Re: the gain-maximizing flaw

[Glenn Bacon <gcbacon@roadrunner.com>]

Roberto,

Some thoughts on your recent communications:

I find it hard to have a polarized argument about risk-min. vs.
gain-max. It's like having an argument about inhaling vs. exhaling.
All systems that I can think of do both.  This is especially clear
when the issue is posed as efficiency vs. reliability.  The human body
strikes me as a wonderful example.  It is incredibly efficient at
converting nutrition into energy and the immune system is an
incredible reliability mechanism.

To carry this view further; when one walks down the street the
efficient thing to do is to walk in a straight line, traversing
manhole covers.  The risk-min. thing to do is to go around every one
and also look out for falling building facades. How do you usually
behave?

I suggest that most programmers who contributed to this problem had
little reason to believe that there would be a problem with system
manhole covers an facades in the future.  In their minds, there was no
choice which compromised reliability in favor of efficiency.  (You'll
miss the point if you characterize this argument as just more
"defensiveness" by old programmers.)

I agree that corporations, as well as most other entities, gain-max.
But just the term gain-max. does not adequately characterize a
decision process. Optimization efforts involve both objective
functions (e.g.. gain-max.) and constraints (bounds on the space of
feasible solutions).  The real issue involves constraints.  Most of us
wish the gain of health, wealth and happiness, but our values
constrain us from rape, pillage and plunder to obtain them.

For reasons stated above, I think all of this has little to do with
Y2K.  It does have a lot to do with larger ecological, financial and
social stability issues.  I find it hard to reduce this to a few
"rule-sets."  Terms like, awareness, connectedness, compassion and
responsibility seem more relevant.

Glenn Bacon

Roberto Verzola wrote:

> My reply to Glenn Bacon:
>
>  >I believe the major hypothesis behind this forum  may be
>  >"deeply-flawed."   The Y2K issue has little to do with gain-max.,
>
> It is perhaps too early to make that conclusion, until this earlier
> debate about gain-max. and the Y2K is settled. But let us keep that
> possibility in mind...
>
>  >beyond the age of punched cards, the cost of memory or storage was
>  >an insignificant issue in the propagation of 2-digit year fields.
>
> In large software systems perhaps. But in controllers and embedded
> systems, maximizing ROM- and RAM-space remained an important
> consideration in the 80s. Even with the PIC microcontrollers of the
> 90s, the limited memory space available makes space savings a
> significant issue, and this can still tempt a programmer to maximize
> space by using 2-digit year fields.
>
>  >To argue that Y2K is the result of corporate gain-max. requires that one
>  >show that 2-digit year fields were a conscious policy decision by
>  >corporate leaders.  I doubt that anyone can find such evidence.
>
> Paul Swann just posted an account of how the Pentagon set the
> standards (2-digit year) for DOD projects, and big business had to
> follow suit. The DOD wanted more "bang for the buck", and I would
> presume that big business conformed because they wanted those DOD
> contracts.
>
>  >Gain-max. may have something to do with the rate of remediation once the
>  >problem was visible and credible.  But I would argue that denial was the
>  >dominant issue.  There is a story that when Clinton was told about the
>  >Y2K issue he didn't believe it was significant because no country with
>  >great companies like Microsoft and Intel could possibly have such a
>
> But I would guess that when Clinton saw the estimated costs of
> remediation, that's when he finally said, "no way!" Externalizing
> costs (by postponing them, for example) is one way of maximizing
> present gains... Perhaps, denial, as you say, has something to do with
> it. But why deny a problem that your experts tell you exists? I
> suggest that the motive is to avoid incurring the costs. If it cost
> very little to solve the problem, I'd guess that Clinton would have
> agreed without thinking.
>
>  >The Y2K management litterature is full of advice on how the CIO can
>  >convince the CEO there is a real problem and get the extra funds
>  >needed to fix it.  Big time Denial!
>
> You said it yourself: the CEO would not provide the extra funds and
> was into big time denial. Why? Obviously, to keep his costs down.
>
>  >Gain-max. is such an intrinsic characteristic of living things
>  >(among many other behaviors) that it is impossible to separate it
>  >from Y2K.
>
> Now, this is an interesting observation, and I agree with you. Living
> things include gain-pursuit among their behaviors. But I will add:
> their behavior also includes risk-avoidance. And I will further assert
> that risk-avoidance will generally have higher priority over
> gain-pursuit among living things. When animals sense both food and
> danger, it is quite obvious how they will generally react -- more so
> if the animal had a recent meal and is in no risk of dying from
> hunger. (I stay away from the max./min. terms because I am not sure if
> individual animals have the computational ability to do such things. I
> would guess that not even human beings use such computational ability
> to the extent that corporations do.)
>
> The alternative I propose is to tilt the balance back in favor of
> risk-avoidance (and risk-minimization, for those who care to apply
> their computational prowess) in our own rulesets, and not to let the
> pure gain-maximizers to rule our lives.
>
>  >But for those who are ideologically committed to flogging
>  >multinationals as gain-maximizers, there are probably better horses
>  >to ride than Y2K.
>
> It is not a question of flogging multinationals but of stating a
> truism that every business and economics student learns by heart, and
> which you can either confirm or disprove: are corporations
> gain-maximizers or not? I assert that they are gain-maximizers of the
> highest degree.
>
> Perhaps you can appreciate the gain-max. vs risk-min. debate better if
> I restate it as a debate between efficiency and reliability, which you
> must admit is a recurrent debate among programmers/systems designers.
> You must have your own stories to tell about how some programmers just
> love to write "efficient" routines using all kinds of tricks and
> short-cuts. (Alan Greenspan  -- now chair of the Federal Reserve --
> apparently was such a programmer, according to the article posted by
> Paul Swann.) And you must know how the debate has been generally
> settled. In instances where the debate is settled in favor of
> efficiency rather than reliability, the result is often a buggy,
> unreliable and crash-prone system. The Millennium Bug, of course, is
> the most spectacular example.
>
> Roberto Verzola