Re: How to make money fast (in science fiction)

On 2008-04-03, Michael Ash <mike@[EMAIL PROTECTED]
> wrote:
> Booting to a GUI has always taken quite a while. This is the sort of
> thing I mean where the speed cost is actually due to more stuff
> being done, not just greater complexity that isn't directly useful.

Yes, I agree.  Though it's still interesting that even the hardware
takes a lot longer to initialise than the more primitive stuff used
to.  It takes about ten seconds for my current computer to even start
looking for a boot-up disk.


> Never say never. A lot can happen before the heat death of the
> universe.  It's entirely conceivable, although beyond anything we
> currently know to be theoretically possible, that ways would be
> found to perform infinite computation in a short amount of time.

I was even including infinite computational power.  For any given
complexity of "oracle" machine there are problems that it cannot
solve in finite time.


> For boring finitely-fast computers like we have today, it's probably
> better to reformulate the question in terms of the percentage of
> users for whom the computer is effectively instantaneous.

Yes, perhaps this would be a better way to phrase it.


> Anyone whose computer usage is limited to driving a car and using a
> phone is experiencing this today. Many people with "e-mail
> computers" and other such low-intensity users probably also fall
> into this category.

Probably not.  As an example, Outlook Express is perfectly capable of
taking many seconds to respond to common user actions even on a quite
decent machine under not uncommon cir***stances.

Also I was just experiencing today the way in which very simple use
can bring a dual-core 2+ GHz computer to its knees just as a result of
web browsing.  In particular, via an emailed link to a webpage so full
of animated craplets hogging the CPU and RAM that there weren't even
enough resources left over for the user to close the window.

Yes, that is a result of very poor software design.  I don't expect
hardware improvements to make the problem go away, though.


> If available computing power continues to increase as it has then
> there would eventually be a point where only a small fraction of
> power users actually have to wait for anything to happen on their
> computers.

I expect it to be much more likely that instant software will be
driven out from the bulk market by software that is tolerably slow for
most tasks, but has "better" bells and whistles.  And in that software
there will likely be features that require more processing still, and
the user will have to wait for those to complete.

What's more, as speed and capacity increases, the range of
applications that are viable in the mainstream also increases.  I
think there will always be things that people want to do that push the
capabilities of the systems of the day.  I know I've many times heard
people ask "why can't computers do X", where X is something that would
require a lot more processing power than we have.  There must be a lot
more people who think it but don't ask it, and vastly more still who
would use such features if they were there but a bit slow.

Don't forget the role of networking either.  More and more software
relies on increasingly wide communication patterns.  Rightly so in
many cases (but not all).  Unless we can crack FTL, latency for
worldwide communication will always be noticeable.


- Tim