Interview with Steve Detweiler

0:00 Okay, so it started, and it's the 18th of August at 1 o'clock, and I'm speaking with Steve Detweiler. Well, usually a good way to start out is with a sort of a historical question. I'm familiar with your early work on the Tchaikovsky formulas and looking at radiation reaction problems, because it was a big foundation for the work that I was doing as a graduate student. So I was wondering how you got involved in that sort of work. So you mean, when you say the Tchaikovsky formulas, you mean particularly, say, particles orbiting black holes, looking at the radiation and stuff. Well, I guess I started out more doing my first, the first research project I did was perturbations of neutron stars, and I looked at Thorne stuff from the 60s. And I don't know, after doing some work on that, then I switched over and started looking at black holes more. and I think probably, I seem to recall that I started working on scalar fields around rotating holes before the Tkalski stuff came out, but maybe it was right about the same time, I don't recall. Because scalar fields were known to be separable and stuff around black holes well before the gravitational fields were. So I was doing some calculations, trying to look numerically for instabilities in scalar fields around black holes. And then when the Tchaikovsky stuff came out, it seemed natural to really be doing the gravitational wave stuff and also to be, there was initially the real question was this business of floating orbits, whether or not you could have enough radiation coming out of the event horizon to keep a particle in orbit at a fixed radius. which I don't recall now, but I think Tchaikovsky and Press did enough work to sort of convince themselves it wasn't going to work. But anyway, I don't know, I sat down and it wasn't particularly difficult. Well, the Tchaikovsky, now let me restate that. The Tchaikovsky equation itself is not an easy one to deal with. Okay, so I'm going to restate that. I must have started, before I did particles orbiting, what I'm trying to recall is at some point I started working with Chandra on this whole thing.

2:30 And I'm trying to recall now whether, I guess I did start working with him. He was the one that got me first working with the Kalski equation itself. and he had a he had his own way of doing things and always and he I wish I could remember how all that went I don't know, he got me fiddling around with the Tkolsky equation and after a while we had a formalism where we reduced the Tkolsky equation itself to more or less an equation much more like the Regi-Wheeler or Zerulli equation. And in fact, took those limits in an appropriate way. When the curve parameter goes to zero, our equations would reduce either the Regi-Wheeler or the Zerulli equation. And once I had that under control, then it was pretty easy to write code to, say, do circular orbits, just because one of the problems with the Tricolsi equation itself is this asymptotic behavior. You know, far out, close to the event horizon or for large values of R, The two solutions are growing and falling at different rates, and that was just a big nuisance. And by having this formalism where it all reduced down to potentials that, say, fell off like 1 over r squared at infinity, made that whole task a lot simpler, and I sort of worked out a lot of that stuff with Chandra, and it was pretty easy to just consider, say, a single particle orbiting a Schwarzschild hole, and in particular then look at energy fluxes at infinity and down the horizon and stuff. As I was doing it, I, of course, was thinking, wouldn't it be nice to detect gravitational waves? But I know I wasn't doing it thinking, this is what we'll detect with gravitational waves. And I didn't have any, I don't think I was getting out any information at all that would have indicated that gravitational waves would be easier to detect or harder to detect or whatever. that it was a problem that I could do that hadn't been done before and was interesting. So I figured, what the heck, I might as well do it. I guess that was my motivation. That was sort of a particle orbiting a whole, and then I did some other things. I think I started working on that when I was a postdoc at Caltech, in fact, in 76.

5:00 a particle in circular orbits, and then I had a graduate student in Caltech that same year, and I, Gene Sednich was his name, did particle spiraling into Schwarzschild. We used the Tkalski formalism because we had all the machinery set up for it, and just set A equal to zero, and did a test particle spiraling into Schwarzschild, and looking at the radiation from that, and I don't know, I guess that was my start on all of this, anyhow. Right. So you convinced yourself that there weren't any floating orbits? Yeah, that was really easy to do, in fact. And, in fact, as I recall, there's a part of it, in order to do those calculations, you need the separation constant for the angular eigenvalue problem. And I got the code for that. I think I got it from, well, I can't remember his first name now, The last name was Larimer, I think, who had looked at the floating orbit problem, concluded that they didn't exist, or maybe he told me this afterwards, but he'd looked at it and concluded they didn't exist and then thought it wasn't worth publishing because they didn't exist. I don't know if he did all the fluxes and stuff like that as well, But I may have gotten the code to do the angular eigenvalue separation from him, that piece of the code. But the floating orbits, I think that was an amusing thing that you could, since you saw energy coming out of a rotating hole, it was an amusing thing you could think about. It was not all that hard to calculate. Then since they didn't happen, well, that was too bad. but it was some motivation, I think, that made the problem a little bit more interesting. Again, since we weren't particularly interested in a real source of gravitational waves, I would say it was the kind of calculation that needed to be done, but it wasn't anything that people were really anxious about having done. Because you could do back-of-the-envelope calculations and get an estimate of what was going to happen, and there weren't any surprises. So you just get the numbers more accurately, but it wasn't tremendously enlightening, I guess. Looking back at the origins of the floating orbits idea, I got the impression that, at least as far as, for instance, Misner was concerned,

