Lady D, a water taxi, capsized in a squall in Baltimore’s Patapsco River North West Harbour on 6 March 2004. Five of the 25 passengers on board died. The US Coast Guard has now released the results of its investigation into the incident.
While the report admits to the complexities of the accident, from human error on the part of the two-man crew, overloading, vessel management, communications, licensing, training and stability, two issue in particular deserve wider attention.
Confirmation bias, in which one accepts information that supports an earlier decision and discounts data that should bring that decision into question, is a common element in accidents and occurred in this case.
Based on an early morning weather forecast, the skipper believed that the weather would be gentle turning to fine in the afternoon. He decided to leave the dock despite an evident incoming squall line. The cloud did not appear to be traditional thunderstorm shapes so he continued to depart the dock even though the weather was deteriorating and increasing wind was making manoeuvring difficult. Despite thunder and a characteristic grey-black squall line he did not change his assessment of the weather.
By the time the skipper received a warning to seek shelter it was too late and the water-taxi was caught in sudden high winds, heavy seas and rain. He attempted to bring the pontoon-hulled vessel’s bow into the wind, which required a turn to port during which he lost control and the vessel suddenly capsized.
Another water-taxi in a similar situation kept the wind to the stern rather than trying to fight against it. Although correctly certificated, the skipper of Lady D did not have experience of handling this type of vessel in heavy weather conditions.
While confirmation bias certainly wasn’t the only root cause of the incident, it did play a key role.
It’s a common phenomenon: Once a decision to do the job is made the focus falls on doing the job, which involves a different mindset, one that tends to set aside new evidence that goes against the decision but accepts, even looks for, data to reinforce the decision that has been made and one becomes blind to changes in circumstances that ought to make us revisit that decision.