Re: merits of Lisp vs Python

André Thieme wrote:
Rob Thorpe schrieb:
André Thieme wrote:
2. He does not back up his argument with facts, just vague views of
history.
Can you back up this statement with facts?

Well, some yes. The burden of proof should not fall on me though, I'm
simply claiming that technological progress won't result in a
"singularity". This is not a revolutionary idea, the burden of proof
should lie on those claiming the revolutionary idea to be correct.

I think he did a good job with his several books, talks and hundreds of
pages of text on his website.

I don't. He leans of specific examples from information processing
which are not analogous to the situation in other fields.

What about his analyses of his "Kurzweil law" for the last 100 years?
It is correct.

Regarding information technology, I think he has a point. Raw
computational power has increased hugely over time. This has not led
to a great increase in the usefulness of that computational power
though. Is your computer really more useful to you than it was 6 years
ago, mine isn't.

My computer is more useful. Perhaps if you would get back a P3 733 with
256 MB Ram and 20 GB HD

Actually my computer is a Duron 1400 with 256MB ram and a 20GB hard
disk.

you would change your opinion.
I personally like to have a faster computer. Even small savings are
worth something, for example the reduced render time of web pages.

I find that is down to the speed of the internet connection. It didn't
improve
last time I upgraded my PC.

If you think otherwise you shouldn't use macros in Lisp which often do
save you to type #'(lambda () ...) and not more.

Yes. But the improvements are different, they are accumulative and
help the important and time-consuming task of maintaining code.

3. He quotes Moores Law, which is really something quite exceptional.
And he worked out himself some more info, and some proofs you were
looking for.

I know, I've read it.

I particularly like this last bit:-
now doubling every year.

My response to that would be "really?" Anyone who's bought a computer
recently will know that, getting something significantly faster than 1
year ago is becoming increasingly difficult. "Dual core" hardly counts
as faster.

It is tricky. He is not saying that your computer doubles its speed each
year. What he says is that ca. every 12 months the number of calculations
per timeunit that can be bought for 1000$ are doubling.
And that I calculated several times over the past years and it seemed ca.
correct to me.

But is it relevant. Compuational power by itself is only of use to a
very narrow field of science. That where 1) computation is important
2) programs can easily be rewritten to take into account new computer
technology.

I work in RF and electromagnetics, you would have thought in that area
the increase in the raw cost of computation would be very important.
Surprisingly it isn't, what is more important is practical performance
of real code. The people who sell electromagnetics simulators for
example have not yet optimized them to work well with
multiple-processors, even though their customers have been using
multiple processors for years.

For me something happened about which Kurzweil is not talking but which
some people think he is:
Last year I bought an AMD 64 3200+ for 156 Euro. This year in august the
price was around 73 Euro (now it costs 55,84 Euro).


I'd agree that is important to people buying PCs.

Btw: for task that can be parallelized well the speed doubles with
2x amount of cores.

Not for many though.

Quote:
"For example, when the human genome scan started fourteen years ago,
critics pointed out that given the speed with which the genome could
then be scanned, it would take thousands of years to finish the project.
Yet the fifteen year project was nonetheless completed slightly ahead of
schedule."
Today the complete process could be done within a few months.

Yes, but this is just a relatively simple computational task. These
things are not really that hard. If the need to do it had existed in
1960 then someone with the resources could probably have done it even
then, by using special purpose computers.

Maybe. But perhaps it would have cost some billion dollars.
With computers as big as New York it might have been possible ;)
Let's take a look at
http://www.kurzweilai.net/articles/images/chart04.jpg

I doubt that. You have to understand the nature of technology a little
more. Technology comes into being when there is a market for it. In
recent times there has been a market for DNA sequencing so great
advances have been made.

If the need had been earlier in the century, and the need had been
great enough then it probably would have happened. As an example, the
Collosus machines used to break the Fish ciphers in WWII were special
purpose computers. Because great effort was put into optimizing for
their exact task it was only in the 90's when general purpose computers
because faster at it.

In the 60ies the price for one base pair must have been around 10k.
2002 it cost 1 Cent and today the prices are probably even lower.

And history luckily begins to repeat, this time with medicine.
Science begins to discover medicine as an information technology.
We begin to understand more about it. Until today medicine was more or
less blind guessing. If we can continue to get more knowledge we can
begin to calculate drugs. Much less tests with animals will be needed
in the coming years and drugs can be found much faster.

I've known people who work in this area. I'm afraid to say medicine is
still largely blind guessing. The problem here is that it will take
many decades, probably centuries, of science to understand even how the
processes in a single human cell work. Information science will
continue to advance and aid the work of scientists, but it cannot do
the work of evidence gathering itself. In medicine it is evidence
gathering that is the most difficult step.

Another point of critic: after something happened it is easy to say
"Yes, but ...".
When in 2030 computers are as fast as Kurzweil predicted all critics in
this thread will say "Of course, I always knew it."

The question is will they be fast in any way that is usable to a normal
programmer.

I expect that in 2030 computers will be of formidable power. Probably
the average machine will have hundreds of processors. But I don't
expect it to be radically more useful to people, and I don't expect the
increase in usefulness to be necessarily driven by the increase in
speed.