7:30 who I think was one of the people who brought it up, part of the idea was to provide a possible explanation for some Weber's results. Was that at all influential? idea so uh i i don't i what i recall is is uh missner's idea about geodesic synchrotron radiation the idea is a test part uh you think about synchrotron radiation got a particle whizzing around in a electric charge in a tight orbit and it beams this radiation and i think he wanted to explain weber's results as there's something orbiting a black hole in the in the center of the galaxy and every and it's got this beam of radiation and when it hits us we see the radiation and that was a way to get around the problem with Weber's result is that if the galaxy were emitting radiation that he was detecting isotropically then the galaxy would have been gone all the mass would have gone away so he had this way of beaming it basically but then when the calculations were done I think Paul Chernowski did the work with him work that out they just didn't see the beam and the real problem is that in for synchrotron radiation you've got a charged particle in a strong magnetic field and there's a lot of acceleration and a test particle orbiting schwarzschild the acceleration isn't all right i mean it's a geodesic but in any sense uh the frequencies just aren't up high enough and i think if you could uh i think it's true that if you put your finger on some test particle and make it, well, what I want to do, if you had some way of forcing an object to move in with much higher angular velocity than a geodesic would allow, if you had rockets, whatever it is that's falling in, maybe then you could get some kind of geodesic beaming, or you could get some of the synchrotron radiation with the beam. But nobody really... I think by the time I got involved in these things, radiation ID had pretty much petered out and that there wasn't much going on with that. So is there a point you were saying that the mid-70s or so the idea of detecting gravitational waves was only invariant to the background as far as motivation was concerned?

10:00 As far as those calculations go there was of course a lot of interest in detecting gravity waves and in fact In fact, when I was a postdoc at Caltech that same year was the year they interviewed Ron Grever and hired him to sort of get the thing going at Caltech. And he gave his talk, and I asked him at the end of the talk, you know, if he had any explanation for Weber's results. And I talked to Weber enough when I was at Maryland that I knew he was an honest guy, and he was trying really hard. issues of just how you're going to do your data analysis and not look so hard for a result that you find it. And anyway, I remember there was a lot of interest in gravity waves. But the calculations people were doing, by and large, I think we were doing them just because it was the best we could do at that time. And they were just interesting problems. Things like what Larry Smarr was doing with his colliding black holes, I think it was the same idea, that what we wanted, of course, was to find that some source was going to give a large amplitude, a lot of radiation, and then maybe we could detect it. So we do what we can. Larry collided black holes head on and saw that there really wasn't much. and I don't know, I wasn't surprised by that because when two things sort of fall head on they hit and it's all over whereas it seemed to me that the real hope was a binary black hole system, comparable masses not a perturbative calculation where they're sitting there going around each other for a while and then you've got a chance to actually see some radiation that orbiting black holes was the calculation to do and that that would have some real effect on maybe how we ought to look for gravitational waves. But I didn't have a clue about how to go about doing it and sort of the best we could do then. And I'm not sure if it's any better now is perturbation analysis. So I guess that's one of the things

