Subject: Re: Bloatware/'de fatso' standards/The faster we run, the behinder we 
	get...
From: mash@mash.engr.sgi.com (John R. Mashey)
Date: Nov 22 1996
Newsgroups: comp.arch

In article <hbaker-1511961315070001@10.0.2.1>, hbaker@netcom.com (Henry
Baker) writes:

|> The economists keep trying to understand why the massive investments in
|> computers haven't produced massive increases in productivity.  Investments
|> in 'IT' (Information Technology) are currently the 'Black Hole' of the
|> productivity picture.

I'm not sure why this long thread is here in comp.arch ...
but in the general interest of promoting rational discussion and useful
analysis that might actually fit here:

(a) Does anybody discussing this topic *actually* understand:
	- What the original statement actually means?
		If so, please post.
	- Is there a statement here that is actually amenable to scientific
		analysis, i.e., like quantification and falsification?

(b) Be reminded that anyone who simply takes numbers from economists without
understanding how they are measured, or what they mean, can get led
seriously astray ... just as people can get led astray by benchmarks
and irrelevant architectural specifications.

(c) I'd love to see a good explanation of this from somebody who knows.
Here are the sorts of questions that one might want ask, that is, if there
are not good answers for these, then the original statement is not useful:

	(1) Presumably, one can measure investments in IT.
	(2) Presumably, one has productivity measures ... but which ones:
		Productivity = output/input
		Output could be:
			Goods produced
			Services produced
			Quality-adjusted goods or services, i.e., must
				subtract problem cases from total shipped.
			Sales
			Profits
			All sorts of other measures, such as lives saved,
				# person-days of health
		Input might be:
			Per/person in producing organization
			Per/capital investment
			Per total asset base

		(Note, of course, that when people throw the terms around
		without thinking, absurd things can happen.  Suppose, for
		instance, you act like manufactured goods are what counts.
		Suppose computer companies A & B are the same size, revenue,
		but differ as follows:
		A: sells you a computer, then every years, sells you a CDROM
		with software updates. (product1+product2)
		B: sells you a computer, and as a separate service, promises
		to update your system once a year. (product + service)
		Of these two, A now looks like a much more productive
		company ... 

	(3) Averages mask distributions, and distributions are *important*.
	The following statements are *very* different:
		(a) The world, as a whole, invested $X, with no increase
		in productivity.
		(b) Some people invested, and productivity actually got worse
			(that happens, after all); others got spectacular
			returns on their investments.

(d) Personal opinion: the generic statement that IT spending is a black
hole is not very useful: it's clear that:
	(1) Some people waste money on IT spending, for negative return.
	(2) Some people spend money on IT, and accomplish crucial results,
		which however, don't show up in typical productivity metrics.
	(3) Some people spend money on IT that has *terrific* return, on any
		measure you'd likely use.

	Of course, some software has gotten bloated ... this tends to happen
	when:
		For the bulk of users, there is a level of "good enough"
		that has already been achieved, but vendors would like to
		keep selling software upgrades.
	Fortunately, there are other application areas where there is
	no level of "good enough" likely to be achievable any time soon,
	and can basically use whatever level of performance can be afforded.

-john mashey    DISCLAIMER: <generic disclaimer, I speak for me only, etc>
UUCP:    mash@sgi.com 
DDD:    415-933-3090	FAX: 415-967-8496
USPS:   Silicon Graphics/Cray Research 6L-005, 2011 N. Shoreline Blvd, Mountain View, CA 94039-7311