RE: Bad Business (was Re: advocating XML)

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

-----Original Message-----
From: David Megginson [mailto:david@megginson.com]

>Then they make a wild guess, and are nearly always wrong.  

I'd be lieing if I didn't account for the utility of well
practiced instincts.  But they don't make a simple wild 
guess, not the competent ones.  They engage their peers 
in a discussion, then they guess.  

>Projectplanning in the first decade of the 21st century is about as
accurate
>as medical diagnosis in the first decade of the 11th century.
>Sometimes we get it right -- generally by accident -- but usually our
>treatments do more harm than the original disease.

Umm... no.  Not the ones who stay in business.  Accuracy may 
vary, but if one knows their market, stays focused on it, 
and doesn't do too many wild projects, it is a reasonable 
task to plan and make profits.  That focus is realllll important.

>The truth is that people are not smart enough to plan that far ahead
 
Maybe that's just the Canadians. :-)  Oops, no I forgot 
about the Big Dig in Boston.

>(look at the US stock markets for another, similar example). 

The US stock markets and the world markets are perfect example 
of what happens when simple solutions get sold 
for complex tasks and fall on their bums.  The real dilemma 
is belief and amplified hype.  Hysteria and fads are common 
problems of intense feedback.  I haven't lost a dime on the 
Internet stocks.  Planning wasn't required. Being knowledgeable 
of the history of the technology, what had been tried, what had 
failed, what were the contexts (one can learn from accurate 
history;  one only guesses if the history was systemically 
distorted) worked just fine.   In the last year, my investments 
have doubled and are increasing at a fairly even rate.  Why? 
Fundamentals of Fair Dinkum.  Buy not on expectation but on 
capability, market share, assessment of management strengths, 
etc.  It isn't that hard to make a reasonable profit.  Greed 
on the other hand, well, that is gambling.  Pleasurable but 
somehwat unreasonable.  Most of the people out their losing 
their condos and Ferraris should reevaluate their analysts.  
Their analysts should reevaluate their sources.

Lies and statistics....

>Now, for a six-month project, she has a 53% (0.9 ^ 6) chance of being
>right.  That's not so good -- if she thinks we'll need XML in six
>months, should we risk investing, say, $750,000, when there's almost a
>50% chance she's wrong?  

If all you have is that statistical rule, you will react like any 
superstitious person does.  The failure is not knowing what XML 
is good for and knowing if you will have a good requirement in 
six months.  That is easier than doing superstitious math.  

I understand what superstitious acquisition means in behavioral 
systems:  reacting to a signal which has no coupling to the 
behavior.  ACK!  She can't breathe underwater.  She's not a 
witch.  Too bad she's dead, but her soul is on its way to 
heaven and she is really better off.  We did her a favor. Next!

>In the real world, where most project managers aren't this good
>(they're lucky to hit 90% for the next day), things get much worse.

The issue is competence over good information.  Both are expensive 
to obtain but not hard to apply. 

>The best solution is to admit our limitations rather than strutting
>with false confidence while we bleed our patients, slap on leeches,
>put them on all-mustard diets, and carve the shape of the cross in
>their foreheads (or the modern equivalents, write 200-page functional
>specifications and draw incomprehensible Gant charts in Microsoft
>Project).

Bad data in a good format is still superstition.  I agree. 
On the other hand, good data in an incomprehensible format is just 
as useless.  The only reason to do the analysis is to see if there 
is a pattern.  Are the metrics good metrics or is the tool in 
the way? Is the pattern real or superstition?

>People -- even smart, well-meaning people -- really, really suck at
>project planning, especially in an area like high tech where things
>change so fast.  

Sure but they still have to cost the problem.  Fixed price contracting 
demands it.  So they do their best and many people are very good at 
it and get better with practice.  Hire these people.

>Even in the dirt-and-concrete world (where things
>change slowly and it is possible to plan further ahead), I'd guess
>that about 50% of roads projects around my hometown of Ottawa miss the
>mark completely.

Does the word graft come into the accounting?  I agree, the larger 
the project, the more the cost overruns are likely, but it comes 
down to competence.  We watched the German rocket scientists build 
machines that ran the first time, every time.  Why?  They'd done 
it a lot and they built the first one locally and measured the 
costs.  When they contracted out the rest, they sent the the 
instructions and the precise costs.   It worked wonderfully until 
Nixon declared this just wasn't the American way and turned good 
engineers into managers of engineers on the other side of the 
country.  Then the game was afoot and the cost overruns became 
legendary.

>This is something Kent Beck and the other XP people got right -- build
>for today, not for tomorrow.

And tomorrow they can build it all over again.  And the next day. 
And the next.  Nope, not on a fixed price contract.  I don't care 
what method you use as long as you hit the marks.

Von Braun got it right.  Build one and cost it.  Sell the heck 
out of that one.  MS got that right too.

They want what they want. If they can afford what they want, 
they get that.  They don't get more than that.

len