12:30 me interested in perturbation analysis around black holes. It was that related to things we're interested in. It's not really solving the important problems, but I didn't see any progress any other way. And again, in some ways, you solve the hardest problem you can, and maybe it'll be interesting. So in the 20 years Now, has the various detection efforts and how they've advanced and become bigger and more important and so on, has that reached the point where you find that the problems that you choose to work on are more directly influenced by possibilities of detection? Okay, so I'll admit, and this is probably just because by now I've been around in this field for a long time, and I see a whole lot of, there are a whole lot of people who are young, have a tremendous amount of energy, who are beating this problem of how do you find a signal and some noise? They're just beating it to death, and I don't think I've got anything to add to that. So that might be my real motivation, that I don't see why. I couldn't do it. I don't think I'd add anything, so I'm not going to do that stuff. So my own thought has been, I'm going to keep solving the hardest problems I can solve that are the most interesting to me. And I keep telling myself that this is completely independent of LIGO, that if LIGO disappeared or never detects anything, I'd like to think that the work that I'm doing will have some interest, say, 10 years from now. I think a lot of the calculations that people are doing, which are very important for LIGO, are so tied into the detector that it's the kind of thing that in 10 years won't be interesting. The detectors will change in some way. So I guess it's because I've been waiting, well, so I've had a standard joke I've told that we're going to detect gravitational waves in the next five to ten years. And I've been making, and it was true about 25 years ago when I started saying it, and it's still true today. And I don't know if that will ever stop being true.

15:00 but my own interests I've just decided that I don't want to focus on issues that are tightly tied to LIGO but I'd much rather stand back and work on more general problems which whether I'm successful with them or not at least have the possibility that sometime in the future there'll still be interesting problems that I'm just not interested in doing work too closely tied to the equipment nice to see that something I've been interested in for a long time is taking on that maybe we will be able to detect gravity waves. And I like the least idea, personally, just because that's the regime of small masses falling into large black holes and that's more closely related to what I've done. I would say the work I did on the quasi-normal modes, I don't think I'd say I was motivated by looking by trying to look for gravitational waves but it was sure nice that it seems to me that that's something, that's some work that I did that may in fact end up being interesting as far as the detection goes and actually as I recall that that was after I'd started working with the Tkalski Formalism and the Reggie Wheeler equation, I was all sort of well-versed in the perturbations of black holes at that point. It was just before I started working with Chandra, and people had been talking about the quasi-normal modes, and I guess Press had a paper, I think it was just Press, had a paper where he was, look, he's more or less found the modes for large values of L by just doing scattering off of the zerillite potential, is basically what he was doing, and he was sending in a pulse, and then noticed that the waveform coming back had sort of slowly damped oscillations in it. And at that time, people talked about how solving the quasi-normal mode problem, there's numerical instabilities because the solution, now I have to make sure I get this right, when you've got a damped oscillation, that means as you integrate out an R, For a t-equal constant surface, the oscillations are getting larger and larger as you go,

17:30 when you've got a complex frequency, which is damped in time. When you're out at a large distance, the radiation, because you're really looking at the radiation that was emitted sometime in the past. Anyway, there are numerical instabilities that make it hard to find these quasi-normal modes. And at that time, the general comment was there are numerical instabilities, and so we can't solve this problem. Well, I looked at it and thought, well, if the damping was very weak, then you could find them. It would be like having a – it's the equivalent to a quantum mechanical problem of putting a particle inside of a – not a square well, but you've got a well with barriers on either side. Put a particle in there and it very slowly leaks out. You have outgoing waves in both sides and you look at the damping of the probability function or whatever, the wave function inside your barrier. And so the whole question, it seemed to me, wasn't a matter of principle. We could do that problem and the only difference between that problem and the same quasi-normal mode problem for Schwarzschild was that the potential for Schwarzschild isn't this nice thing with a well in the middle of it. It's just a big bump, and so the wave isn't bound together. So that says you've got the lowest modes or the best modes are going to be damped away pretty quickly. But it's still a question of practice rather than principle. So how weakly damped are these modes, and I found that with a fair amount of effort, I could sort of integrate the appropriate equations and make sure that I wasn't being troubled by the instabilities and I could still identify the lowest frequencies of the normal modes. And I was real pleased when I did that. I then gave a seminar at Chicago, and nobody seemed interested at all. really, I thought it was something really neat, and in the audience was Chandra, Jim Ipser, Cliff Will, and Bob Grosch, and they all sort of seemed pretty, I don't know, so I walked away thinking, well, I thought that was neat, but what do I know, and about three

