Notes

Back in 1962, most programming was done using punch cards. This method involved a piece of paper that a programmer would punch holes into. In this form you could translate the code to card as the holes in the card represented a one or a zero. On average, each card held 80 characters thus anything long form took thousands of cards and mistakes could not be spotted by looking at a screen or modern syntax highlighting. To check the validity of a card you would need to punch another card to compare to the first using a card verifier. A program therefore was quite hard to validate, as a card verifier is not efficient, leading to any possible error such as this one to be left unchecked. In turn, a transcriber who hand wrote the guidance program missed the ‘R dot bar n’, which lead to the person programming the punch cards to not include the vital symbol. When the cards were read into the program this defect was carried over. The fault came to fruition when the antenna failed, in which the code, took over the role of navigation, that possessed the missing overbar.

Good software stems from good practice of employing principles to develop rigorous and polished systems. A defect was left undetected throughout the prior Ranger launches, and only upon failure of the hardware and reliance on the software, did the defect turn into a failure. Maintenance is a critical aspect of software engineering practice as it is the continual development of code after deployment. Because the program had been previously used on the Ranger this shows a violation of practice through lack of maintenance on the code. As mentioned earlier about the failure of debugging, this also shows shallow testing of the code. If they had tested for the outcome of the failure of any hardware, namely the guidance antenna, the code would have been maintained to an operational standard, and the fault, would have not occurred. Modern unit testing would’ve been a valid way to find this issue, as output of the hardware signal and the software guidance system could be compared to see if there was an error in either.