The software for things

I have been musing about the intersection of the internet of things and software maintenance.

The internet of things has already had issues with poor security and the lack of software updates, with DDOS attacks being mounted from things. Economic factors work against companies fixing the software and making an update — as the sale has already been made, so updating a thing is just a cost. In addition the interface for updating a thing is usually dire compared to the process of updating software on relatively fat systems, such as phones and notebooks (which are tricky enough).

Compounding the issue is the need to maintain software on the things for a long time, to update the like of encryption services and removing flaws in protocols. Five years is considered a long time in software, but it is just the warranty period for a kitchen hob or a (normal, stupid) LED light bulb. The software industry, as a whole, does a terrible job of maintaining old code. Normally most of the old code is thrown out and the coding started again, shipping new code on new hardware. However, shipping new hardware is impossible with things, as the devices are already out there. On the up side, software like web browsers are continually updated, so there is hope.

Currently things are built around the view of a personal computer: a computer for each device and a network to bind them. I wonder if this view is hampering us, and we need to rethink what a computer is and how it is all assembled, distributed, and made to fit around the lives of people. I also have a nagging feeling that maybe we should be looking more seriously at GNU Hurd, but that may be a totally unrelated fever-dream.