In the US, where ubiquitous computerisation has long since become a fact of life, two applications have thrown the problems of closed-source into sharp relief. The first is the Diebold electronic voting machine. The principle is straightforward — it displays the available options to the voter, records the votes and sends them to a central counting-house to be tabulated. In practice, the Diebold is an overcomplex, labyrinthine machine, the correctness and integrity of which have been called into question by several watchdog bodies.

Now that should be easy to resolve — simply appoint a taskforce to audit the machine for correctness and security, ensuring that it does what it is supposed to do, and nothing it is not supposed to. Well, it would be simple — except for the fact that the Diebold is closed-source. There is no transparency in how it works — the disturbing, terrifying fact is that a private company can, through malice or incompetence, invisibly influence an election in one of the world’s most powerful countries.

Exhibit number two is the breathalyser, a device which estimates the amount of alcohol in a person’s bloodstream. It is commonly used as a “field sobriety test”, and its results are admissible in court. Since a person can be convicted on the basis of a breathalyser test, it is imperative that it be above reproach. In practice, however, how the sensors are read, and how their readings depends on software that is, again, proprietary and closed-source. Several breathalyser manufacturers have indeed been taken to court and asked to turn over their code — which has been found to be buggy and error-prone. Others have resisted on the grounds that revealing their source would “compromise trade secrets”, something that, sadly, seems to outweigh human rights in much of the world.

... contd.