There is an old English saying: “look after the pennies, and the pounds will look after themselves.” If you are careful with even the smallest amounts of money, you will never need to fear that you are poorer than you realized. The same thinking can be applied to time, and to revenue assurance. When I worked in revenue assurance, I looked after milliseconds, as well as pennies. The challenge of revenue assurance is often treated as a question of data in vs. data out, where the data that comes in has already been simplified to a whole number of seconds. But if you only think in terms of abstract data, whizzing between computers, then you have already lost some data, and failed to see some opportunities for fraud and error. Events in the real world do not have durations that are measured in whole numbers of seconds. There is money in milliseconds, and where there is money, we should take an interest in the possibility of error and fraud.
To illustrate my point, let us ask how we feel about the following scenario. Suppose some unscrupulous business distributed a smartphone program which makes all calls last 0.5 seconds longer, by introducing a delay between when a user presses the button on their touchscreen, and when the phone signals that it wants to end the call. No matter how call durations are rounded, half of all calls will be recorded as lasting one second longer than previously. If calls are charged per second, and the average call has a duration of 90s, then revenues from affected customers will rise by 0.56%. Whilst 0.56% does not sound very much, revenue assurance has shown that there is very significant profit to be made by chasing these small fractions of percentages.
Having given the scenario, can I now ask who is responsible for ensuring this does not happen? Is it revenue assurance? Is it fraud management? Would we be secretly glad if somebody did this? How would we feel if handset manufacturers improved their code, and reduced the delay between when a user presses the button and when the phone signals the termination of the call?
Let me now translate this into a different problem that nobody seems to be taking that seriously. (Apologies to anyone who is taking this seriously – if you are one of them, then please respond, and set a good example for me and everyone else!) Everybody has a naive understanding of when a phone call starts, and when it ends, even if they do not really understand the technical details and propagation delays involved. But who has an understanding of how much data needs to be transmitted in any given situation? How would you be able to tell, if the data was deliberately padded to increase the volume? This might seem like a crazy scenario for telcos, who worry about the cost of servicing increasing demand for data services. But on the other hand, there are various ways that telcos currently get paid for data volumes, and there might be more ways in future. On the flip side, if networks are worried about capex spending on infrastructure, what more can they do to improve the efficiency of the data they carry? To get some feel for the numbers involved, take a look at the blog I wrote about accuracy expectations in relation to IPDRs and DOCSIS.
As you may have noticed by now, I am banging at doors, asking questions, without offering answers. My interest is in challenging existing practitioners about whether they have looked behind those doors. To be brutally frank, all of us will suffer from a lack of imagination about the ways that complex networks and systems can be subject to error or manipulation.
I have no desire to open every door, and explain what lies behind them all. For a start, that would involve a lot of hard work. Also, it would be thankless. But most importantly, many people should alreay be facing these challenges, so they have as much duty to share their insights as I do. With that in mind, let me ask some other questions that are perhaps left outside of the scope of our work, but need to be inside the scope of somebody’s work.
- Using your phone to make international money transfers sounds like a great idea. If somebody can manipulate the time when a currency transaction takes place, they can manipulate the foreign exchange rate that is applied. What is best practice to ensure customers are not cheated?
- So you want to charge Netflix, and businesses like them, for all the burden they place on your network. Fine. The internet being what it is, sometimes packets need to be resent. Who pays for that slice of the burden?
- 14-year old Suvir Mirchandani has identified how the US Government could save hundreds of millions of dollars by printing documents in typefaces that consume less ink. Telcos are also getting smarter in similar ways; they even hold conferences about energy efficiency. But who is responsible for analysing the cost of things like data which is stored in the cloud, but then never used?
Admittedly, some of these questions are a little odd. But that is partly my point. You cannot expect me to have three insights as good as Suvir’s idea about saving ink. We need to work together, and share ideas, to identify those which will have most impact. And if we do not, then cheaters and fraudsters will come up with similar ideas, with the hope we remain ignorant of what they do! History tells us to look for money in unexpected places. We may look in the wrong places many times, but one surprise finding may yield a jackpot. It is our responsibility to look in those places, even if we are often wrong, or find it difficult to solve the mysteries we have identified. With that in mind, take a look at this recent story from 60 Minutes, a news program on the American CBS Network. It explains how a smart, persistent man identified the way some stock market dealers convert just a few milliseconds’ advantage in their fibre optic network into millions of dollars of profit.