Thursday, 30 July 2009

Thought experiment: International parcel post

Yesterday I had a chat with a colleague who's looking at transformative ways of reducing the cost of sending parcels to China. We came up with the following thought experiment, which was quite fun - I thought it was worth sharing.

We started out with the assumption there is a fixed cost of cargo planes along the route from the UK to China and a very much lower cost of sorting a parcel, which varies based on the labour cost in the territory the plane lands in. The main determining factor on the cost per parcel is the utilisation ("degree of fullness") of the cargo space in any given plane, the secondary factor being the number of times it is sorted. A parcel takes a pre-determined route through the system, with a fixed number of interchanges, independent of the utilisation of the planes that carry the parcel at each stage.

It occured to me that this is analagous to the switched telephone network; our thought experiment was to imagine that the international parcel post worked on a "best endeavours" basis, analogous to the IP communications protocol on which the Internet is based. In this scenario, parcels would arrive at a sorting office and be routed to another destination based on the utilisation of the cargo planes that were at the office, always seeking to maximise utilisation and therefore minimise cost per parcel across the whole system, while moving individual parcels to their destination in a "random walk" fashion.

We came up with a number of challenges. Firstly, routing depends on having accurate information about a parcel's destination that is easily (electronically) communicable to the sorting office. RFID tagging of parcels at the point of despatch is the simplest way of achieving this. The technology is well understood and unit prices are low.

Secondly, routing must be calculated based on knowledge of utilisation in the total system, meaning that a great deal of computer power will be required to run the system and that information on parcel volumes and airplane loading must be shared by all sorting offices. Neither problem seems insurmountable, the first part could easily be solved through cloud, grid or distributed computing with a commercial model based on a fixed price to route a parcel. The second part is political and based on corporate strategy and business model. True to form, if the largest couriers and carriers adopted it, the rest would have little choice but to follow suit.

The third challenge is about customer expectation. In a truly "best endeavours" system, it would no longer be possible to guarantee the delivery date and time of a parcel at the point of sending, unless a highly sophisticated probability- or chaos-theory based model could be constructed to predict likely delivery time. Alternatively, a QoS model could be forced on the system - my opinion is that this is a bad idea (I have a similar view of corporate NGNs that try and do the same thing, but that's for another day), however it could be done by having a sense of importance and priority on each parcel.

The third challenge made us think about pricing models. In our experiment, a customer could be charged a variable price for delivery of their parcel, perhaps within a given range. This would enable the carrier's pricing to become more real-time, enabling them to respond to demand in a more flexible manner. Similarly, the consumer will have better visibility of the trade-offs inherent in international shipping - price increases dramatically when time is the priority, whereas it falls precipitously if efficiency is maximised.

Billing will be an interesting challenge, particularly if it becomes factored into the routing algorithm. Similar systems are used for the exchange of international minutes between telecoms carriers, so the technology is understood. There may even be "parcel arbitrage" opportunities.

The primary upsides of this scheme surround efficiency, both financial and carbon-related. By emphasing utilisation of resources, there is less wastage and hence less cost, benefiting the system as a whole. In addition resiliance to disruption such as demand peaks, weather, strikes etc... would be markedly increased.

So, a really enjoyable exercise - I'm sure you can come up with other interesting implications of it. To close, I must say that the degree of cooperation required is probably beyond the World at this intermediate stage in our societal evolution - it'd be fun to try though...

No comments:

Post a Comment