Talk:Learning Learning/Parable 3

Absolutely. You hit it right on the nose.

And while we're at it, there's no reason for them to learn to use a pencil or pen either. A keyboard is so much faster. And even that will be obsolete when we get better voice recognition. Technology for reading and speech synthesis is improving rapidly, so soon it will not be necessary for anyone to learn to read. We also need to take a look at the interface. Written characters are highly inefficient for storing and transmitting information. Human language could be stored much more efficiently in machine form, and that would eliminate the need for (1) paper, (2) keyboards, and (3) visual displays.

As long as children can carry around a laptop, computers are so good at remembering things that there's very little need for them to remember anything. In fact, they won't really need to learn much at all except how to get along with each other and how to turn on the computer. Socialization is probably best taught at recess, so there won't be much need for classes, or classrooms any more. With improving communications we are rapidly reaching the time when face-to-face interaction is no longer needed, so even the recess can go.

Machines can do our work for us, and even most of the chores will be done with machines. So that really doesn't leave much humans will be needed for, except, of course, programming. All this will happen much sooner than we think -- probably even before Firefox quits leaking memory, some time by mid to late 21st century. Now if I can just get this &*#$@ program to compile and quit crashing....

Laptops? Yes, laptops for all. Just one problem, though. It's hard to learn number theory until one has a thorough concept of numbers.

I disagree very strongly with your first paragraph. It's complete nonsense. This whole idea about audio being some kind of future for human/computer interfacing is fluffy thinking. Let's look at this: Speech synthesis/analysis has major problems of a very fundamental nature. It's a broadcast medium. With screen and keyboard (or book and pencil) any number of people can work together in a room. Add speech synthesis and you have an unbearable cacophany - and even with headphones, you have an impossibly noisy environment just as soon as you add speech analysis - not to mention that the things I say get picked up by my neighbours computer. I can't use audio in a noisy environment - how do I use my computer in a factory environment? Furthermore - speech is a serial, almost one-dimensional process - it's hard to hear the output from a dozen programs running at once without getting confused. Vision is much easier to spatially separate. When a program outputs data - it's easy to see which window it appeared on. Worse still, audio is inherently temporally bound. The program says something - if you didn't pay attention - you have to ask it to repeat it...with video, the data can stay there for as long as the program wants you to see it. Further - the BANDWIDTH of your ears is at best about 20kHz - and about the fastest we can process data is the rate of human speech - perhaps 100 words per minute. The bandwidth of our eyes is roughly 12 million pixels at maybe 10 frames a second with colours - that's maybe 200MHz - twenty thousand times the bandwidth. Try analogies like 'cut' and 'paste' with audio - it's an exercise in total futility. There are a few things you can do with audio that you can't do with sound - one is that you can broadcast to many destinations at once - but networking solves that with computers - you can just tell one of them and have it pass the message along to the others. Also, audio let's you communicate around corners...maybe useful for ordering robots around...but generally, audio is a crippling medium. So - no - there are VERY good reasons why we don't have practical audio interfaces that seeing people actually use - and we're not likely to change that in a hurry.

I don't know what makes you think that storing human speech would be more efficient than text. A long-ish word in English might be 10 characters of text. That's 70 bits. Text is pretty easy to compress and there are plenty of algorithms that'll compress ASCII text to one third of the size - so we're talking maybe 3 bytes per word. How much audio can we get into 3 bytes? Well, 4KHz is recognised as pretty poor quality - so if a word takes 1/4 second to say, it's 4,000 bytes - compression of audio can be fairly agressive - maybe 10:1 but you're still looking at 400 bytes per word. So text is more than 100 times more efficient. Then look at the processing power to decode speech - it's a nightmare! No two people sound the same - you have tons of amgiuity that requires AI to sort out - even one person sounds different when they have a cold or talk with their mouths full. Searching ASCII text for every occurrance of some word is a trivial task for a computer - yet finding a specific word in a bunch of audio recordings from different speakers is still a research topic! One day it may be POSSIBLE - but compared to searching text, there is no way you could ever describe it as 'efficient'.

SteveBaker