 picture presentation, it's also only for Southeast Asia and South Asia. It doesn't yet include China. Here is the big picture message. You can see here a number of interesting things. We, based on what the scientists have come up with so far, we expect greater gains in production in South Asia than Southeast Asia. We expect the source of most of those gains to be genetic improvement. We expect a greater share of gains in rain-fed environments than in irrigate. So it's quite interesting and some of it is slightly counterintuitive. So that's the big picture. If we get more specific about the sources of the gains, we see that basically germplasm for abiotic stresses is the single biggest source of gains. We see a lot of similarity in the patterns for Southeast Asia and South Asia, perhaps even too much similarity. The principal differences would be that the gains from management for abiotic stresses would be much higher in South Asia than Southeast Asia according to these figures. The gains from improved nutrient management are higher for Southeast Asia than South Asia. Otherwise, most of these are quite similar. And now down to what are the top 10 at the research solution level? So these are the top 10 sources of yield gains according to what we now have. The top portion of the bars with the patterns, that would be for Southeast Asia. The bottom solid portion of the bar is for South Asia. So these are the top 10 considering those two collectively. And the number one source of yield gain is submergence tolerance, according to the latest estimates from the scientists. That's followed by site specific nutrient management for NPK. Now the third one is a little bit controversial. We have two sets of estimates for weed competitive varieties. So this is one of the sets of estimates. The other set of estimates for weed competitive varieties would be very different. So I think that's an interesting point for a lot of discussion, whether that's really feasible. That's followed by hybrid yield potential, bacterial blight resistance, inbred yield potential, brown spot resistance, which is again perhaps a slightly controversial one. Drought tolerance, coastal salinity tolerance, and brown spot management. So I urge you to take a careful look at this. We want feedback on this. This is I think an area that you may all have some ideas about. Now there's another interesting aspect. I can put it back up when I finish the presentation, so if people want to refer to it. Another interesting aspect is so in terms of our aggregate yield gains, we expect more yield gains in rain fed environments than in irrigate. But actually something very counterintuitive happens when we look at our poverty maps and we overlay our ecologies with where the poor are located. We actually have more poor people in irrigated environments than in rain fed. So how it will all weigh out when we get to impacts with the poor we have yet to see. So this is very much an analysis that is still in process. We hope to complete the estimates for scientific solutions by sometime in January. Then we'll have a process of more reconciliation particularly with the yield gaps and yield trends. And then once we really feel that we have good confidence in the numbers, we will begin the process of modeling expected impacts for the poor and for the environment. And we hope to have that by the end of March next year. So thank you. We welcome your feedback. If it helps to do so I can put up that figure on the top. No, okay. Okay, so thank you. So before you start throwing apples and oranges at David, let me reiterate that this is very preliminary. And I could argue myself for at least an hour about some of the lampposts shown in the very final slides. But please provide some feedback on what you generally think on this approach and how we can make it better. Floor is open. I see a hand up there. Eric Wales talking to you. David, this actually is a critical book with the presenters. I don't know exactly how the exercise of estimating impacts perceive it, but you talked a lot about the uncertainties. And so do you have, did you collect or save the information on the uncertainty, sort of the second and third moments of everything been put up here as point of estimates? And I know this is already a very complicated exercise, but if in fact the uncertainty of impact is critical, that how the confidence interval around these impacts, how broad it is and whether it's skewed in one direction or another, you have an empirical distribution. I guess if you solicit it from scientists what their estimated impacts is, is it possible for you to utilize that information to add value to this exercise? We are doing that to some degree in terms of certain parameters we ask for a range, but we don't ask for a range for every parameter because it will just double the amount of data to be collected from the scientists. So particularly the greatest uncertainty is about adoption. I think most scientists feel the lowest level of comfort in terms of their adoption numbers, so for those we're using ranges. But you have different only on their point of estimate? Well, we're working, we have a group approach for this, so generally what's happened is that when there's divergence there's discussion and usually the divergence narrows down. Often it's a matter that a particular individual may have a different interpretation of something than someone else once you get common understanding, I think. Probably one of the areas of future improvement to bring in elements of uncertainty analysis, and there's various