This is going to be difficult to do without tooting our own horn but, I believe it’s important and relevant.
While the forum doesn’t have the best track record (and some of the posts are dead wrong) the gCaptain blog has a very high record of sucess predicting the cause of incidents.
After the final incident investigation report is published we always go back and look at our initial predictions… which we sometimes make just hours after the event. We are always suprised how well we nailed it (and, as Ken says, how wrong the forum is).
How do we do this? I’ve never sat down and analysed our methods but maybe this will spurn me to do just that.
Also interesting to note is the fact that our best predictions come with the most hostile feedback. It has almost become a joke in the office… when our phones and emails start lighting up with hostile vitriol Mikey will call me to say we must be pretty close to the truth on this one.
I will say that the personal predictions I post to this forum (vs what I post to the blog) have a very low sucess rate but that is because the ideas I post to the forum are very different from those I post to the blog. I post to the forum mostly those aspects of an incident I don’t understand very well. Often I post knowing that a specific theory is wrong but not fully understanding why it is wrong. I suspect that the forum’s low sucess rate is because other people do the same thing.
As said, I’d like to enumerate our methods but I will mention what I believe to be our single most important asset… perspective.
Before making a prediction we ingest a large amount of data (we have built many tools over the past ten years that allow us to do this quickly) then we eliminate as much of it as possible. Take the Cosco Busan for example. Many initially said that was a terrorist act… for a number of reasons we eliminated that option quickly.
A lot of theories just wash away with more data.
Once we have eliminated the ridiculous (e.g. Russian subs torpedoing the Deepwater Horizon) and the improbable we are left with a small subset of data, each individual piece points to a different set of probabilities.
I use my own intuition to sort this data then I get on the phone and start talking with people who can offer perspective on the information we have uncovered.
What we are looking for at this stage is information AND personal bias. We call people with specific expertise in whatever peice of the information bubble we are looking at and ask them two types of questions… questions that will lead to possible answers as to the cause AND questions that will reveile their personal bias.
Of these two chucks of information the biases are most important. Every expert we contact has a personal bias and an adgenda… we do not uses their ideas as to the cause unless we fully understand their bias.
The biggest source of bias is that of the editor. Many of our competitors have abysmal records in predicting the true cause of an incident and, it’s gotten to the point when I know what “bend” each website and magazine will take before they even publish them because I know the biases of their editors.
How do I eliminate my personal biases? I kinda get two gut feelings after I’ve digested the initial facts. The stronger gut feeling is the cause I want it be… exploring this feeling brings some amount of joy and intrigue when I pick at it. The smaller feeling is wholly negative and picking at it makes me feel ill. It’s this second gut feeling I dive after.
I also enrolled in a two year training course for elite authors that focuses just on this topic… eliminating personal bias from your own work.