And in the eerie predictions by prescient geniuses department…
I’ve been examining the whole “post PC” buzzword that everyone is using these days on the eve of the Microsoft Windows 8 release. I first heard the late Apple CEO Steve Jobs use it in 2007. He was referring to an eventual world where mobile devices would replace PCs and Macs altogether. My old boss, Oracle chief Larry Ellison, used to talk a lot about the same when we were designing the web-based New Internet Computer (NIC) on a custom Linux platform by Wim Coekaerts. But then I found out that David Clark, a senior scientist at MIT’s Computer Science and Artificial Intelligence Lab (CSAIL) think tank, not only coined the term first in a 1999 speech called the Post-PC Internet, but he made a number of predictions about the future of technology going out a decade.
He nailed it, even calling the developments of tablets, digital TVs, Internet streaming radio and Internet services like Skype. Amazing. This speech is so good I’m including the transcript he sent me this week in full below.
MIT Laboratory for Computer Science
Transcription of a talk at the LCS 35th celebration
April 13, 1999
Let me explain this “Post-PC” phrase. I think several of us started using that term for different reasons. I picked it up because I wanted to tweak my friends at Intel, but of course now they are saying it too, so the fun has gone out of it. There is a spectrum of change that we are on. To start, I’m old enough to remember time-sharing and so I want to go back to the good old days when the CPU was this monstrously expensive thing and we all gathered in a circle and bowed down. So, you have all these displays clustered around the computer. (It filled the entire 9th floor of the building. It was wonderful. It was called “Multics”. Actually, I’m even old enough to remember CTSS too.) Moving forward in time, the CPU kept getting cheaper, so it needed to serve fewer and fewer people. And as the CPU got cheaper, we reached this magic point where there was a 1 to 1 ratio between the number of CPUs and displays. This stage is called the PC. 1 to 1 is a very powerful ratio. It lets you forget about all sorts of things for example we forgot about the X-window system. We forgot how to split an application into two halves. We are going to have to relearn all this. And because that’s a very powerful ratio we’ve sat at that point for a long time. But that hasn’t stopped this trend, where CPUs continue to get cheaper and more widespread. So, that ratio is eventually going to come unstuck. And when it does it’s going to do it violently. We can see this happening now. What you are going to end up with is this final vision here in which there is one display surrounded by a bunch of CPUs. You don’t want to think of that as PC 1, PC 2, and PC 3; that’s stupid. Those are all of the computers that are embedded in all of the objects that we encounter as we walk through the world.
So, what are the essential characteristics of this future? Processors are getting cheaper. That’s sort of straightforward. I hope I’ve talked you into that. The second is user interfaces go with people; they don’t go with processors. I have a display on my desk. I don’t want to have one display for each computer. I want to have one display. I’d like a small one with me when I walk around. I’d like a big one in my office. Maybe a huge one someplace else. I’d like one in my car that whispers in my ear. So, that’s the second point. And the third point–notice my picture had this downward trend with time. There is a trend that is going back toward the top of the picture, although we no longer call it time-sharing. It is now called “services in the net”. The idea is that if you put something in the net it actually may be easier to manage. You pay somebody else and he takes care of it. You don’t have to install it yourself for nine hours so that you have an excuse to fly to Asia. So, there is a force in this picture that is pushing us back to the top, there is a force in this picture that is pushing us to the bottom, and the PC is simply a way station along that path.
There are students I have met (it’s really depressing) who are so young that they’ve never seen anything except the PC. And they think that’s the way it’s always been. So, of course that’s the way it will always be. So, there is a role for us old fogies. If you’ve outlived the present into the past, maybe you can remind people that things change. We did TCP for the PC in 1982. That’s a long time ago. And this gives you a false sense of stability that traps you when you think about what applications are going to look like. I’m an Internet guy, so I like to think of applications that have the word “Internet” in their name: things like Internet telephone, Internet radio, Internet music, and Internet television. I had somebody say to me “I don’t want to go to my PC to make a phone call.” Wrong thought. I have a telephone in almost all the rooms in my house. Ten years from now what will you have in every room in your house? It won’t be a telephone; it will be a “something”. It will be on the Internet, we know that. It will probably have speech input and speech output. It will have a display. It probably won’t have a keyboard, if you believe Victor Zue. These devices will be ubiquitous. And that will be where you do your telephony, listen to music and so on. Now we might disagree as to exactly what that object is, but envision it and say that’s the context in which new applications are going to emerge, because that’s the user interface space in which applications are going to happen.
Now there are a couple of obvious things that follow from what I just said. Consumer networks have to be goof-proof. The consumer today thinks they can go to a drug store, buy an RJ11 jack and some wire, screw it all together and actually put a new phone extension in. They don’t have to run HP Openview in order to do that. So, whatever tomorrow’s data networks are, they have to be consumer grade. You can’t break it, and if you do, it fixes itself. You can buy it at the drug store. Second, it has to be cheap. There is an old saying that somebody made up, which is that the cost of networking something should be no more than 10% of the cost of the something. So, if you have a dollar object in your hand, putting it on the net shouldn’t cost more than 10 cents. And that implies that networks have to be heterogeneous.
Why is the future so heterogeneous? Well, you are going to have somewhat expensive (like a dollar) high-speed networks that are good for entertainment video and high-speed audio. And you are going to have really cheap nets, like a tenth of a cent, which are going to hook together little light switches and things like this. And all of this has to work together. Somebody said to me that consumer appliances aren’t taking off because we haven’t figured out what the standard is for networks in the home. Wrong thought. Multiple kinds, so glue them together. They have to be glued together. (Great new market for Cisco.) But accept heterogeneity. Now those of us who have lived in Internet for a long time have remembered that we didn’t always have one local networking answer. As Bob will remind you, if I were to drop my hat, I tried to invent something called “Token Ring”. Oh well. But during the time that we were debating this we could still talk to each other because of a set of protocols called Internet that ran on top of that mess. And we need to take the Internet approach for device networking.