20:00 months after that, Chandra said he was interested in this quasi-normal mode problem, and I thought i'd done it and i had but uh i don't he got me going on it again and he had a different way of integrating the equations which was um oh technically it was done the computer code that he wanted me to write the integrate the equation it was different but it was pretty much the same kind of problems or difficulties and instabilities and stuff but anyway we resolved the problem again and then it then when i did it with him all of a sudden it got a little more attention and people seem to seem to it made a little bit more of a splash I don't know I think I'd have to say that out of the things I've actually done that might have an impact on LIGO in the future it's probably that quasi-normal mode the calculations particularly for the rotating holes but that's not going to be early LIGO since this is sort of the tail of the waves that we're seeing as it dies off that's not going to be for a while I was a little bit curious about some of the division of labor that you seem to see in relativity work between when you have collaborations such as you just spoke of with Chandler where one person is doing the coding and the other person is maybe just helping with the algorithmic side. I guess I'm just kind of curious sort of the sociology of that, if you like. I mean, first of all, if it's mostly just a generational thing, it often happens that you just have older physicists who are interested in making use of numerical techniques but don't have the experience with it. I think I'd say that in this day and age, most of us do have experience with it, we older guys, but we're tired of it. And after all, what are graduate students good for? On the other hand, I personally haven't had much luck working with students. And it's always been the case that, I guess I would say in general, the difficulty with a student is you tell him to go do something, and then he goes and does it. And that's not really what you want.

22:30 What you really want is a student to figure out what the real problem is and go after that. And that's usually something rather different from what the more senior person is saying. I don't know, that sort of happened with the work I did with Chandra. That happened a couple times. I mean, at some point, he had a way of, quote, simplifying the Tchaikovsky formalism that was going to reduce it to a simple scattering problem with a well-behaved potential. But his scheme, in order to do it, in order to find the potential, you had to integrate, you had to solve a differential equation on a computer to end up finding the equation that's equivalent to the Tchaikovsky equation. differential equation to generate the real differential equation you're looking at. And I thought, well, this is pretty silly. And he wanted me to work out the code for that. And it wasn't working very well, and I just knew that that wasn't the right way to be doing it. And there was one time where I visited Chicago, and we spent a couple days together, and he outlined all this stuff, and I didn't get any peace at all. I was continually wanting to know how the progress was going. It was real frustrating. Then I left, and I went back to Maryland. Then he would sort of call me about every day, and I was able to say, oh, the computers are down. I'm not making any progress at all on it. Meanwhile, I was madly working away, trying to figure out a better way to do the whole problem. and in the end so I did end up with a sort of a simpler version of a way of rewriting the Tchaikovsky equation which in the long run I don't think is all that important but it was certainly better than the scheme that he had developed and I know I couldn't have done that if I'd been at Chicago while I was working on it I just wouldn't have the time and I guess just what I like to is that I was proud of myself that I could keep them at bay while I could think about the real problem that we were interested in solving rather than this particular version of it.

25:00 And particularly with coding problems, if you don't have a grasp of the physics of what's going on, it's really hard to write code that's going to work the way you want it. side of that is if you do understand the physics, it's much easier to test what you're doing as you go along. And so I've found I'm still writing my own code for stuff that I do. And I've got a couple of graduate students I'm working with now, and they write their code and I write mine, and then we can compare our results and stuff. And I sort of like the way that goes. And in particular, if a graduate student disappears, I've still got and I have some confidence in the way it's working and stuff. But my own thought is that these big collaborations like this, the Grand Challenge business, that trying to work together in a big group, particularly with software, complicated code, I don't see that as really being the way to solve problems. It might be okay if you're Microsoft and you're writing an operating system or something. of people that to do all the parts of it but uh and i think the benefits of the grand challenge has not been that there are a lot of people all working towards a common goal it's been there's a lot of money that gets pumped into the field and then it goes to individuals and i see individual people out of that whole collaboration that have done some interesting things but i don't see that the collaboration itself had much to do with that other than as a conduit for getting money out to and resources out to the people doing the work. And that might change someday, but I don't see that it's changed yet. I don't know. Maybe my remarks are, maybe I'm saying these. I've never been a real fan of the Grand Challenge. And I'll admit some of that sour grapes because I don't have anything to do with it and it's not my style or anything else. but as I look at the result of it there are specific things that have been real interesting like the stuff Matt Choptwick turned out there are specific results that are interesting but as a whole I'd say that the Grand Challenges

