 And so, I looked at the global data that we have, which is a matrix that was set up, what we call the global origin destinations matrix based on census data, stock data. And that's become the standard building block of any discussion of global migration, in fact, any discussion of regional migration. It's passed into the received wisdom, I think we might say, in migration studies. And the points I wanted to get across is, yes, this is a good thing, but it's also got weaknesses, and so we've got to know how to deal with those weaknesses. A stock figure gives you an idea of how many migrants are in each country at a particular time, but it doesn't say when they got there. And it doesn't really tell you much more than these are the number of people in that country at that time. And so I was arguing that we need to move forward with this, and the most obvious way to do that is to look at flow data, to try and develop more accurate flow data, which we can do in two different ways. One is to improve registration data, a continuous system of registration of how people come into the country and go out, because very few countries have records of how many people leave, and also continuous records of people moving within countries. So that's one direction forward. The other direction would be to look at the censuses, the same as we do with stock data, and try and find out how many countries have direct information on time of arrival, last place of permanent residence, and using these data to try and generate flows. There are other methods, more sophisticated mathematical modelling. Then I went on to look at what I think is extremely important, is that we tend to look at interstate movement, in other words movement from one state to another, let's say from China to the United States. And I don't think that's particularly meaningful, because you really want to know from where in China migrants come, and to where in the United States migrants go. And we know that a lot of that is urban-to-urban migration originates in the big cities of China, and it goes to the big cities of the United States. There are exceptions. You do find migration out of small-time China, or rural parts of other countries, such as Pakistan and Bangladesh, going to the big cities of Europe, particularly the United Kingdom, or at least in the past that was where the most went. But my point was that we really need to look at sub-national units, and migration to large cities. Now I think this is going to be one of the most important ways to improve migration data as we move forward. Yes, states control migration, that they have the national migration policies, how many people should come into a country, but really it's the cities that have to provide the education, the housing, the jobs, and most migrants go to big cities. If we looked at the United Kingdom, my own country for example, you would find that the population of London was about 50% foreign born, whereas on average in the United Kingdom it's 13%. That's a huge difference. So most migrants, most international migrants, go to large cities.