The Psychotic State Number 14 Oct 31, 1999 Halloween Musings on the World To Come Halloween is not usually a holiday associated with thoughts of the future and past, yet that is what I am pondering today. Perhaps it was the sight of the little kid trick-or-treating dressed as a Palm Pilot that set off my current psychosis. Whatever the reason, I have spent this particular holiday musing over what the our industry is likely to change during the next few years, and what it's likely to mean for all of us. I'll begin by making a list of what I think are some important improvements in technology. Nothing I'll talk about is really off-the-wall - everything in the list already exists in some form or other, and the changes I'll predict are just what one would expect from ordinary progression of things. The list is not in any particularly logical order, but it goes roughly from what I think is increasing order of coolness and wierdness. *) More Moore's Moore's Law refers to the historical trend for semiconductor makers to put ever more transistors on the same size of silicon for the same amount of money. It's pretty powerful - it's what has been responsible for our spending less money today to buy 500 MHz Pentiums IIIs with 64 MB of RAM than we did at the beginning of the decade on our 10 MHz 80386's with 640kB. So computers will continue to get smaller, lighter, cheaper and more powerful. This is supposed to go on for at least 5-20 years more before we start to hit fundamental physical barriers to miniaturization. (You can't make a bit of memory occupy less than one electron, for example.) Why did I put this first, i.e. least interesting? Because (a) we're kind of used to the changes this brings, (b) lots of people have thought about it already, and (c) I bet that if the computers we use never get any better than they are today, we could still do everything I'll talk about below anyway. *) You know where to find me... This is an easy prediction: ubiquitous bandwidth. Every telco in the country is spending buckets of cash to put cellular and fiber networks all over. In just a few years, there will be high bandwidth connections everywhere, literally everywhere. You'll be able to read all your e-mail on your PDA (or whatever it is), you'll be able to surf the web (or whatever it is) from the beach. Today, we have the sense of being "on the net" or "off the net". This distinction will disappear. When you go to visit your friend's house for dinner, you don't think about whether there will be electric lights there. In twenty years we'll think the same way about bandwidth. *) You're on Candid Camera. To my mind, the interesting parts of Moore's law happen on the edges. We are on the cusp of a revolution right now: the end of film photography. Today, it's still just a bit better and cheaper to shoot on film than on a digital camera. We haven't really been at this digital photography thing for too long, and I bet that there's lots of room left to improve. Right now you can go out and buy a digital camera that costs under $500 and will take megapixel shots. The problem is that images on film are a lot like words on paper - fine for what they are, but hard to store, manipulate, and transport. We'll want 'em electronically. I'm betting that Mr. Moore will have something to say about this all (since fabricating CCD cameras is, I suppose, not all that different from ICs). Thus the prediction: within a few years, it'll cost $5 for a high resolution camera, better than film, which can be hooked to the network. That latter won't be hard, because the network will be everywhere, as I said. No more fumbling with flash cards or floppy disks (hah!). How many places would we install cameras if they were that cheap and simple? Will Big Brother be watching us? I don't know. *) I've seen that face somewhere before! With cameras everywhere, it would be pretty surprising if the computers couldn't do something with their eye on the world. Not to worry: as I write I have friends who are developing software that can recognize people's faces. There's a spooky part of this, but then again there's lots of interesting applications as well. Imagine: in a world where there is bandwidth everywhere, and cameras everywhere, and software that can recognize your face, why have wallets? Why have charge cards? Why remember two hundred passwords? You could walk into Han's deli, stick your face in front of the machine, debit your account and buy that yogurt. If a mugger attacks you, there's nothing to take but your mug. *) You talking to me? Ain't no one else in here, you talking to me? There will be software to recognize your face, and there will be software to understand your speech. There's nothing new about that - it's already out on the market today. (I have friends in this industry as well.) The problem is that today's software is not fully baked: it needs to be "trained" to your voice, its error rates are significant, it's expensive. These issues will all go away. Even today there are systems that can recognize a limited number of words as spoken by any person at all. With a few more years of work I expect that it'll be possible to have a computer take a written transcript of everything that's said in a room during the entire day. What will computers be like when there aren't any keyboards on them? Which leads me to a question: will people 