27:30 didn't live up to what it was supposed to be doing I don't know what they claim to have done but my guess is that they can orbit two black holes for a fraction of a revolution around each other before things get fouled up. I think I'd also say that actually the problem they're working on is the wrong one. I think that trying to sit back and just use brute force on that two black hole problem, that we're just not ready for it yet. I believe what I'm going to say now is true. I've read it, I think, from a reliable source. and that is that all the improvements in the last 30 years for solving, say, Laplace's equation, that the software using things like multi-grid methods and stuff like that, that those are responsible for the improvements in our ability to solve Laplace's equation, that those improvements are much grander than the improvements in hardware. And we know how much hardware has improved in the last 20 or 30 years, the algorithmic improvements are still more important. So using today's algorithms, using yesterday's computers, we're better off than using yesterday's algorithms with today's computers. And I think that's the way it is with a black hole problem, that you need to focus on what problem you want to solve and understand the physics of it. And I don't think that waiting for the next generation of computers is going to do it. And I think it's going to be somebody really understanding how that system works and then figuring out a clever way to get the computer to solve that problem. So could one draw a parallel between your experience with Chander where it could actually be a disadvantage to just sit down and bury yourself in coding up the plan you've already got when you maybe might be better spending time? Well, I'm sure that's the case. Well, okay, so it's that way with me. I think most scientists are a strong individual tendency. Everybody wants to be, I think it's safe to generalize. We all want to be the boss. We all want to do what we think is interesting the way we think it ought to be done. And I think that it's just not a good procedure to have sort of a corporate structure where you've got the higher levels in the hierarchy telling the lower levels

30:00 what to do. I just don't see that working unless you've got a very well-defined project where you really can break it up into little bite-sized pieces and have somebody go do it and then expect that when you put it all together that it's going to work. And I think for this sort of front-line new stuff like trying to get black holes to orbit each other, We're just not at that stage. So I think that the, I think, first off, the best scientists want to do what they want to do, not what somebody's going to tell them to work on. If nothing else, we've got big egos. And I don't know, I just hate it when somebody tells me what to do or what things I ought to be working on. And I don't think that's unusual. So that mode does seem to work in, say, high-energy experiment. And maybe that's because you really can take a graduate student or somebody and say, okay, build this piece of equipment. And they more or less know how to do it and go out and get it built. And when you plug all those pieces of equipment together, now you've got your particle detector or whatever it is. Maybe that works. But there it's, I think that it works because the engineering can all be done and that you really know that when you get it together it's going to work. But with this black hole stuff, it's so, I don't know, I think the problem is just too hard for us still. We're not at the point where that sort of approach makes sense, is my thought. It's still at a stage where you can't really automate the problem to the extent of breaking it down into these independent problems. Right, right. And, yeah, once you can break it down, well, then you can farm it out. Right. And then a big operation makes sense. But I don't think that any of the research going on right now with gravitational waves really fits into that category. I mean, I think the interesting things that have been done in the last five or ten years have been things like, oh, what Eric Poisson's been doing with looking at the details of, say, particles spiraling into black holes. And, again, even though it's a perturbative analysis and stuff,

32:30 and he's directly relating it to LIGO a little more closely than I would if I'd done the work, nonetheless, it's nice work that just sort of stands alone. Yeah, and it's more or less done by a small, like the important things are all being done by small groups of people still. And I don't see that changing anytime soon. I guess every time, so I was at one of the LIGO meetings they had last spring was at University of Florida. And I just sort of went and sat in because they let me at some of the data analysis sessions. because there was a room with people that I know have done sort of interesting science and they've got all this focus on, you know, they've taken the different tasks of templates and stuff and broken them up and different people are supposed to be working on different parts of it. And I guess I'm glad they're doing it because that's all an important part, but I don't regret not doing it myself. Yeah. Yeah. I don't know. It's too much like that thing of, you know, not quite being told what you have to do at least. Here's the stuff that we have to do and we're all going to have to do it. We all got to do part of it. And again, I'm glad they're doing it because it needs to be done. And I imagine when the paper comes out, there'll be a hundred names on it. And I don't know, I guess I just wouldn't get much enjoyment out of that. And so for you, actually, what I suppose one might call the traditional way of working for relativists in this small group, as you say, is still actually the most effective. Oh, yeah, no question about it. Actually, at Florida, we're really too isolated, more isolated than I'd like. And I'm glad to come to a meeting like this one just to hear some other people and this whole business about the radiation reaction that we've been talking about the last few days. I'm certain that what I'm doing is the same thing that other people are doing, although it's clear that they all have a different language, different style and stuff. And I'm not sure that they understand the methods I've used to attack some of these problems. And nonetheless, it's really good to sort of sit down and hear what they have to say in their approach. So I think I'd say I like to work in small groups. It's nice to have good interaction with people and to push ideas back and forth.

