10-31-2019, 11:20 AM
(10-30-2019, 06:44 PM)Warren Dew Wrote:(10-29-2019, 06:16 PM)John J. Xenakis Wrote: ** 29-Oct-2019 World View: Exponential increases in computing power
(10-28-2019, 10:16 PM)Warren Dew Wrote: > While I agree with the "fast enough computer" theory, I don't
> necessarily agree with the idea that we'll get computers fast
> enough and efficient enough.
> Here's the issue: for decades, Moore's law was driven by reduction
> in processor size. Unfortunately, within the past decade, that
> reduction reached a physical limit as circuits on the chips shrank
> to the point that inductance effects between adjacent "wires"
> became limiting.
> At this point, Moore's law is being driven by parallelism instead
> of more powerful processors: having larger numbers of the same old
> processors running in parallel. That can still result in ever more
> powerful computers - but they will also be ever larger and consume
> ever more energy. That limits their usefulness in certain
> applications, in particular mobile applications.
> In 2050, the best artists may well be computers which fill
> buildings. We may still be traveling around town in Uber and Lyft
> vehicles driven by human beings, however.
Nobody believes this. You're saying that progress in computer
technology, which has been proceeding exponentially for decades, is
now stone cold dead.
Not at all. Let me try putting it in different words, since those didn't work for you.
Moore's law is a law about computing power and costs. It is not a law about computing power and size. Sometimes the computing power will come in the form of machines that sit on desktops. Sometimes it will come in the form of machines that fill rooms.
In the current phase, which is likely to extend to 2030 at least, the machines get bigger as the cost of computing power goes down. That will change if and when the trend jumps to a different technology, as it may have several times in the past, but the next such jump is more likely around 2050 than 2030.
There's also no guarantee that the next jump will go to smaller rather than larger machines. It may turn out that quantum computing can be done much more efficiently in space with large volumes of hard vacuum, and we'll have huge quantum computers in orbit which we access over a wireless internet equivalent, as just one possibility. They will by that time be much smarter than human beings, but they may not be able to walk around on Earth.
Truth be told -- some of the reader devices give easy access to calculating power that the Apollo 11 astronauts had available -- and that was less than the typical scientific calculator that college students had in the mid-1970's. There remains a size limit on computers, and that is a minimum size: human fingers are not getting smaller. Human capacity to relate to data is not increasing beyond a certain level.
I would guess that the most sophisticated computing now goes to cinematic CGI effects. People obviously relate to the results by attending movies that could never be made without them. People see but a tiny fraction of the computational sophistication -- and that is under human management, and then by people with much creative talent to judge the results before those can ever meet the quality necessary for the finished product.
Creative people, unless limited to crafts and traditional media, will need more intellectual sophistication just to control machines that turn out material for mass consumption. As it has been with painting, any blockhead can toss paint onto a canvas, but editing the results requires some intellectual sophistication. Tyro as I am at the hobby, I know enough to not show incompetence through bad perspective and to avoid making images of automobile graveyards or of the interiors of slaughterhouses.
The ideal subject of totalitarian rule is not the convinced Nazi or the dedicated Communist but instead the people for whom the distinction between fact and fiction, true and false, no longer exists -- Hannah Arendt.