ways to do this. At some point I think I may have suggested, I can't remember, the thing about fuzzy sets for example, but we want to go one thing at a time, so I think there's many ways to get better. Basia had a comment on that. Back there, please. Wait for the microphone, please. You know, it's great. You mentioned in the field games when you did the South Asia versus Southeast Asia, you highlighted the rain fed and then germ pleasant, being kind of the area greatest field game impact. And then towards the end, you know, it was kind of like, okay, that's hardly relevant. We're finding more core and irrigated than rain fed areas, even though now we're projecting greatest field games in rain fed areas. So where do you think there's opportunity for us to help with distribution in those irrigated areas that are going to be benefiting or not benefiting you? You see where I'm drawing those connections. Well, yeah, I think it's a matter of, we'll have to see in terms of the technologies themselves where the balance even comes out. But it will be a matter of which technologies benefit the greatest number of poor taking into account where the poor are and where the technologies can apply and what can be expected of them. Is that factored into your modeling? Yes, but that's when we get to the point of processing the impact estimates. I mean, we're looking for what options maximize benefits for the poor. This time I got your name right here. A few years ago, we had a problem in a particular country and they invited a pathologist to look at this problem. Obviously, it was a pathological problem. Later, they invited a sorrel scientist to look at this problem. Obviously, it was a nuclear problem. You incorporated Erie staff, 70% of emerging jockeys. I'm not surprised you've come to the conclusion that breeding is the problem and they're all focusing on submerging. I think we have to open this up a little bit and get outside of Erie. Otherwise, you're going to come to the same conclusion we've been doing for 20 years. Should I respond to that for you? Yeah, yeah. I would say that in terms of the expertise involved in this, there hasn't been a strong disciplinary bias towards breeders. Actually, breeders tend to be less interested in this kind of exercise by default than the agronomists have. So, I think actually the participation has been reasonably balanced and there's actually the teams that have put together the estimates have been interdisciplinary. So, usually there's been an agronomist plus a breeder, sometimes plus a molecular breeder, sometimes a pathologist. It's not that you have just a single individual or a single discipline filling in the sheet for their own discipline. But I want to maybe make an own comment because I know where Ed is coming from and it is a problem that we still grapple with. And Bas has alluded to that quite a bit actually too and that is in such an exercise, how do you account for the quantitative impact potential of, let's say, a technology solution which is actually the stepwise implementation of many different little components. So, the six principles of better agronomic management that Flare and Ed and his team in South America have promoted very successfully in different versions in different countries. If you would evaluate each of those separately, they would come out very small. But we have seen that where it is done all together we have had the highest yield growth rates of anywhere in the world in the last six years 200 kHz per hectare per year. And there are things in there that individually we don't know yet how to put it all together. It is like the six principles such as optimizing yield potential by making sure that you plan at the right time and choose the right variety that maximizes the yield potential for that climate. How are you going to quantify that? So, we still need to improve that. But that alone doesn't help you if you don't do the other five steps, which are things like have high quality seed, irrigate early enough to make sure that you get a vigorous crop, make sure that you get your nitrogen fertilizer on at the right time and at the right efficiency to preventive measures of best management when necessary, things like that. So, the complex solution that actually represents the set of best management practices that a farmer can use to close the yield gap and that combines the high quality seed on the right variety with the management package. How to quantify that and also account for differences in the ecologies and regions, that is probably still the biggest unresolved challenge. If we were able to do this right or better, I think some of the preliminary results would look quite different, particularly with regards to the relative proportion of genetic gain impact versus management impact. So, I think with this type of exercise I agree with Ed, we tend to underestimate the potential impact from the agronomic improvements, which individually may not look like much, but when you put them together, then they actually have a very big impact. So, it's one of those things, I think, that we will need to carefully look at, but it's not easy to do. So, if anybody has ideas on this, we'd certainly welcome those. Also, not mentioned just to finish this, is we have in this exercise also other things, which we look at like post-harvest losses, for example. Not mentioned, but we look at that too. Or improvements in grain quality that could lead to a higher market value. There are many aspects still, which we are working on, but not yet showing them. So, pass yet another.