35:00 But when it gets down to really getting something accomplished, I don't know, I think a small group of one or two or three is usually plenty to really make progress. and particularly for something like writing code where, I don't know, I think people have done studies that show that the bigger the group of people writing the code is, the longer it takes to get the code written once you get up to a certain size. But all the time is spent not actually writing the code you're supposed to do but dealing with interfaces between all the little bits of code that are being written. And so the big computer projects, and this is really sort of an operating systems and stuff big computer projects are still best done by a smaller number of programmers adding more people, more programmers isn't necessarily the right way to get the job done maybe it's make life really good for the few programmers get them so they don't have all the extraneous stuff bothering them so have fewer people concentrate harder on the problems And I'd say that's the way it goes with science. How much of an obstacle is the difference in language that you mentioned between different groups of researchers? I mean, you mentioned that people here are using somewhat different language and terms. Is that just a minor obstacle to be overcome if you really want to make use of what they're doing? It's not so difficult to... I guess I tend to do things a little bit differently from other people and I find that if I can't somehow get it worked around to be using the same kind of language that they're using that what I've done won't get noticed. So I think the language is really that. It's communication. And if you can have two people doing the same thing in, I think I'd say, different styles, and they may not even know it. So I've found that that has been a problem with some of my own work. I think in general I'm not sure about that. I think if I'd been keeping closer tabs with the people who are here at this meeting, for example. So at the end of this meeting, I now know how to go back and rewrite my own talk

37:30 and use the right kind of words in the midst of it to make it sound a whole lot more like some of the things that they were doing too. I know the right phrases or whatever. But I think when it gets right down to it, that it doesn't make much difference as far as the work goes. Sort of like what we were saying earlier, I feel that a lot of science is somebody finally understanding some work that somebody else has already done, into their own language or their own style of doing it. And I think that happens all the time in science. I don't know. I guess it keeps us all busy. And I was particularly interested when you did the quadrupole formula stuff, just because about that time, I'd finally really understood the quadrupole formula. And I was thinking to myself that I didn't really understand some of the other approaches that other people had, but I understood mine, and to me it would have, in some sense, made sense to sit down and write it up as a paper. But on the other hand, I knew that it would just look like another paper of, here's somebody who finally got it. And so, I don't know, I didn't write that stuff up. But on the other hand, I was pleased to finally, what I felt, really have an understanding about the quadruple formula, how to use it or whatever. It was amusing to me, just about the time you, there's a little history of it, it was finally making sense to me at a level that I felt very comfortable with. Was that arising out of your perturbation work? Yeah, in a way it came from the perturbation stuff. I started, so I've had a long-standing, a problem I've wanted to solve is the two black holes orbiting each other. I mean, everybody wants to solve that one. And then there are just different methods. And one method I've had along the way is to simplify the problem by not having radiation going out at infinity. If you lose radiation, the system evolves, and then you've got a big, complicated mess on your hands. But if you artificially put standing waves at infinity, and what you mean by standing waves is a little bit tricky,

40:00 but basically don't let the radiation go off to infinity. Bounce it back on the system, and then look for a quasi-stationary system where the holes go around each other, radiation goes out, but radiation's being pumped back in. the object's orbit, but there's no evolution going on. So that to me makes it, that's a clean problem. You can formulate it in a way that's real clear, and that problem seems tractable to me, because you're not doing the time evolution anymore. You're just trying to find, if you found the geometry at one moment, you just lead transport the geometry, where you've got a d by dt plus omega d by d phi killing vector. And once you've got your geometry, that's just the way it's going to stay while the holes go around. So in some sense, it reduces to solving the initial value equations with these standing waves in the gravitational field. So it seemed to me that that's an idea I had a long time ago. And I've been trying to implement it ever since in one way or another. And when I got onto the quadrupole business, I decided to try to do it using a post-Minkowski kind of formalism, similar to the stuff that Blanchet and DeMort were doing. I guess my thought was that I've never liked the post-Newtonian because I don't like this expansion in small power and the overseas. I don't mind weak feel, and that's why I was thinking of post-Minkowski. small v over c. And partly that's because from a perturbation problem, the test particle can be moving at any speed you want. So you're not limited to small v over c. And that realm then between, you have post-Newtonian comparable masses, slow speeds. You have particle, test particles, small masses, high speeds. And what we really want to do is combine them and get large masses, large speeds, we can't do that. But maybe somehow using post-Newtonian, post-Minkowski ideas combined with perturbation analysis, maybe we can build up to start out with a big mass and a small mass, but let the small mass start getting bigger and bigger

