Welcome to Lecture twelve on Networks, Friends, Money and Bytes. And today's question is a followup on the question we saw last lecture. Back then, we looked at why am I being charged $ten a gigabyte in certain markets. And now we're going to look at how can I pay less for each gigabyte. Before we go into this very practical question. We would like to highlight that the field of network economics is a broad one. We had been focusing on the specific area called, network access pricing. For example, on the mobile cellular network. There are quite a few other important areas in the field. Including The interactions between the ISPs and ISPs. In the next lecture, we will look at the inter ISP agreements, the interaction between Internet Service Providers and content distribution network, for example, between AT and T and Akamai. And, the interaction between, vendors such as Apple of cellphones, iPhones and iPads and the ISPs. Okay? The way that they arrange payments agreements are, are very interesting to examine. Another topic is the proliferation and demise of technologies. Trying to look at the economic and the networking technology aspects of these trends. Yet another important area is the decisions of making certain investment. These can range from pro-tracing spectrum for 3G or 4G network. And these can cost tens of billions of dollars before each auction. To, the, adoption, of a certain say, security or new access technologies. So access pricing, interaction among the ISPCDN and vendors, the modeling of the rise and fall of technologies, and the decision of investments, they all can benefit from our understanding of network economics. And this field is interesting in part, because it's a 2-way street, new network technologies can rewrite the valence equations in the economic. Modeling. The same time, economic forces shape the rise and fall of network technologies. It is also a challenging feud, mostly because, data is very difficult obtain. Whether these are the data about the cost, or revenue. Models of the ISPs, or these are the, say, data of consumer behavior. For this kind of data, we'll need proprietary information from the industry. For this kind of data, we need to carry out realistic consumer trials. We'll see some of these challenges, we've met later, in today's lecture. So now, let's look at, specifically, access pricing. Now, access could be access for voice, or access for data. And our focus is on data. So we call this data pricing. Now, we have seen data pricing. In the last lecture, we have seen flat-rate data pricing. We have seen usage-based data pricing. And we're going to see some interesting extension that we collectively call the smart data pricing. So this is an umbrella term. And there are many different meanings of this term. We're going to look at a few of them, in this overview module of the lecture. Now, one is an hourly accounting based data pricing. Instead of looking at how much you have used. It looks at, how long have you been using it. Another one is priority pricing. This is, instead of asking how much you should charge, you ask what to charge. For example, if I offer a premium service, Okay? Maybe I can charge more. In some of the countries you have what's called anti-throttling. Now, throttling is reduction of transmission speed. Anti-throttling says that I will boost your speed up a booster button. If you're willing to pay more. There are also other kinds of answers to this question. What to charge, for example. Why you know that different product will lead to different prices. It turns out that you can also charge different prices and that will automatically differentiate the products into different tiers. In the advanced material part of the lecture we will look at an example of that copy. Paris Metro pricing, where you label the same cars in the metro subway service into premium or first class versus normal or code class, has set the different prices. It turns out that they are charging the same product differently, actually you differentiate the product, as long as people's, consumer's utility depends on the congestional utilization level of the service. Yet another variation is whom to charge. So far we've assumed that the ISPs will charge you and me, the consumers. But there are also notable exceptions. For example, the 1-800 number in the US and similar kind of idea with different numbers in other countries is a big success of creating win-win across callers and callees, instead of you paying, Okay? You call 1-800 number, you don't pay. Back in the old days, calling long distance can be quite expensive, even on land line. Instead of the. Other side will pay for it. Why will the other side be interested in paying for it? Because they would like to sell you something else, another service or product. So similarly that you can think of was called 1-800 number for mobile data or sponsored content. Now this is not identical to Google sponsored content or ad space option on a search page. But the idea is that, is that if a consumer is paying for the content. The content or app provider may be able to chip in part of that, and subsidize the consumers. And when the elasticity is just right the bottom increase will over-compensate for the amount of rewards that you provide to the consumers. So it could be a win-win for everyone. Okay, yet another answer is to the question of how to charge. You may say that instead of charging everything the same, I may say, certain applications I would charge less. This may depends on the kind of applications, Okay? Certain application may occupy a lot of capacity. And related to that is the idea of congestion-dependent pricing. It says, if the network, here's a pipe, okay. In the network, and in general, there are many such pipes. Okay. I look at this network and I say, maybe this part is very congested. So all those flows that go through this path should pay more. Okay? Whereas, the flows that go through other paths that are not congested. They don't need to pay as much. Because they are not really competing against many others. And there's more than capa-, more capacity than, they need anyway. So for example, this can be location dependent. Certain locations are just so crowded that, you're presence presents a strong