by sim­ple sheer force. The pro­to­cols of the Internet were imple­ment­ed and prop­a­gat­ed every­where. They were freely avail­able. Bob Kahn and I gave that away on pur­pose. And so it was a real­ly pow­er­ful tool. No bar­ri­ers to adop­tion. So if you want your ideas to pen­e­trate, maybe you should think about giv­ing them away instead of patent­ing and con­trol­ling them. So that’s one answer, anyway.

Audience 3: Thank you.

Audience 4: Hi. My ques­tion is also about kind of get­ting Internet to the rest of the world. And it’s a lit­tle mul­ti­fac­eted, I guess. What do you think the best way to get devel­op­ing coun­tries online would be, and what do you think the impacts of— I guess now sta­tis­ti­cal­ly 100% of devel­op­ing coun­tries some­where have access to the Internet, but kind of the pro­lif­er­a­tion of that access to more peo­ple. Like, what are the impacts of that? And do you think there’s a rea­son (You said you’re an opti­mist; I guess maybe I’m a pes­simist.) that these peo­ple aren’t online right now. It’s been eigh­teen years, and it seems like it’s going slow­ly, or like you’re say­ing, American Indians don’t have access. It seems like they live in the coun­try that the Internet was cre­at­ed in, so why should­n’t they have the same kind of rights as we do?

Cerf: So, let me let me come at this by telling you a sto­ry about the refugee camps. And this present sit­u­a­tion espe­cial­ly with ISIS, and the sit­u­a­tion in the Middle East has caused an awful lot of refugees to migrate away from all the risk. The first thing they ask for is not food, shel­ter, or cloth­ing, or any­thing. They ask for Internet access. And the rea­son they do that is that’s how they get infor­ma­tion. That’s how they can con­tact their fam­i­lies and let them know where they are and that they’re okay.

So there is no doubt in my mind that Internet access is impor­tant in all coun­tries of the world regard­less of their eco­nom­ic sit­u­a­tion. And you’ll hear some peo­ple say, ​“How can you pos­si­bly waste mon­ey on get­ting Internet infra­struc­ture in place when they need clean water and food and hous­ing and so on and med­ical care?” All of which is true. But the Internet, to the degree that it is a use­ful tool for caus­ing actions to hap­pen, for get­ting infor­ma­tion to find out what to do, what’s avail­able, where is it, how do I get, it’s a real­ly pow­er­ful tool.

So I’m big fan of get­ting every­body online. I’m dis­ap­point­ed to tell you that in the United States, accord­ing to Lee’s work— I hope you don’t mind my doing it this way, Lee. I can hear bet­ter this way.

Rainie: It’s awesome. 

Cerf: I don’t mean to just turn my back on you.

Rainie: No.

Cerf: You could stand here, if you want to. 

Sorry about that. I just was con­scious of the fact that I’m show­ing my back to Lee.

His work shows that in the US, only about 80% of the pop­u­la­tion is online and the oth­er 20% does­n’t want to. And there are a num­ber of rea­sons that are giv­en for that. Some of them are eco­nom­ic, some of them are you know, ​“I hate social media and I don’t want to have any­thing to do with the Internet.”

So that’s dis­ap­point­ing and a lit­tle scary, because it might mean the same sta­tis­tic would show up else­where in the world. But I do believe that we have an oblig­a­tion— Maybe you don’t, but I feel an oblig­a­tion to do every­thing I can to make sure peo­ple can have the oppor­tu­ni­ty. They can decide whether they want to get on or not. If it isn’t avail­able, and isn’t accept­able, and it isn’t afford­able, they can’t make that decision. 

I’ve been involved in some­thing called The People Centered Internet for the last year or so. The whole idea here is to make use of the Internet in order to pro­vide infor­ma­tion to peo­ple that they will find use­ful, that will improve their lives. Quality infor­ma­tion about health­care being an obvi­ous exam­ple. And so I still believe that hav­ing access to infor­ma­tion on the net­work is one of its great­est val­ues, and this theme of shar­ing what you know is still there and still powerful. 

However, there is one oth­er prob­lem that you have to deal with. There’s a lot of misinfor­ma­tion and disinfor­ma­tion on the Internet. The Internet has no idea what it’s car­ry­ing, just like a post­card does­n’t know what you wrote on it. And so we inher­it an oblig­a­tion to think crit­i­cal­ly about what we’re see­ing and hear­ing on the net­work. Because there are peo­ple out of igno­rance who will put wrong infor­ma­tion on the net. Some peo­ple put bad infor­ma­tion on pur­pose­ful­ly. Sometimes they do it in order to get you to do some­thing like go to a web site and give away your password. 

