One of the concepts I've been somewhat enamoured with lately is that of Pareto optimality. I first heard of the idea of Pareto optimality in my secondary school economics class and recently realised it makes a rather good aid to decision making - on the road, and elsewhere.

Pareto optimality (also known as Pareto efficiency) is a concept first referenced by Italian economist Vilfredo Pareto in the 19th century. When an economy is in a state of Pareto optimality, it is in a state of allocation of resources where it is not possible to make someone better off without making someone else worse off. For instance, if there is a finite amount of electricity, in order for an electricity company to service an additional customer, they need to reduce what is supplied to other customers - i.e. for one person to be better off, the others are worse off.

Most resources are not in a Pareto optimal state in real life because our supply of resources is constantly growing. However, I think the idea of Pareto optimality actually translates to the behaviour of road traffic in most urban scenarios very neatly.

As an actor in the road system, there are a series of rules that govern how I must behave. These rules are necessary to prevent collision, to protect the safety of users of the road system and to ensure fair and orderly flow of traffic. Often collisions happen because expectations of the behaviour of other actors break. This can happens either due to lacking information (as commonly experienced in bicycle collisions where a driver cuts up a cyclist who they didn't see and thus were not expecting to be present) or because an actor behaves in a way that flouts expectations (for instance, a driver running a yellow light).

Of course, where these rules are commonly broken, the expectations of how actors behave change. These expectations, however, are still common amongst the majority of road users - allowing traffic to flow (albeit, often, at increased danger to participants). For instance, in India, is it given that traffic laws are disobeyed. However, for the most part, they appear to be uniformly disobeyed. Lane markings act as guidance more than any hard and fast rule of where your vehicle must travel, so drivers are accustomed to using their horn when a vehicle that is changing trajectory is at imminent risk of colliding with theirs. Equally, drivers listen for these beeps when driving. (There are many more examples of how the road system there actually manages to function without rules and with a billion participants...)

So how does Pareto optimality fit in with this system of rules on the road? Simply, given that a particular traffic situation within the road system is Pareto optimal, my actions as an actor may benefit me but will make another driver worse off. In the aggregate, this doesn't serve to benefit the world at large - and often the benefit to me is minimal.

Before I illustrate this with a few examples, it's worth considering when and when not a traffic situation is Pareto optimal. Pareto optimality is concerned with the allocation of resources - in the traffic sense, I like to think this is when the given capacity is fully allocated. For instance, let's consider my the last segment of commute, through San Francisco from the Powell Street BART station to work (a taxicab distance of 7 blocks).

In the mornings, there is heavy traffic and the first entire block of my trip, just south of Market, is often filled with standing traffic up to the intersection between Mission Street and 5th Street. In this situation, traffic almost behaves like a sliding block puzzle - for there to be space for your vehicle, another vehicle has to move out of the way. Similarly, an intersection can be considered full allocated when every pathway into that intersection is filled with traffic, pedestrian or vehicular.

This specific fully allocated road situation lends itself to my first example - that of running a light. It is clear that this situation is Pareto optimal - if I was to, as a driver or a cyclist, run a red light, it is very likely going to make someone else worse off at the benefit of saving me about a minute. Besides the obvious danger to crossing pedestrians, it will likely cause an entire row of traffic to (in the best case) wait a few seconds, and (in the worst case) have to brake to a halt, depending on the timing of the lights.

Any city cyclist will bemoan the irrationality of traffic lights, particularly when there is no traffic (this often happens to me when cycling past, say, 10pm in the evening). However, this situation is not Pareto optimal - there are plenty of spare resources, and it is possible to jump a red light without making any other road user worse off (albeit at some small risk to your own safety).

Another example where Pareto optimality becomes evident is lane shifting on a busy highway. In California, at least, traffic speed on the highway is fairly uniform - often the leftmost "fast lane" will be traveling only marginally quicker than the rightmost "slow lane". Slower traffic stays right and people move to a left lane to overtake. When there are three or more lanes, this system works reasonably well, and traffic is evenly distributed over the lanes. It breaks down, however, when there are just two lanes. This happens because the speed differential between the slow lane and the fast lane is significant - there are fewer lanes but drivers are still as slow and as fast as on any other highway.

Under heavy traffic, then, the left lane becomes a bottleneck for faster drivers. This results in a nuanced situation where the left lane is fully allocated and therefore Pareto optimal but the right lane is not, due to gaps in traffic. Any action affecting the left lane is likely to make actors worse off. A common action is where drivers in the left lane use the right lane for undertaking. They gain a few car positions and are slightly better off, but given a fully allocated lane, do not manage to increase their overall speed. On the other hand, this action forces an entire stream of cars to brake, thus making them worse off (increased fuel consumption and annoyance).

These two examples illustrate, hopefully, how Pareto optimality might be a useful decision making aid, particularly when commuting around town. When I'm driving or cycling around, it's nice to think about what effect my action will have on other people. If I rush through an intersection, is my gain (usually in time) worth the annoyance, discomfort and inefficiency forced upon other users? Equally, if I'm at a pedestrian crossing at night and there isn't a car in sight, no one is likely to be worse off if I jaywalk.

As for elsewhere, there are several situations where the concept of Pareto optimality has utility. I like to think the outcome of negotiations can often be improved by striving to push them to the Pareto optimality boundary. For instance, if a hypothetical startup is making a deal with a potential customer and the terms of the deal are not at the point where asking for more will make any party worse off (financially or in terms of goodwill), that deal is likely not optimal.

Of course, looking at another type of negotiation: that of a tenant and a landlord, that situation often is Pareto optimal and it's not possible to change the terms of the agreement. Changing the rent would very evidently make one party worse off and another better off.

Hopefully this makes sense and I haven't totally abused one of the fundamental principles of economics!