Subject: Re: The value (?) of popularity
From: Erik Naggum <erik@naggum.no>
Date: 1999/09/29
Newsgroups: comp.lang.lisp
Message-ID: <3147591653434816_-_@naggum.no>

* Dobes Vandermeer <dobes@mindless.com>
| But the decrease in quality comes with the apparent advantages of growth
| and prosperity.

  yes, that's exactly it: apparent.

| Without interest there is only disappearance and death, and this makes
| popularity seem essential to me.

  popularity is essential for the mass market.  the sustainability in other
  markets, of which most are far more interesting than the mass market, is
  damaged by popularity, meaning interest from the general population.  for
  instance, enormous markets are dependent on company-to-company sales and
  would be completely unable to educate enough people to deal with regular
  joes as customers.  these are the professional markets.  I stay clear of
  the consumer markets whenever possible, and go through a lot of work to
  get professional products.  the really sad thing is that they often cost
  _less_ than the consumer products, but if they volume went up, so would
  the price.  for instance, doctors, dentists, and vets use this highly
  absorbant material to clean up leaking body liquids of all colors and
  contents, and I got a bunch of this material when my cat had an operation
  and she did not have bladder control while recovering from anesthesia.
  so of course I bought 10 lbs of this stuff from my vet -- it far exceeds
  the absorbancy of _any_ of the consumer paper towels on the market and a
  box of 10 lbs costs the same as about 4 lbs of consumer paper towels.
  now, the reason for the low price and high quality is that people don't
  know about it, don't need to buy small units of it that are convenient to
  households (it comes in stacks of individual sheets, not rolls) in all
  sorts of stores, and there's no marketing for it at all.  which means: no
  sales force to talk to all the stores or chains that might stock it, no
  shipping costs to absord into a retail price (the purchasers pay for the
  shipping from the one outlet in the entire country), etc, etc, etc.  this
  stuff would end up costing more than five times as much as it does today
  if it were to hit the consumer market (of course I asked the people who
  make it).  which probably explains why the paper towels are all low
  quality: the cost of producing a roll of paper towels is less than 10% of
  the price customers are paying for it.

| LISP is by no means on that path (yet), but it seems (to me) to be almost
| a general opinion that it is not as popular as we'd like, either.

  Lisp is a professional product in the professional market.  it would be
  very seriously harmed by becoming a consumer product.

| I would hesitate to say that I am completely happy with them, but I
| prefer that computers are popular, and thus readily available and
| growing rapidly, than that they would have "fallen to the wayside"
| because they lost a battle with a competing inferior technology...

  huh?  Intel and Windows _are_ inferior technology and they're winning,
  leaving all the quality designs by the wayside.

| I can tolerate the lowering of quality on the microscopic scale because
| of the increase in quality overall.

  but we have a decrease in overall quality!  40 years ago, producing a
  business letter had a cost relative to company expenditures in general
  that was about half of what it is today.

| The choice in CPU on a machine only seems relevant for assembly-language
| programming and nitpicking;

  really?  ever heard of any of the numerous Intel bugs?

| Standards such as CL, CLIM, and CLOS protect us from the details of the
| inferior processor we are running on and allow us to work in a more
| comfortable programming environment whatever CPU or operating system lies
| beneath.

  this is not true.  somebody has to care about the CPU to make the
  software run on it.  making highly optimized code for something so
  braindamaged as the Intel instruction set, which isn't even relevant to
  optimization, anymore, as the underlying parallelized microcode-like
  instructions have everything to do with the pipeline, cache, etc, is so
  hard that it takes _years_ of effort to get it optimal.  I have seen code
  that ran 8 times faster just by reordering the instructions on a Pentium
  II.  stuff like this matters a lot more than people would like to believe
  when they use high-level languages, because the computer doesn't go away
  just because we don't _have_ to worry about it all the time.  we _should_
  worry about it, but not all the time, and not for everything we do.

| Thats part of what attracted me (and others) to LISP, is its rich library
| of abstract constructs, and the powerful ways you can add and extend the
| library with minimal effort.

  all fine and dandy, but it has to run on a real CPU, and there's a
  dramatic difference in CPU's that make running anything useful on them
  easy or hard.  this translates to sluggishness of development and
  adoption of new instruction sets.  look at the SPARC -- it has managed to
  upgrade itself 6 or 7 times, and have changed a lot in the meantime,
  because the users were professional users who upgraded everything or were
  satisfied with software emulation of various instructions.  the Intel CPU
  can still run 8086 code, because consumers hold on to old software even
  when they upgrade the hardware, and the third-party investment in the old
  instruction set is so enormous that they couldn't change it dramatically
  without risking that nobody would follow them.  this leads to CPU's that
  are _very_ expensive by today's technological standards.  done right, a
  new chip could cost a lot less and run a lot faster, and we have it:
  Digital Equipment Corporation (rest the blessed soul) produced the first
  processor to break the 100MHz barrier, and it's still amazingly fast.  of
  course, with despicable Compaq buying it all up, it's going to run the #1
  sluggish software in the world: NT, so it basically runs just as slowly
  as Linux on a computer two Intel generations ago.  that's how the world
  doesn't win through popularity contests.

#:Erik