Dan Ingalls wrote:

> No sad words from me

Oh, I was recriminating my own gloomy mood.

> Now why haven't we made such amazing progress in software?  Huh?

One reason is that we have been sidetracked into a dead end path:

   http://www.dreamsongs.com/WorseIsBetter.html

Bob Jarvis asks:

>         Why and how does hardware get faster?

Mainly because the integrated circuit fabrication process is being 
improved continously. If you were to build a 0.13 micron 8086 it would 
be 100s of times faster than the original simply because the smaller 
the transistor, the less charge you have to move around to switch it 
(the faster it goes, ignoring wires). There has also been some 
improvement in microprocessor design, so the Pentium III is a few times 
faster than that modern 8086 would have been (two instructions per 
clock vs several clocks per instruction), but that is mostly an 
indirect consequence of the process improvements (we can fit many more 
transistors in the same area, now, so "wasteful" designs are possible).

> and
>
>         Can we do similar things to improve software?

Yes and no. Faster hardware makes the old software fly (try Smalltalk 
V/286 on a 1.2 GHz Athlon machine...) but

Michael Rueger commented:

> Well, we have: todays programms run 100 times slower, 100 times more
> memory and in color ;-)

so we don't care about running old software better. Which is why the 
answer is no: most of the added transistors we get work in parallel 
while most of the added software is serial (the more there is, the 
slower it goes).

-- Jecel