We show how adversarial risk analysis may cope with a current important security issue in relation with piracy off the Somali coasts. Specifically, we describe how to support the owner of a ship in managing risks from piracy in that area. We illustrate how a sequential defend–attack–defend model can be used to formulate this decision problem and solve it for the ship owner. Our formulation models the pirates’ behavior through the analysis of how they could solve their decision problem.
When Richard and I recorded the podcast a few months ago, we had no idea that there were actual research projects involving pirates and operations research. What a pleasant surprise!
Here are a few show notes and links to extra material.
Walter Mebane, Professor of political science and statistics at the University of Michigan, has written several articles about under-staffing voting locations, election fraud, and election forensics (he may be a political scientist, but his papers have figures that look like they are made in R. Not too shabby). Check it out.
You are more likely to die in a car accident on Presidential election days. Here is the JAMA article that summarizes this disturbing research finding (and a Scientific American podcast about it).
University of Cincinnati PhD student Muer Yang developed a simulation model to reduce and equalize voter waiting times across different voting areas. This is a good example of using OR for elections.
How did the University of Illinois student body elect a gnome and snail for the student body president and vice-president, respectively? Technically, they didn’t. The gnome and snail were disqualified after winning a plurality of the vote, and the controversy led to an investigation. Here is the full story according to the snail. Amazingly, all of this is real.
We followed up on our last episode on pirates and OR. We found a new article in the Journal of Transportation Security that presents statistics on pirate attack success rates.
Don’t forget to email me with your OR jokes. And don’t forget to vote!
A search for “pirates and operations research” turned up a link about Pittsburgh Pirates player Ross Ohlendorf, who was an OR major at Princeton. I am interested in OR and real pirates. I have a new podcast out on pirates (see below).
Here’s the gist of the piracy problem plaguing the Gulf of Aden and Indian Ocean. Pirates operate small, maneuverable skiffs disguised as fishing boats to board trans-Pacific cargo ships, hold the crew ransom, and disappear once they receive their ransom. They don’t steal the cargo on the ships very often. Even in the modern world, it is hard to track down the pirates after the ransoms are paid.
Various governments have tried to combat piracy with warships that patrol the pirate infested waters in the Indian Ocean. The warships “can’t keep up with the region’s elusive pirates. The hijackers’ simple, brutal tactics are too effective. Their business model is too attractive. And they’ve got nothing to lose but their lives.” (see this Wired article).
Finding the pirates before they attack is like finding a needle in a haystack, since the pirates disguise themselves amongst many fishermen. The US Navy began a randomized screening approach to identify potential pirates.
[U]ntil they brandish weapons, pirates are indistinguishable from legitimate fishermen. Naval crews must stop and interrogate a lot of innocent seafarers, as seen in the U.S. Navy video below, in order to have any hope of disarming pirates before they attack. With tens of thousands of fishing boats plying the Indian Ocean alongside just 30 warships, lots of pirates are sure to slip through.
Preempting pirate attacks by screening is therefore not like screening passengers at security checkpoints. Everyone looks the same up front, so random screening is truly random, not risk-based. It’s not surprising that this doesn’t work. My research with Sheldon Jacobson has shown that aviation security resources are not good at detecting threat passengers if they do not discriminate between risk and less risky passengers. The problem is, it’s harder to identify risky fisherman than risky air passengers.
Mercenaries–Rent-a-cops with real weapons–can protect the cargo ships at a much lower price. This is cheaper than the government solutions are also more effective. A “few armed guards should be sufficient to defeat a pirate attack” and the mercenaries are always at the right place at the right time whereas navies usually arrive after the attack. Perhaps mercenaries could be placed on cargo ships much like Federal Air Marshals are placed on air flights: using an optimization-based approach that solves large Stackelberg games.
Looking at piracy as a supply chain, one goal is to disrupt the supply chain as much as possible in order to deter would-be pirates. Since the equipment used in piracy is so simple, this may not be very effective. But pirates still need some kind of training to become successful menaces to society. This could lead to a kind of shortest path network interdiction model, where the pirates-in-training are trying to find the shortest path to becoming a pirate and we are trying to force them to take the longest path, thus disrupting the pirate supply chain. if it takes longer to become a pirate, then there will be fewer pirates and fewer pirate attacks. The network of nodes and arcs with associated costs would be akin to modeling the path to one’s university degree, where the nodes are courses and the edges represent prerequisites. Becoming a pirate would likewise involve certain skills (nodes), some of which must be performed in a certain sequence (arcs). If there were such a thing as Pirate University, I hope they don’t teach operations research.
Richard and I discuss pirates more on my podcast. We also talk about sports analytics, OR in the news, . I attempt to tell a couple of OR jokes, but as you will hear, I am in desperate need of better OR jokes (send them to punkrockORblog@gmail.com). You can listen to the podcast here:
How would you use operations research to catch pirates?