I want to look at some slightly more interesting observations about system structure. And this relates a little bit to what Anant was saying and Bill Gates said a little while ago. If you think about the devices we use today, a lot of them are in the word processor stage that Gates was illustrating in which you have a device with a custom application built into it. My cell phone is really only one application. It does cell phone for a living. A pager is a one-application device. Then we’ve got the device that’s the amazing shrinking PC. It turns from a desktop to a laptop to a palm top. They are getting thinner and thinner. But they are still PCs. They have complete programmability in the way that the shrink-wrap model of today assumes. So, the first stage in the process of getting these devices into the future is to make them smaller. But the second stage, which I think is actually really more interesting, is when they start to be dependent on services in the network so that they are not quite so self sufficient. The exciting applications that run on your palm object are the ones that have matching services back in the net. My assistant can make an appointment in my calendar, and if my palm thing were only continuously on the net, it would be able to tell me about it instantly. So, what does this say about software? Well, it says we have to do something we’ve talked about for a long time. We have to write client/server software. Now today the server is on the PC, but the services are just going to diffuse into the net as a set of things that you can either purchase or rent. There will be lots of companies selling you services on the net.
The next thing that’s going to happen to this device that you carry with you is it’s going to blow up into parts. Here is a cool device. Can you see it in the back of the room? This is a little earphone with a little microphone. You put it in your ear like this and you talk. Now what era does this live in? For the moment, this lives in the one device, one application era, because the plug on the other end goes into my cell phone, so the only thing you can use this for is to make a cell phone call. But as Victor is about to explain to you as soon as I get off the stage, we want to talk to every one of our applications. So, we are going to move from the model of one device, one application, to one device, one function. This is the listening function. This is the talking function. My digital wristwatch is the two-line display with four button functions. My palm thing is the slightly larger display with a pen function. And now when you want a new application, for example, a cell phone, what happens? What you are going to do is download a piece of preliminary software that looks around your world and says, “Oh, he has asked me to build him a cell phone. Can I do that? Well I found an earpiece, I found a microphone, and I found someplace that the caller ID could show up. I found his little black book out on the net; I found this other service over here. If I stir and mix, that will be a cell phone.”
So applications, as somebody previously said, are not shrink-wrap. What next generation applications have to do is come down into your world and organize themselves to match. And notice that applications get organized differently for each person because I might like the caller ID information on my wristwatch, you might like the caller ID information gently whispered in your ear, and somebody else might say, “That’s really stupid, I just want this thing in my pocket to vibrate, and I’ll answer it.” So, there isn’t a uniformity of user interface. There isn’t a uniformity of application function. It’s the opposite of shrink-wrap. It’s going to be self-organizing and personally configured, making use of both the devices you carry and the services that are out on the network.
Let me make one more point about this future world. Everything is going to have a computer in it. Everything is going to be networked. Why? Do I have some grand vision? No, I don’t have a grand vision; it makes the cost lower. Lowering the cost equals inevitable. Somebody was explaining to me that when you manufacture an appliance (I don’t mean a computer appliance, I mean something like a microwave) it’s cheaper to put a network interface on the back of the box than it is to put a user interface on the front of the box, because a user interface on the front of the box has buttons and displays and things, and those cost money, and they break. So all of these computers embedded in devices and appliances are going to get networked. And here is the point: each one of those devices represents a little piece of cartilage that is linking the real world to the cyberworld. When you have an object that has a computer inside of it, you can deal with it in two ways. You can walk up to it in the physical world or you can talk to it in cyberspace.
Consider the Coke machine of the future. The Coke machine of the future does not have to have a place where you put money in. Boy, does collecting money add cost, because not only do you have to have the machinery that can tell real quarters from fake ones, but then you have to send a human out to empty the money and you have do things to make sure that he is not cheating. No, a Coke machine in the future is a machine for dispensing Coke that just has a little placard on the front. And the placard has some sort of a machine-readable name across the bottom and then above it it says, “Scan and click for a Coke”. So what do you do? You take out of your pocket some little object that scans that name that is on the little placard and you push the “Do It” button. This is like a mouse in real space: you can point and click and you can drag and drop. When you push the button on your hand-held device the message goes through the Internet to the place where I pay the Coca-Cola franchisee a dollar, and then the Coca-Cola franchisee sends a message across the net to the Coke machine and it dumps a Coke out. Why is this inevitable? Because it is a much cheaper Coke machine. So what space links real and cyberspace? The names must be very flexible and must be machine scannable. You don’t want to sit there and type in a URL in order to get a Coke, so this has got to be automatic.
So what does the future look like? Well, it is a network full of services – all kinds of services. Calendar maintenance services, e-mail services, voice mail services. It is a world in which there are devices that you use as user interfaces: devices you carry with you – the Handy 21 is an example. There are devices you use when you are near them, like the Enviro 21’s. The applications that you choose to run are constantly adapting to these new devices as you walk into a room, the application has to reconfigure itself to use the different elements that are present. Then there are the computers that are embedded in all the devices and appliances that you encounter as you move through the world. These are devices with which you must interact to purchase a Coke, pay for parking, control your microwave. You will want to interact with these computers, and you will do this by finding the linkage between their manifestation in real space and their manifestation in cyberspace and dealing with them using the things you carry with you – your interface devices. That is the post PC world that I think we are about to move into.”