We have all heard the phrase “Lies, damned lies, and statistics,” often used to describe the use of statistics to bolster weak arguments. The phrase is by many erroneously attributed to Mark Twain, who in turn attributed it to the British Prime Minister Benjamin Disraeli (which, apparently is not correct either since the earliest known appearances of the phrase were years after his death).
However, if not abused, statistics is a tool related to facts and stating a truth more than a lie. Furthermore, analyzing statistics can help us predict the future or estimate a value. There are many ways this can be done; perhaps the most typical one is to use a trend line graph set up by past data of whatever development you are studying, whether it is population growth or house prices, and use this to extend the line as a prophecy of the future (looking at the temperature drop these last weeks I can predict winter is coming).
To me, an even more fascinating property of statistics is often referred to as “The wisdom of the crowd.” If you are unfamiliar with the term I strongly recommend that you Google it; I promise you will find many interesting articles. “The wisdom of the crowd” principle is used for an incredible range of subjects, from capital management, business decisions, even politics, and – wait for it – it can also be used for weight estimation of a vessel!
The “wisdom of the crowd” can be explained by the following simple example: Put a jar full of jellybeans on a table. Let a random group of people each take a guess on how many beans are in the jar, and note down the answers. If you pick a random guess from one person and compare with the actual number of beans in the jar, chances are you will not be very close. But, here’s the interesting part: Take the average number of all the guesses and you will most likely be astounded at how close this will be to the true answer. Have a couple of thousand people guess and you’ll probably be spot on. The explanation is as simple as that for every one person guessing a number too low, there is a person guessing a number too high, and on average, they cancel each other’s errors. You’re just an internet search away from reading about actual studies and experiments that confirm this description.
So, how does this relate to weight estimation? Well, I’m not suggesting that you should line up thousands of people and have them all guess the weight of your ship (although that would be an interesting experiment), but there is a more practical and even better approach: By dividing your vessel into several weight groups and doing an estimate of each of them, you will in many ways obtain the same effect. Each weight group represents a ” person’s bean guess” and a higher than reality estimate in one weight group is most likely cancelled by a lower than reality estimate in another weight group.
Now you might object to this and say that even if we are talking about several different weight groups, the guess for each weight group is taken by the same person; and this person may very well have a tendency of leaning too low or too high for his guesses. Therefore the errors would be correlated, thus not giving the desired effect and ending up with a bad overall result. This is a valid objection; however, the estimation for each weight group should be derived from historical data – statistical facts – rather than “guesses” and the correlation between errors should be minimized this way.
To conclude: Divide your estimation into several weight groups to create your own “crowd” and use historical data to obtain uncorrelated “guesses” between the weight groups. Harvest the “wisdom” to get an accurate estimation of the weight and CG of your vessel. And ShipWeight is your software tool to get the job done efficiently.
PS: Another statistical method that is related to this is Lichtenberg’s Successive Principle, which I will discuss in a future blogpost.