Bloggo back to the blog
Discovering Context: the first Titanic failure – by Erik Petersen-->
After a century, we have apparently discovered the context of the sinking of the Titanic. Rather than simply being an iceberg in the way, it appears human error is to blame. I always wondered how an unsinkable ship created at the peak of many centuries of ship design, and known to sail through icy seas, could sink in such a way. Apparently the cover-up happened for legal reasons, and is now being rewritten by the granddaughter of the only surviving officer.
The chance of the error occuring was increased by a fundamental non-intuitive change to an interface, human habit and binary mistakes. These are all issues that can still affect technology today, and are just as challenging.
When airline pilots convert from Airbuses to Jumbos and viceversa, they flip switches in the opposite directions. This is just a fundamental design difference. Luckily the joysticks stay the same. When ships converted from sail to engine, the action of the steering wheel (the joystick equivalent) reversed. When the First Officer Murdoch called “hard a-starboard” (veer right), the helmsman Hitchens turned the wheel right as he would have done in the days of sail. For a steam ship, he should have turned it left. It took 4 minutes before this mistake was noticed and corrected, at which point the ship must have been much closer to the middle of the iceberg, with disastrous results.
As well as force of human habit, a non-intuitive interface, there was also the little known aspect of binary mistakes. They are the simplest form of decision, do this or do that. Ray Panko in Hawaii has studied human error for many years. He lists a 1983 study where subjects had to turn a control left or right under extreme stress and failed half the time. With these sorts of factors at play, there was actually a higher chance of the helmsman making a mistake than not!
What lessons can we learn from this? Well if stressed people can’t even turn a control the right way, how do they design, code or test? We need to find ways to avoid or find mistakes. The Prince2 methodology does not view a deliverable as complete until it has been reviewed . Even better is the agile practice of pairing for planning, design, programming or testing. Plane pilots also verbalise their actions, describing what they are doing as they do it. This is a good practice when pairing to check the intention matches the action.
Another less known practice is simply avoiding overtime , so people are not tired and hopefully less stressed. To minimise simple human mistakes, checklists are reliable tools for pilots and software teams.
So not only have we found out that human error seems to have caused the Titanic sinking, it was literally a disaster waiting to happen. It was probably the first case of blaming the machine too, as a coverup for human error!