Mailing List Archive


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [tlug] [OT] Say _no_ to the Microsoft Office format as an ISO standard



Edward Middleton writes:

 > I presume you are talking about People in general because a lot of
 > people do say no.

Yes, more specifically those who don't say no.

 > Thus we are talking the average person, who doesn't get down and
 > read the source.

No.  We have no idea whether the yes-man we're talking about is
"average", or if he reads source when available or not.

And the choice to never read source can be the correct one if the cost
of learning to read source, or of your time, is sufficiently high.

 > The uncertainty isn't about whats in the source so much as whether
 > it is the best product to fulfill their needs.  I don't believe
 > most users are well enough informed to make that kind of decision
 > about software most of the time.

If *they* don't know, what makes you think *you* or Linus or Theo or
Richard knows better?  Getting the information about the "best"
product can be very expensive; it often makes sense to just take "good
enough" off the shelf at Fry's, even if you are subject to lock-in.

On the other hand, there are plenty of large companies with IT
departments charged with making good decisions about company-wide
software purchases.  I know we all love to deride the intelligence of
the "suits" and the "Microserfs" who make those decisions, but I have
to suspect that maybe they know something we don't know.
Specifically, their employers' needs.

 > Is obesity (1) or (2),  is gambling (1) or (2),  is living off debt (1)
 > or (2), or are these all "good for him"?

Obesity is not good for me, but I like to eat and drink with gusto,
one of life's little tradeoffs.  The fraction of people who gamble to
excess or live off debt is small enough that I'm comfortable in
putting them in the "cognitively impaired" class.

Note that both gambling and living off debt have the property of
lock-in; a small amount of cognitive impairment and/or lack of
information can leave you in a position where it makes sense to double
your bet or your debt, although you now know the odds are against you.
This is much less true in the case of computer software or hardware,
where you can change your mind every year or two.

 > > So despite the enormous attraction of behavioral economics for people
 > > who dislike conventional economics, the biases[1] that evidence
 > > strongly suggests actually affect ordinary people (from the janitor to
 > > the CEO) in real life are few and small.  The most important ones seem
 > > to be second order (eg, the Allais paradox).  The Kahnemann-Tversky
 > > "prospect" theory doesn't seem to have any behavioral predictions at
 > > the market level that can be distinguished from ordinary risk
 > > aversion, for example, although it is clearly verified by
 > > introspection, and weakly confirmed in laboratory experiments.
 > 
 > Am I right in saying that the crux of this argument is that risk
 > aversion not irrational biases are the basis for consumer software
 > purchases?

No, the argument is that we don't know what consumers are doing, but
there's no compelling evidence that conventional economics is
consistently inferior to some other model.

 > If so, is there any evidence of this.

Wrong question.  The question is, "is there a model that is either (a)
significantly better at describing behavior, or (b) is easier to apply
to construct larger models?"  People like to eat, so they discount the
risks of overeating, and become obese.  Just because obesity has a
dark side doesn't mean that becoming obese is irrational (according to
economics).  "Rational obesity" is a working model.

 > How many windows sysadmins list security as a reason for not
 > deploying Linux?  Is this risk aversion or irrational bias.

Most likely it's organizational incentive, ie, "Nobody ever got fired
for buying Microsoft."  In other words, if you define "security" as
"no varmints get in", you may prefer to learn about Linux.  But if you
define "security" as "no varmints get in that I can be blamed for",
you may rationally prefer to stick to the product you know, which
offers fewer ways for you to go wrong.

The problem with behavioral economics is that you look at people doing
something you know is stupid and self-defeating, and you go, "OK,
Windows sysadmins by nature are stupid and self-destructive."  If you
start by assuming they know what's good for them, *then* you can
uncover the organizational incentive aspect.

Of course you can do both, but I see no advantage to doing that.


Home | Main Index | Thread Index

Home Page Mailing List Linux and Japan TLUG Members Links