Re: Mainframers ... according to RW (was: Infinite Loops and Explicit Exits

On 13-Nov-2004, docdwarf@panix.com wrote:

> Your experience indicates that it is physically impossible for you to be
> familiar with what a fraction of mainframe programmers do, hence the
> request for evidence of *any* sort.

My experience also is extremely limited.    I've only worked in 4 states, wi
th
virtually all of my work on mainframes from various manufacturers.  I've onl
y
worked at a dozen or two shops out of the thousands and thousands out there.
So it isn't surprising when my experience varies tremendously from his
conclusions.

I would have to do extensive research before I could have a similar degree o
f
confidence.

In article <cnajck$hjq$1@peabody.colorado.edu>,
Howard Brazee <howard@brazee.net> wrote:
>
>On 13-Nov-2004, docdwarf@panix.com wrote:
> 
>
>My experience also is extremely limited.    I've only worked in 4 states, w
ith
>virtually all of my work on mainframes from various manufacturers.  I've on
ly
>worked at a dozen or two shops out of the thousands and thousands out there
.
>So it isn't surprising when my experience varies tremendously from his
>conclusions.
>
>I would have to do extensive research before I could have a similar degree 
of
>confidence.

Other folks do other things, of course... but please pardon my imprecison
in what you've quoted; of course Mr Wagner is familiar with 'what a
fraction of mainframe programmers do', the question is the significance of
this fraction.

DD

On 15 Nov 2004 11:21:45 -0500, docdwarf@panix.com wrote:

>In article <cnajck$hjq$1@peabody.colorado.edu>,
>Howard Brazee <howard@brazee.net> wrote: 
>
>Other folks do other things, of course... but please pardon my imprecison
>in what you've quoted; of course Mr Wagner is familiar with 'what a
>fraction of mainframe programmers do', the question is the significance of
>this fraction.

It depends on the deviation in the population. The smaller the
deviation, the more likely a sample represents the population (plus or
minus the confidence interval, CI). Suppose, hypothetically, we were
sampling a batch of data files (somehow) known to be identical. In
that case a sample size of one would tell us everything about the
batch.

The intelligent questions are:
.. How much deviation was seen between samples?
.. Were the samples biased? In other words, were they all taken from a
specialized subset.
.. Was the finding positive (it was in this case) or negative? A
negative finding requires many more samples.
.. Might there be a third factor causing both variables? In this case,
self-selection and employer-selection would be a likely suspects.

In article <hulip0logb9765jkq22pc5b0g8jrjj8frd@4ax.com>,
Robert Wagner  <spamblocker-robert@wagner.net> wrote:
>On 15 Nov 2004 11:21:45 -0500, docdwarf@panix.com wrote:
> 

[snip]

>Suppose, hypothetically, we were
>sampling a batch of data files (somehow) known to be identical.

Suppose, Mr Wagner, we try not to compare human beings, with their
associated complexities, to 'a batch of data files (somehow) known to be
identical'.

DD

On 15 Nov 2004 20:43:25 -0500, docdwarf@panix.com wrote:

 
>
>Suppose, Mr Wagner, we try not to compare human beings, with their
>associated complexities, to 'a batch of data files (somehow) known to be
>identical'.

Tell it to A.C. Nielsen and others who use statistics on human beings.
Tell them 'I'm unique, just like everyone else.'  Also, stop taking
prescription drugs because they were all validated using statistics.

In article <c66jp0l51s6vraqafqjg0265tea5mn0hkl@4ax.com>,
Robert Wagner  <spamblocker-robert@wagner.net> wrote:
>On 15 Nov 2004 20:43:25 -0500, docdwarf@panix.com wrote:
>
> 
>
>Tell it to A.C. Nielsen and others who use statistics on human beings.

When this person posts to comp.lang.cobol I just might do that, Mr Wagner;
until then you've made a supposition and I countered it.

>Tell them 'I'm unique, just like everyone else.'

When I think such a thing is worth telling, Mr Wagner, I just might do
that; until then I suggest that the 'we' you mention above try not to
compare human beings, with their associated complexities, to 'a batch of
data files (somehow) known to be identical.'

