Documentation IS Rocket Science

As some of you may know, I am an avid space fan.  Larry Klaes (candidate for GBU? - an inside HCI joke) in his mailing group has pointed to an article detailing the issues of reviving the Saturn 1 moon rocket to power the next stage of human space exploration.   Reading the entire article is highly recommended and it can be found here.  It basically details the issues involved when compilling documentation for the retirement of a project that at some future date may be reinitiated.  The problems of incomplete documentation, tacit knowledge in the scientists's heads (to be redundant) and lost documentation and tools for supporting manufacturing tasks parallel issues we have today for many of our large software technical artifacts. 

One advantage software engineers have is that we have very low manufacturing costs for multiple copies (clearly not the case for the Saturn 5 F1 engine).  This clearly helps in keeping old software running.

Some interesting discussion topics would be how do large software projects differ from large hardware projects, beyond the obvious manufacturing issues?  What can we do to better preserve documenation and the knowledge that swims in the heads of the designers?  Is there something beyond wikis?

I have been remiss lately in posting -- I need to catch up to meet my goals for an entry a week. Hopefully soon as the Spring semesters begin to wind down. Later!

Comments

[Arun Hegde]

That was a interesting one on infact brilliant. If you like space, cosmos I am sure you will like this link too. http://web.mit.edu/newsoffice/2006/planet-birth.html
Even I a small fan of Space and alien objects.

Cheers
Arun

[Rob Van Dyk]

These days, design review packets packaged in PowerPoint (try saying that ten times fast) serve as almost semi-useful documentation. :)

===

The article has a point on losing knownledge of legacy systems, though...

Being able to come-up-to-speed on a decade or more of prior work has been something that I've come against at my present job.

Heck! After a recent corporate merger, my company's name lost a significant namesake (see: http://en.wikipedia.org/wiki/Guglielmo_Marconi, http://en.wikipedia.org/wiki/BAE_Systems) which dates back to the foundation of the industry [btw: I consider it serendipitous that I was even able to link Guglielmo Marconi's work to the present day BAE Systems].

In my view, this renaming intentionally obliterates a very critical bit of documentation. How long will it take before wireless technologies are completely disassociated with Marconi (and the rest of the people who helped discover it)?