42:30 in some kind of a perturbative, iterative scheme. and that's sort of the approach that I've had in the back of my mind. Most of these things, once you get past sort of the first order in the perturbation theory, it really becomes complicated partial differential equations, and I don't have any experience solving those. So I decided a while ago that I was going to stick with problems where I can separate variables and just integrate ordinary differential equations because I feel very comfortable with that. So that's limited the kinds of problems that I can work on a bit. But, I don't know, there's still so much that we don't understand about two black hole systems that I'm still finding a lot of fun, as long as I'm finding it fun. What's important? I was curious, actually, if the work you did on the perturbation problem, looking at radiation from small particles around big black holes, in the 70s, if you found any part of the quadruple formula controversy impinging on that work you were doing. Well, I thought about it, and it was clear to me, I mean, the way that would impinge on it is take that test particle and move it out to a distant orbit, and you get the quadruple formula, and that's real easy to do, and you get the right answer. And so I believe that analysis. And then there was sort of a classic paper with Ehlers and Arnold Rosenblum, and I'm not sure who the other authors were. That's right. Where I think because Ehlers' name was on it, that paper got some attention. And it basically said maybe we don't know all we ought to know about the quadruple formula. I guess I hadn't paid much attention to it. I mean, I'd read the land-downed lift shits. And I guess I didn't see what the big deal was. I'm not sure I understood the issues involved. And from my own experience, namely the perturbation analysis, I didn't see a problem. So I pretty much ignored it. But when I got around to doing the two-black hole problem,

45:00 and I was trying to do this in a post-Minkowski manner where you have two big black holes far apart and you start out having them move around each other more or less in flat space, and then I had this idea of trying to build up in an iterative way, adding on the gravitational field from the black holes onto the Minkowski space. And when I started doing that, I real quickly got into early papers by Kerr in the late 50s and some EIH stuff. And the EIH stuff didn't make much sense to me, but Kerr's papers really did. And I sat down and read those real carefully. And then I found, I guess what I liked is that he wasn't using post-Newtonian. It was really a post-Minkowski kind of thing he was doing. and maybe it was just because I had a lot of respect for Kerr in the first place and that made me pay a lot of attention to these papers but as I recall there were two of them that I read pretty carefully and the first one, the earlier one of the two to me really approached the problem in a way that made sense to me and I think it was really looking at that paper that I came to grips with a quadruple formula myself how to do those sorts of problems. But I guess it all boiled, it really boiled down to, again, the EIH question of you've got a black hole or a strongly, you know, some object with a strong field, and how is that going to move along a geodesic in what space-time, even if it's real small, a small object, how do you show that that moves along a geodesic in a bigger space-time? There were got some interesting issues. Again, curse analysis I liked a lot. It just sort of clicked with my own thinking of the problem. So that's what I thought. So it sort of relates to what you were saying earlier, that each individual scientist has to sort of come to a realization of the problem in his own particular way. I sure think so, and I think having that any other way just doesn't work. It always helps to, if somebody else has solved a problem, and you sit down and you read it and you can understand it, that's always good. But I find that science is hard enough, but that's pretty rare.

47:30 I mean, usually what happens is I get an idea, and I start working on it, and then I get confused, and then I'll go read some other stuff, and I won't quite get what they're doing, and work some more, but eventually I'll find out that maybe I do understand work that other people have written. It's more of a... I think passively you don't get it. Science is too hard. Sitting back and just reading it, you get an idea, but you don't really understand what's going on. But you sit down and you try to do it yourself, and you see how hard it is, and then you go look at the answer in the back of the book. You go read somebody else's paper, and then it... Often, if it's a good work you're looking at, that you can sort of make some sense of it and finally come around to, at some level, understanding it at a deeper level, I guess. And would you say that there can be a sympathy then between the individual styles that you spoke of earlier? So, for instance, Kerr's work in particular appealed to you as something that you could relate to rather than, say, EIH or... Yeah, yeah, that's right. And I'm not sure if it's just why that happened. But when I saw his, all of a sudden, I think he was more or less approaching it the way I was trying to approach it. And so it was pretty easy for me to take what I'd been doing, and it was easy for me to see what he was doing, because it was the same kind of thing I was thinking of doing at the same time. Whereas I know the EIH stuff, I didn't. and I'm trying to remember now just the things that bothered me about it and it might have been this question of small speeds because most people right away make that the assumption of small speeds and I've never I just have never liked that and Kerr really kept away from it as long as as I recall his second paper he had small speeds but the first paper was a straight it was much more and you didn't have as many sort of real results. But it was this nice putting sources on... Gee, I'm not even sure I can get it straight. But putting sources on sort of weakly curved backgrounds,