50 years from now know how to spell? Sure, I suppose they'll be able to read, but if they never actually have to put letters together to form words, is there any reason they'll know when to write "their", "they're", or "there"? Frankly, will there be any reason for most people even to learn how to hold a pen? (This may seem pessimistic, but really, how many people know that distinction today? I know how few people can do arithmetic since the invention of the calculator. I have more than once seen someone use one to multiply a number by 10.) *) In the beginning was the word. Search engines keep getting better. Altavista, for example, indexes a substantial fraction of the web and can find all documents with your keywords in a fraction of a second. If you pause to think about it for a moment, you'll realize just how cool that is. But simple searching is just one of the amazing things people are doing with text. Software exists that finds similarities in documents and how they use words to try to determine what they are about. Thus if document A uses words the same way as document B, and document A is about trout fishing, we can probably determine that B is about trout fishing as well. No more hunting through search results of "10,000 documents found". It gets better: There are people working on programs that can grade students' essays (http://www.linguafranca.com/9907/nwo.html) and detect plagiarism in programs submitted in student's computer science courses. It all sounds like magic, but it's nearly here today. All of this has to result in some pretty slick e-mail response management systems and improved customer service, right? *) MegaGigaWhat? The machine I am typing this essay on is my 25 MHz 68040 NeXT. When I purchased it, in 1991, I bought an external hard drive for it: 200MB for about $1000. At the time I thought that was pretty cool, because a decade earlier one would spend that same amount of money for a 360K floppy drive. Checking a MacMall catalogue I got in the mail this week, I find that one can buy a 50 Gig hard drive for that same $1000. So in the last 20 years, storage cost per bit has come down by a factor of nearly 1 million! What can you do with 50 Gigabytes? With good compression, you can store a feature film in about 5 GB, so 50 Gig is good for about 10 such films, which you could watch in a day. I suppose the reason people don't yet do this so much is that it's still $100 to store a film, which is way more than the couple of dollars a video tape would cost. Suppose that this rate of cost reduction continues for the next 20 years as well. In 2020, for $1000, we can buy a 50 MegaGigaByte drive (it may not be a spinning magnetic disk, but whatever). What can that store? Why, ten million feature films, of course. Has there been that much film ever shot? Even if there has, you'll never watch it - at 10 films a day it would take 1 million days, or more than 2500 years. Such a drive can hold every word ever written and every sound ever recorded. What will we do with that? (Of course you should bear in mind that MS Windows NT 2020 will take 15 MegaGigaBytes of this drive, but you'll still have the remaining 35 for films, software, etc. Unless of course you want to load MS Office 2020, in which case you should buy an external drive.) *) Take me to the Holodeck. There are going to be all sorts of improvements in display technology. In a few years, these bulky, powerhungry CRTs we spend our days with will became a thing of the past. People are researching flexible versions of the LCD displays on laptops - roll 'em up and tote them around like a newspaper. There will be virtual reality glasses that can project images directly on your retina at the full resolution of your eye. Think about that: whatever the computer wants to project on your eye, you will see. What kind of information architecture do we build in such a medium?! This last is the wierdest, and also the most speculative. But not really: these sorts of displays are certainly coming. I really have kept away from stuff along the lines of computers that think or quantum computers or whatever. Those may happen too, but I'm in the mood to be conservative. We've reached the part where I promised to tell you what this will all mean to us. You'd be a fool if you listened to anything I said. A society in which people live with the tools like the ones I've spoken about above is just too different from ours for us to contemplate meaningfully. The biggest changes will be wrought in _ourselves_, or more properly in our descendents. Their expectations of what it's like to be in the world will be radically shaken by these changes. For example, with cameras everywhere, what will people desire in the realm of their privacy? What will the class structure be - who will be rich and who poor, and what will poverty and wealth be like? To see the futility of predicting such things, imagine a person in the year 1800 who wanted to make predictions about the meaning of that radical new invention, the steam engine. Would anyone sane be led to predict in the coming century the abolition of slavery? I'm not much of a futurist. I always loathed listening to Newt Gingrich, and I found Alvin Toffler and his whole movement rather naive. I don't know whether these changes will be good, bad, or indifferent. But it sure is pretty damned cool to think about them.