 That's very easy so about one and a half years ago. There was a big conference in Bonn Connected with my 65th birthday actually it was 11 months earlier the birthday, but the conference was just before my 66th and In the week before the conference Ken Ohno who was one of the organizers of the conference, but actually couldn't stay But he was in Bonn the week before he asked me a very puzzling question and Kind of as a challenge and so I went home thought about it told him the next morning I'd spent three or four hours kind of half of the answer and Then the next day I finished getting the complete answer to his question this question Made no mention of the Riemann hypothesis the Riemann Zeta function It was on something completely different called the partition function, which every mathematician every number theorist knows very well Then he left because he had another conference to go to one week later in the mail. I Well the email I got a paper with four authors Ken Ohno, it's two postdocs sort of Griffin and Roland and myself on the Riemann hypothesis and application I have no idea that that's why it was why he had asked me this original But it turned out that what he asked me was a kind of a toy problem To prove something like what was needed for this paper on the Riemann hypothesis but in a simpler situation and So he gave me the simpler one so that I wouldn't be prejudiced and I'd solved it So then based on my solution they wrote up the solution of the full problem Actually turned out it was not yet a complete solution There were the analysis turned out to be much more complicated than they had thought and to adapt what I done for the toy problem To the Riemann case actually took it's another eight months till it was correct. It was not finished in a week But eventually we we finished it and the paper then was submitted and Accepted by the proceedings of the National Academy of Science I think there was something very beautiful in this toy problem Which did not have to do with the Riemann state function It's called the partition function another famous function famous because Euler found it Studied its properties of data Ramanujan and hardly proved wonderful results about it and When I saw that toy problem But I actually showed was that the same answer the same method would apply to a very large class of functions And there was a kind of a universal answer What you have to show is that certain numbers these are the roots of some polynomial Would eventually if this parameter that I thought about being big enough would eventually all move apart and become distinct That was the problem But what I found with the computer and then proved was that they didn't just become distinct They stabilized all was to the same thing that these zeros these numbers always became in the limit the same numbers Independent of the other parameter, but also if the problem it was true for partitions It would be true for any generic function And that was the one that was then used for the Riemann's data function It was the same they had the same limiting value and this really isn't eloquent result It's very surprising you ask something about certain numbers roots of a certain polynomial And then it turns out that the answer is always the same the problem can come from many many different sources There's a parameter that goes to infinity so you have a limiting distribution and that limit is always the same It's a universal answer So that is something I think any mathematician Consider eloquent because you don't expect you think each time it will be different, but no it's always the same There's a universal answer and this was certainly a nice discovery So that I think is the actually the most interesting part of the paper that there's this universal Formula that you can apply to the toy problem partitions to the not so toy problem about the Riemann's data function in principle to many other Questions I don't any important applications in theory many other questions at this sort and you will always find the same answer So that's that was a big surprise Well, that's an amusing story. So many years ago, maybe 30 years ago I was having a conversation with Enrico Pompeii who's very famous Italian number theorist But we've worked most of his life now in America fields medalist to absolutely one of the great living analytic number theorists and he was visiting one we've known each other for years and we got in Conversation about the Riemann's hypothesis He is was then still is a firm believer like a Believering in God you just cannot imagine that it's not true It's an article of faith and I was taking the point a few temples advocate. Probably. It's true. Maybe it's true Maybe it's not true. I would say this I said there's not enough evidence. I would be happy to make a bet 50-50 either way, so he immediately dumb dumb. Let's make a bet So then we decided what we said two bottles of good wine because he's a wine connoisseur and that's what we both love good wine But then we have to decide how will we ever decide who won this if somebody proves three not hypothesis He wins if somebody disproves that I win But what if nobody ever solves it and we go to our graves? We didn't what we wanted a criterion So I offered which was stupid of me. I said, okay, look Enrico Today it was 30 years ago. Don't remember exactly Three million zeros of the Riemann's eight of them should be computed and none of them is a counter-example the conjecture yet If they succeed in getting three hundred million a hundred times more and they haven't yet disproved it I will confess defeat and I will pay up So then ten years went by because this was a very complicated calculation And I knew if they wouldn't be people wouldn't go from three million to four million There's no point you you have to make serious progress the ten years later Computers have become more advanced and so two teams of the scientists and mathematicians one Brent and his colleagues in Australia one Terril and his colleagues at Amsterdam Worked and they worked in competition but also writing letters to each other no email yet And they decided they wrote preprints which came out at the same time still no archives And they stopped at 200 million so I breathe the sigh of relief my bet is still okay It will now since they stopped it 200 million it will be another 10 years for people go to a building Of course, I knew I would eventually lose but I have a few more years But I have a friend Hendrik Lenstra another very famous number. There's Dutch He's dodges from Amsterdam and he told him he knew about my dad So he told Terile secretly why don't you guys go on that don will lose his bet So then they went on and they went to 300 Peltin So I lost my bet I gave one Gary his two bottles of wine one he shared with me So I drank half of this bottle But to do that calculation took 700 hours of CPU time sample processing unit That was not today's sample processors on your own pen or on your smartphone Then it was a building the size of the ICTP and one hour of CPU time cost officially $1,000 actually the universe never paid it was in a grant money this black at night They would let it run with nobody but in principle so in principle these people in Australia spent $700,000 700 hours at $1,000 an hour just that I would lose my bet So I always joked that I've drunk the most expensive bottle of wine in history because this bottle of wine cost $350,000 on the other bottle also $50,000 of course it didn't really cost that probably because $100 the actual bottle, but that's the story of the bet I'm not sure I prefer it, but it's kind of what I often end up doing is first of all it says I'm actually very good at it I often I mean really concrete problems like recognizing numbers or solving some identity I'm good at I enjoy most mathematicians don't care about it And if they do they're not as good and so I have a certain reputation and quite often people ask me Sometimes by email I've many papers people have never met that somebody sends me an email I've a problem and somebody said you might be able to work on them sometimes I can do it and sometimes I can't Some as it takes a day some as it takes six months, but I like that kind of challenge I mean it's not the main main object in mathematical life to solve challenge problems, but it's fun And so it's happened very often and sometimes those I become the car through the paper and sometimes I say no I don't want to be a car through because I didn't contribute enough, but they said no your solution is important Then I write an appendix and the appendix is written by Don Saguir The pending is in a proof of statement 4.2 And so I have I think something like 25 of my papers are not papers by me There are appendices by me to papers of other people, but often the other authors say no no this is more than appendix We really this was a crucial point and so you should be a car through and a few times It's been very embarrassing because then people assume I know that subject because I'm the author of some these papers are quite well No, and very important but not because of my contribution, but I solved some problem that the office needed