negative externalities to the other. Or it could be time dependent. At certain time of the day, the network is not in much use and there is not opportunity cost. So go ahead, I will charge you much less, or maybe even free. A typical idea is the nighttime versus daytime tariff or pricing for voice calls. Nowadays, the carriers, at least in the US, are trying to switch from limited. Data, a limited voice, and unlimited data to the other end into limited data and unlimited voice. Because they realizes that its mobile data traffic that's driving their cost revenue equation. But, back in the old days, you get unlimited data, but your voice minutes are limited. However night times and weekends don't count. So that's a simple static. And two period. It's, there is just a night time and day time. It's static because you pre-define what is night time and weekend and the price. Well, the price is zero, cuz it's free. Day time, it will count towards your fixed quota. That is a simple case of time dependent static pricing. We'll look at time dependent dynamic pricing with multiple pairs coming up. So whether it's location dependent or time dependent, eventually, it can evolve into congestion dependent. Look at the particular time and location, and decide the price based on the congestion condition. So the phrase SDP, smart data pricing, refers to any of these, okay? Hourly based, priority pricing, implemented by a variety of mechanisms such as parismetro pricing, two-sided pricing, or 1-800 number generalization of sponsored content, or application of congestion-dependent pricing. Now today, we will be soon focusing on just one specific case: time-dependent pricing. We'll go into a little bit of detail of this one member of the family of SDP, and then advanced material will say a few words about product pricing and two-sided pricing. But before going into any of these, I want to highlight again, the difference between static and dynamic pricing in our terminology. What we say, pricing is static, it doesn't mean we charge the same thing to all the people, all the time. Simply says that we have a predefined, fixed definition of. Time slots that's predetermined. And the price is also fixed, even though we can have different prices for different time slots. So the daytime versus a nighttime and weekend time voice call. Tariff practiced by many US carriers in the last decade, falls into this kind of, pricing. Okay? You fix the time slots. You fix the price. Which may change, but very slowly. It may be, you know, once a year or once a couple of years. The carriers may decide to move the price. But it is a much longer time scale than, say, a monthly bill basis. You can have different prices, it's just that prices are predefined, fixed for fixed time slot definition. Now, in contrast, dynamic pricing says that the price at different time slots can change in a fast time scale. Or alternatively can say that you're not even defining or predetermining the notion of time slot. You can say that well, today APM it is $ten a gigabyte but tomorrow's APM maybe $nine a gigabyte, Okay? Now, how fast is the time scale? It got to be faster than, say, the monthly bill's time scale, which is monthly, Okay? But there are choices. Now, you can say, I will do day ahead. I will announce the price, at least for this location. Or maybe for this area of locations that you are likely going to be, situated in, one day ahead of time. So you can prepare for it. But you can also say, I don't announce a day ahead, I do spot pricing. Meaning that I will price right on the spot. I don't announce any day ahead. Maybe one second ahead, or something. Another dimension of time scale is what's the granularity? Now the granularity here, of, one of a duration in which the price is held the same. It could be say thirty minute. Okay? Or it could be one minute. It could be, maybe, two seconds. Now, anything below that, it might be infeasible just to go through all the signaling, that control this bidding process. But you can go down from, you know, hours and minutes, all the way down to seconds. Now, of course, as you go down further to seconds, you better make sure that you have auto-pilot, without human interactions. And that you are basically going into from congestion, from time dependent to congestion dependent pricing. Okay? You're looking at instantaneous congestion condition. So all of these are examples of dynamic pricing, in contrast to static pricing. Clearly dynamic pricing are more general, and more, powerful but it also suffers a potential risk of complicated message to, the consumers, so we must have a very simple user interface, UI, and in fact, as we will later discuss, you can have a dynamic pricing but not dynamically viewed by consumers. There are many ways to hide the complexity of dynamic changes from the consumers, so that they do not have to be bothered, even though they are given the power to choose among alternatives. So the ideal win-win situation through STP is the following. That the carriers or internet service providers would generate more revenue, reduce cost, and lower the term, because the consumers are happier. The consumers, at the same time, will be able to pay less for each gigabyte. Now, what is the catch here? Consumers pays ISP, how can they both be happier? What the consumers may be paying less for each gigabyte. Say, instead of $ten a gigabyte, on average, say you pay $five a gigabyte, you think that you are getting a much better deal. At the same time, if they consume more, in general, without incurring more cost to the ISP infrastructure, then the ISP gets more revenue, lower or the same cost, and thereby increasing their profit. This also implies that the content providers in such a win-win will attract more eyeballs and therefore afford better subscription or mobile advertising revenue. And the vendors will be able to sell more innovative hardware and software. So the question is how can we achieve such a wonderful win-win? How can we go from the current state, which is $ten a gigabyte, to such win-win through different smart data pricing, STP. Now we will look at a specific case of time dependent pricing in most of the rest of this lecture and then advanced material go into a couple of other topics.