So we are now oblig­at­ed to think more care­ful­ly about the infor­ma­tion we both give and receive in the net. We’re oblig­at­ed to teach our kids that. Because they’re going to be faced with infor­ma­tion sources not just on the Internet but every­where. Movies, tele­vi­sion, radio, news­pa­pers, mag­a­zines, and so on. Their friends. And so still think­ing crit­i­cal­ly turns out to be real­ly impor­tant. So one thing we should do as we try to bring this gift to the devel­op­ing world is to make sure they also under­stand that crit­i­cal think­ing is part of the deal. Okay.

Audience 4: Thank you.

Rainie: We’ve prob­a­bly got time for about two more ques­tions, I’m afraid to say, because we’re off to oth­er things and you have to go to classes.

Audience 5: So, I was won­der­ing, when you first start­ed work­ing on this project what sort of things were you imag­in­ing, envi­sion­ing back then that are here in the Internet now, and what things are here in the Internet that sur­prise you?

Cerf: So, this is ques­tion #102, which in short form is, ​“What were you think­ing?” So, first of all you should under­stand that this project arose out of a Defense Department inter­est in using com­put­ers in com­mand and con­trol. The the­o­ry was if you could put com­put­ers to work man­ag­ing your resources, you might be able to over­come a larg­er force by man­ag­ing your resources bet­ter than the oppo­si­tion. And that meant putting com­put­ers in air­planes, ships at sea, mobile vehi­cles, as well as fixed instal­la­tions. That’s what led to the basic Internet design, mul­ti­ple dif­fer­ent net­works, radio, satel­lite, fixed-wire line, and so on, all hav­ing to inter­work in some uni­form way. So that’s what drove us towards the Internet protocol.

The sec­ond thing is that we knew the mil­i­tary had to oper­ate on a glob­al scale, because we did­n’t know where they might have to go next. So we designed it to be non-national in char­ac­ter. So the address space does not have any­thing in it that says ​“this is a US thing and this is a French thing.” At a high­er lev­el, you see then in the domain name space, you see .us, .fr, .ee, but you also .com, .net, and .org, which are glob­al and not nation­al. So we were dri­ven by neces­si­ty into the archi­tec­ture that you see today. 

The com­mand and con­trol part also pushed us towards test­ing pack­e­tized voice and pack­e­tized video. So, in the 1970s, we were doing pack­e­tized voice and pack­e­tized video. Not very much of it because the data rates that were avail­able were fair­ly low. But today when you do stream­ing video, and when you do a Skype call, you’re exer­cis­ing mech­a­nisms that we were exper­i­ment­ing with and rec­og­nized the pos­si­bil­i­ty of forty years ago.

Now, I won’t tell you that we envi­sioned every­thing that’s hap­pened. I’d like to say that, but of course we did­n’t. But we actu­al­ly had had quite a bit of expe­ri­ence with the ARPANET and the social ele­ment that around arose out of the email. And so we had some fair­ly clear sense of some of what would hap­pen. When the World Wide Web showed up, that was much more shock­ing in some sense because of the graph­i­cal inter­face, the col­or­ful nature, and the com­po­si­tion of voice and video and text for­mat and every­thing else. 

The adver­tis­ing thing was not entire­ly obvi­ous. There were a num­ber of peo­ple who tried ban­ner ads and failed. Until Google came along and fig­ured out that the search engine bound with the adver­tis­ing was a good com­bi­na­tion. The search engines them­selves weren’t nec­es­sary until the avalanche of con­tent showed up on the net and peo­ple could­n’t find any­thing. So the search engines were the tool for for doing that.

What you should rec­og­nize is the Internet has evolved in steps, where a new capa­bil­i­ty is cre­at­ed when it’s need­ed. This is also true insti­tu­tion­al­ly. So, we invent­ed the Internet Society in order to sup­port the Internet Engineering Task Force when the gov­ern­ment said they don’t want to pay for that any­more. The ICANN got cre­at­ed when the USCISI said, ​“We’re in a space where we’re afraid we’re going to get sued,” and Jon Postel did­n’t want to get stuck with that, and so he went off to take actions which trig­gered the cre­ation of ICANN. The Internet Governance Forum got cre­at­ed out of a big debate about how we should do gov­er­nance of the Internet.