In Windows Vista for example there are useful new features, and the
whole system is slower than XP. But these things are not interlinked.
The features are there to persuage people to buy Vista. The slowdown
is there to encourage people to upgrade their machines and buy more
Vista licenses.

4. His arguments concern the drivers of technology, assuming that
anything is possible, which is not the case. Once the drivers hit
problems with what is possible they must slow down. Similarly, what
happens when demand slows down? Personally I can't find anything
useful to do with a fast computer these days, from my own point-of-view
fast computers are only for playing games.
He also talks about the slow downs.
If you personally can't find useful things that you can do with computers
it doesn't mean that others see it in a same way.
Jon Harrop for example seems to want faster computers. He is waiting
for systems that allow Stalin to compile a program in less than one
second.
Google also would like to have computers that are some trillion times
faster than todays.

Yes. But it is the mass-market that makes other things possible.
Today people aren't as worried about the compuational power of their
computers as they were once. They're more concerned about style and
usability.

I agree: speed is not a big priority anymore (besides those of us who
are forced to work with the Scheme Stalin compiler).
I for example invested in a silent computer system.

Exactly, and with more people making that choice those who demand speed
will not be able to ride on the coat-tails of everyone else, as they
have for a long time.

This is something people forget. Kurzweil talks about the huge
increase in bandwidth of wireless communications networks. He talks as
though its created through better technology. Really this only has a
small part to play, systems with similar capabilities to todays GSM
networks could have been built far earlier than they were. The real
driver for change was financial, the network owners managed to get a
large user base.

But what you say is true for any technology. Now as everything we have
already is developed it is easy to say "It could have been done earlier."
I don't buy that.
Of course the market plays a big role. We can't sell things that are
percieved as too expensive. Anyway, check the number of Patents that are
made,

Many more patents have been made recently not because more things have
been invented, but because the USPTO have stopped checking patents.
You can patent anything you like these days, it doesn't have to work or
be useful. Most patents don't and aren't useful.

Besides, citing patents as evidence that technological progress will
increase in the future is false. Patents give revenue the assignee, to
everyone else they are a disadvantage, a tax. Increasing numbers of
patents hold back the use of new technology. I for example am working
on a hardware project currently. I'm doing what I can to avoid
patented technology, even at the cost of time, money and performance.

about success in science, nanotechnology, IT, medicin, etc.
When we look at the bigger picture there is something happening.

I don't see science or technology increasing at any faster a rate than
they were 10 years ago. I guess it's a matter of point of view.

And it is not going to advance with the speed it is happening today.
It is getting faster. At least this is what happened in the past 10k
years. Saying that the paradigm shift rate will stop to increase will
need strong arguments.

No, what needs a strong argument is that information science and
technology is so important to the rest of science/technology.

I am looking forward to 2010/11, when Intel will ship its 80 core cpu
for webservers.

Each to there own. I've no idea what I would do with an 80 core cpu,
probably turn-off 78 of the cores and save electricity.

Absolute legitimate. Target of this chip is not the private person but
the industry.
The funny thing is: people don't "need" faster computers or video cards
as long the hardware is too expensive. If the price falls then people
suddenly begin to argue: "Oh yes, I really need this new hardware. I
just can't continue without it."

Yes. And computer hardware companies convince them of that.

When you can buy a 80 core cpu for 60$ you might become interested as
well.

Yep.


I speak only for myself in this message.

.

Relevant Pages

  • Re: Physicists dont deliver
    ... scientists the means of constructing the technology that makes you ... I was thinking more in terms of CDs, computers, modern cars, electronics, ... Do you need your subscription to Popular Science renewed? ... dave hillstrom mhm15x4 zrbj ...
    (sci.physics)
  • Re: How to Jeep Exit a C-130 in Air Assualt
    ... in the entire universe of science, computers, logic, and technology ... that are silly enough use F1 keys is the idiot Government and IBM. ...
    (sci.military.naval)
  • Re: O.K, how do I give 98se access?
    ... computers - and will continue to do so. ... My response was not a rant - I have arthritis in the fingers and practicing ... are witnessing the shift in technology from business to domestic consumer ... Providing management methods that satisfy all is a substantial challenge - ...
    (microsoft.public.win2000.security)
  • Re: Blu-Ray to bring price hike
    ... I tend to agree with the Mayor in that a lot of this technology serves ... to do it the hard way first, and use computers and other tools AFTER ... the teachers have to be involved and trained. ... If we take technology out of schools, as the mayor of nothing, nobody ...
    (comp.sys.mac.advocacy)
  • Re: Why?
    ... There is always a market segment where one technology or the other has advantages, but this is _comp_.misc, remember? ... I can go to the computer museum to see computers built from valves. ... Your original post in April asked why the decision to use transistors rather than tubes was the right decision, ... How many companies have stuck to tubes in computers, how many have stuck to discrete transistors, how many have stuck to "dicrete" ICs, how many have stuck to AT bus, then VL bus, ... ...
    (comp.misc)