>Also, stop taking
>prescription drugs because they were all validated using statistics.

When such statistical studies, Mr Wagner, compare human beings to 'a batch
of data files (somehow) known to be identical' I just might do that; now
that you've tried to throw up some dust to obscure the issue the issue
remains: of course you, Mr Wagner, are familiar with 'what a fraction of
mainframe programmers do', the question is the significance of this
fraction.

DD

On 19 Nov 2004 11:54:36 -0800, riplin@Azonic.co.nz (Richard) wrote:

>Robert Wagner <spamblocker-robert@wagner.net> wrote
> 
>
>That may well be true. Perhaps the Grade A shops wouldn't hire you, or
>didn't need contractors - _because_ their employees were motivated and
>progressive.
>
>You seem to imply that you personally are the _only_ programmer above
>'Grade C', at least in IBM shops.  You have stated that your
>statistics are faultless and thus _every_ programmer in IBM shops is
>two levels below your standard.
>
>Actually, I believe the real reason is that any shop above Grade C
>wouldn't let you touch their code with a barge pole.  That fits your
>'evidence' exactly.

You have no evidence for any of these claims. You're making up 'facts'
to support a pre-formed conclusion.
 
>
>Strange that, they seem so much like you in that respect.

I don't pre-judge skill based on nationality or other social factors.
Some of the best code I've seen was written by people who seemed
unlikely candidates based on meeting and talking to them. Conversely,
I've met people whose convincing talk was given lie by their crappy
code.

Truth is in code. All the rest is just rhetoric. I've posted lots of
code here. I wish others would do the same.

Robert Wagner <spamblocker-robert@wagner.net> wrote
 
>
> It depends on the deviation in the population. The smaller the
> deviation, the more likely a sample represents the population (plus or
> minus the confidence interval, CI). Suppose, hypothetically, we were
> sampling a batch of data files (somehow) known to be identical. In
> that case a sample size of one would tell us everything about the
> batch.

This is exactly why your opinion of yourself differs from what others
may think of you.  It is quite true that that if there is a small
deviation then a small sample represents the population.  The problem
is that you _start_ with your preconceived _conclusion_ that 'all
mainframe programmers are the same' and then use that to justify the
vast generalisations that you make from inadequate sampling.

> The intelligent questions are:
> .. How much deviation was seen between samples?

Given that the samples you report and samples reported by others, and
by my direct observation, then they show a large deviation.

> .. Were the samples biased? In other words, were they all taken from a
> specialized subset.

The samples may also be biased by, for example, you being selective
about which programmers you include as 'mainframe'.  If they don't fit
your preconceived notion about how you want to characterise them then
you may put them in the 'PC' categorey but working on mainframes 'by
mistake'.

> .. Was the finding positive (it was in this case) or negative? A
> negative finding requires many more samples.

There yer go then.

> .. Might there be a third factor causing both variables? In this case,
> self-selection and employer-selection would be a likely suspects.

Exactly. Your employers may be selecting the sites that you work in
and thus the samples you are exposed to for much the same reason that
they have the employees that you see. You may be working in decrepet,
old-fashioned, poorly programmed sites, _because_ of your particular
qualities and experience.

Or possibly, not working in modern, progressive sites for those same
reasons.

Robert Wagner <spamblocker-robert@wagner.net> wrote

> Suppose, hypothetically, we were
> sampling a batch of data files (somehow) known to be identical.

The problem is Robert, _How_ did the tapes get to be 'known' to be
identical ?  That is the bit that you just gloss over.

In the case of 'mainframe programmers', it seems that you 'know' they
are all identical because it suits your political agenda, you 'made it
up'.

On 22 Nov 2004 02:27:33 -0500, docdwarf@panix.com wrote:

>In article <1ri2q0daadhjft28akko2cnf3spp0i5922@4ax.com>,
>Robert Wagner  <spamblocker-robert@wagner.net> wrote: 
 
>
>Mr Wagner, who posted that 'Russian programmers don't do well because
>they're so opinionated' above?  That certainly seems to be a pre-judgement
>of skill based on nationality... what am I missing here?

I did not comment on their programming skill.