Re: Does software-engineering focus on the wrong subject?
From: CTips (ctips_at_bestweb.net)
Date: Mon, 20 Dec 2004 17:38:22 -0500
Bjorn Reese wrote:
> CTips wrote:
>> We know that there is a huge difference in productivity between the
>> best and the average programmer - anecdotally, the spread is something
>> like 20x to 50x. But very little literature discusses exactly how to
>> improve an average programmer.
> Numbers like these are not supported by empirical data. See for example:
I'm actually aware of both the 1:28 claim, and its various refutations.
However, I'm speaking from personal experience, and from discussions
with colleagues. And I'm referring to the entire process - from
requirements through analysis and design to impementation, and including
testing and test-environments and maintainence and maintainability - not
just to maintainability. The difference of overall productivity is not
just the implementation (where may only be a small reason for the
difference) but in the overall design and architecture, and in the
maintainability of the final product.
In a previous job at a big company, two people put together in 3 months
a tool which was somewhat similar to a different tool that had taken 12
man years+ to develop. Its not exactly a perfect comparison, because
each of the tools had different strenghts, but they are close enough
that I would feel comfortable stating that we got a 25x productivity
>> Now, why isn't there much done on the subject? Is it because most
> Well, researchers like Lutz Prechelt, Janice Singer, and Timothy
> Lethbridge springs to mind (there are more, but I do not have a
> comprehensive list.)
It would be nice to get some specific papers where they show how to
I looked at the Prechelt's home page, and there were too many papers to
be really go through.
Just out of curiosity, I looked at his paper "An empirical comparison of
C, C++, Java, Perl, Python, Rexx and Tcl"
and decided to implement the phonecode program based on the problem spec
in page 2. It took me about 2hrs, 15min. I then looked at the results of
the 5 C programmers; it looks like they ranged upto 16 hrs; thats a 7x
difference w.r.t. masters students in CS (I have no clue wether they
were average or above average programmers). The solution is in