And so one of the nicest things about the Internet sto­ry is its abil­i­ty to adapt and to cre­ate new tech­nol­o­gy and new insti­tu­tions, at need. And it’s con­tin­u­ing to do that today. And so from my point of view it’s been for­tu­nate that you did­n’t have to pre­dict every­thing. That the sys­tem was so flex­i­ble that it allowed you to invent on the fly. And it’s still true today, and I’m hop­ing some of you will wind up invent­ing some stuff that nobody thought of, and maybe that will turn out to be a big, impact­ful effect. And that might be you.

Audience 6: Hello, Dr. Cerf. Thanks for com­ing. You touched on this theme a bit as you went through your talk, but I just want­ed to hear some more of your thoughts on it. What do you think about the state of opti­mism in 2016? It’s been said that it’s unpop­u­lar now—

Cerf: Wait wait wait. I’m miss­ing part of this. So we’re going to do this the easy way. [steps down from stage] I’m sor­ry about this for the guys on the cam­era. The state of…? 

Audience 6: The state of opti­mism in 2016.

Cerf: Optimism, okay. And should we be opti­mistic or pes­simistic about the net?

Audience 6: And also how oth­er peo­ple have… I feel like there’s this idea that cyn­i­cism is the truth a a lot more more. And you said you’re an opti­mist. What do you feel we can gain from opti­mism, and what do you think about the fact that cyn­i­cism is the in thing now?

Cerf: Okay. So I’m still not sure I got all of this. You want to know why I think that the Internet is some­thing I should be opti­mistic about as opposed to pes­simistic about? Or—

Audience 6: For the most part. Just in the world in gen­er­al, like how do you—

Cerf: Well, okay. So, look. We have to be real­is­tic about this. The Internet is a place where bad things can hap­pen. It’s a plat­form. It’s an infra­struc­ture. Look, let’s take cars and roads. People get into their cars and they drink and they dri­ve and they run into things. So they run into each oth­er, they destroy prop­er­ty. And at this point, we don’t aban­don all cars and shut down the roads because it’s too valu­able, and I would argue the Internet is like that. It’s a valu­able infrastructure. 

So what do we do about bad behav­ior which would dri­ve us into the holy cow, this is a pes­simistic envi­ron­ment? The answer is we adopt poli­cies and prac­tices that say to peo­ple there’s some things on the net which are not accept­able. And if we catch you, there will be consequences.

So, I have three solu­tions for the bad part of the net, the things that hap­pen that we don’t like. First one is to find ways to pre­vent the bad thing from hap­pen­ing, tech­ni­cal­ly. So we defend against denial of ser­vice attacks. We try to write soft­ware that does­n’t have exploitable bugs. There are a whole series of tech­ni­cal things we can do. We use cryp­tog­ra­phy in order to pro­tect peo­ple’s pri­va­cy and confidentiality.

But then, if we can’t stop all the bad things from hap­pen­ing, then we say okay, ​“If we catch you doing this thing, we con­sid­er that unac­cept­able and there will be con­se­quences.” That’s sort of post hoc enforce­ment. And we do this in law enforce­ment all the time. We tell peo­ple, ​“Here are the rules. We know we’re not going to stop every­body from doing these bad things. But if we catch you, there will be consequences.”

Then there’s a third thing you can do, and it’s going to sound weak. And that’s to tell peo­ple, ​“Don’t do that, it’s wrong.” Now, that sounds like sort of…it’s as weak as grav­i­ty, because grav­i­ty is the weak­est force in the uni­verse. Except, you notice that when there’s enough mass, grav­i­ty is real­ly pow­er­ful? Social memes and accept­ed behav­ior pat­terns are pow­er­ful. And so if we tell peo­ple, ​“Don’t do that because it’s wrong,” and if enough peo­ple say it’s wrong, then you will have an impact on peo­ple’s behav­ior. So like I say, I am opti­mistic about the human race, although there are times occa­sion­al­ly when I feel less opti­mistic. But on the whole, I think we’re going to be okay.

Alright let’s get this one over here.

Rainie: I think we will have to make this the last question.

Cerf: Oh, real­ly? This poor guy was waiting. 

Rainie: Okay.

Audience 7: I feel bad.

Cerf: It’s not his fault that I have long answers.

Audience 7: First off, I just want to say that I’m quite a big fan of your work. I don’t know if you get that all the time. But I feel like a lot of the ques­tions have been ​“what do you think the impact is on the Internet on every­body else.” What is your favorite use of the Internet?

Cerf: Honestly, short answer, Google. And it’s not because I work there. I use it all the time to find stuff on the net. I can’t even sit down to write a paper with­out using Google. Once when the pow­er went out I tried to write a paper long­hand and I could­n’t do it because I got three sen­tences in and I need­ed to look some­thing up, and the net was­n’t avail­able. And I’m just sit­ting here fib­ril­lat­ing. So the hon­est answer is being able to find stuff on the net, is what real­ly turns me on about its avail­abil­i­ty. So I use it all the time.

Audience 7: Thank you.

Cerf: Last question.

Audience 8: Well first off, thank you for actu­al­ly tak­ing my ques­tion. You kind of touched on it in the last answer you gave, like the bad actors that are on the Internet and sort of try­ing to stop bad actions. I want­ed to talk about the sort of dis­con­nect between what you do online and the social con­se­quences that you may or may not face offline, and sort of what we can do to stop these bad actors in terms of gov­ern­ment reg­u­la­tion, whether it be poli­cies to say you can or can’t do cer­tain things on the Internet, with­out infring­ing our right to free­dom of speech. Because one of the things that sets the Internet apart from oth­er mass medi­ums is that there is the aspect of inter­ac­tion, and that’s what makes it unique. So, what do you think?

Cerf: So, there are sev­er­al things here. You pack­aged up what turns out to be a fair­ly big issue. The first thing I would observe is that those of you who dri­ve in your cars, prob­a­bly espe­cial­ly if you’re alone, feel pro­tect­ed by the wind­shield so that you can say things that you would nev­er say in pub­lic to any­one, espe­cial­ly about the per­son that just cut in front of you. Or the light’s green and it’s just sit­ting there, and you’re say­ing, ​“You [mum­bles]…” So, we feel a lit­tle pro­tect­ed by our lap­tops and desk­tops and mobile screens, like they can’t reach us through there. And so there’s some behav­ior which is a con­se­quence of that. 

The sec­ond thing is that there are some peo­ple who imag­ine that they can say and do any­thing they want to because they are ​“anony­mous” and nobody will know. I think that’s turn­ing out to be less and less true. That dis­cov­ery of the ori­gins of things is becom­ing more com­mon than it used to be. Part of it is because we have bet­ter tools for doing that. But there are peo­ple who think that because they can’t see the con­se­quences direct­ly, that there­fore there aren’t any. 

We had a huge debate at Yale a cou­ple years ago exact­ly on this top­ic. We were talk­ing about multi­user games on the net, and the sorts of things that hap­pened there. And some peo­ple said it did­n’t mat­ter if you were real­ly nasty and brutish, because it was just a game. And yet there were peo­ple on the oth­er side of those avatars, and there were real con­se­quences to that. And so we were argu­ing that. And the team that took the posi­tion that there are real con­se­quences won that debate on the grounds that there real­ly are con­se­quence. And so we’re back to rec­og­niz­ing the impact of what you do. And maybe we have to make that more vis­i­ble some­how, so peo­ple are con­scious of that. But we’re back to sort of social engi­neer­ing, where peo­ple need to under­stand that when they’re tex­ting, when they’re play­ing the multi­user games, that there are human beings on the oth­er side and that you have an oblig­a­tion as anoth­er human being to rec­og­nize that. 

This is not always an easy les­son to teach, and we’re back to sort of val­ues again, and try­ing to help peo­ple rec­og­nize that the val­ues they expe­ri­ence in face-to-face inter­ac­tion should apply in this online world as well. And that prob­a­bly takes a lit­tle bit of train­ing for peo­ple to rec­og­nize, because the con­se­quences are a lit­tle dif­fer­ent than hav­ing you get very angry and punch me in the nose for hav­ing said some­thing because we’re only two feet apart. Well…six feet. (I’m and engi­neer, I have to get this right.)

So I think we have to have a job ahead of us to help peo­ple rec­og­nize that there is real­ly no dif­fer­ence between the inter­ac­tions we have through this medi­um and the inter­ac­tions that we have face-to-face.

Cerf: I guess that’s all time we’ve got, Lee.

Rainie: So, thank you. I have two final things to say. Everyone who is here today can and should either pick up the phone or broad­cast in some way that you were in the pres­ence of great­ness. Call a friend, call your par­ents, call some­body, or post to some­body who’s impor­tant in your life say­ing you were in the pres­ence of great­ness. And no one should ever be in the pres­ence of this man with­out begin­ning and end­ing the con­ver­sa­tion by say­ing thank you for the won­der­ful thing that gave us. So, thank you very much.