50:00 you know, Minkowski space plus something else, and then seeing how the sources move in it. and when collaborating with people to continue this theme of different styles of working that people have do you find it easier to work with people who have a similar style of your own or can it be a help to have contrasting styles in collaboration I don't know, I think it's probably more personality than anything else, it's not even so much style but just sort of on a real personal level, you get along person or not i've uh i was working with a with somebody a bit ago and it just seemed like every idea he i'd have you tell me why it wouldn't work i didn't want to know why it didn't work i want to know how to make it work uh it just seemed like it wasn't worth my while i mean and some of the sometimes he was right sometimes he wasn't but it was um i don't know i guess i i've always been an optimist because if you aren't if you're a pessimist you don't make it in science you think oh I can't do that well you won't so I think I would say that the really crucial thing is to get along with people on a personal level and the collaborations I've had over a number of papers have all been like that just sort of people whose company I enjoy and I think you get people together who can talk to each other, then they're going to come up with more or less similar approaches to a problem and you get enough sort of communication back and forth. I think that's really important. I think one of the things I've read, the work I've been doing most recently, at Florida our group's pretty small. It's not as active as it ought to be, and I've had a lot of trouble getting people. I could really use somebody to just go and talk to about every other day about what I'm doing, and I haven't had that, and that can be really useful, I've found in the past. If nothing else, just to have somebody sort of sit there and nod their head while you're talking and every once in a while have a puzzled look. That's about all you need sometimes to just sort of have a framework

52:30 where you try to make your ideas clear to somebody else, and then that always makes them clear to yourself. So I'd say I've missed that in the last, oh, about five years or so. I could have used some more of that kind of interaction. So perhaps the key reason why it's, of course, by the most important thing I don't know if the person is that being in personal contact is the important part. I really think so. Yeah, I think so. I don't even think it's necessarily getting ideas, sometimes questions you need from somebody else. Why are you doing it that way? But the amount of feedback you really need is pretty minimal. But without any, it's hard if you're just sort of sitting there. And as I've been working on this most recent stuff, sometimes I wonder if I'm just way off in left field. In fact, as I came here to give my talk, I was afraid it would be viewed that way. I mean, I knew that nobody else was doing things the way I've been doing it. got some results that nobody else has, so that's always nice, but when you work, it's a fairly big project, and when you work on a big project, there are always sticky points, and as I was preparing my talk, I thought to myself, well, I was trying to think of places where people would complain and say, oh, what about, you've got to show this, you've got to show that, and I had about four of them, and sure enough, one of those things people focused in on, and it was an important point that I was aware of, and I sort of thought about it, and I thought, that really, I had a way of working out in my mind where I was pretty sure it wasn't going to make any difference, but I hadn't really nailed it down, and that was out of the sort of four or five things that I was concerned about, people making comments about sure enough that that that was the the point uh and so that's that that's made coming here just to talk to other people that's made it really valuable to me because i'm i know it's not going to be a big problem i'm going to go back home and figure out in a week or so how to uh how to

55:00 deal with this issue uh but there are really sort of four or five things that i thought people might be concerned with, and the fact that the others, I'm not going to, the other issues that weren't raised, I'm not going to take as evidence that I'm doing everything right. But it isn't blatantly, it isn't something that just sort of stands out. And it was clear when I came to this particular point, it was just real clear that I had glossed over a place where ran into problems, and I wasn't aware of that just because I hadn't been following other people's work enough. So I think that interaction is really crucial every once in a while that you really need that people help, giving you some feedback and stuff. It's too easy to just go out in the left field and wander around for a while. So there's a kind of a tacit interchange between people when you have a face-to-face contact. So probably, if I judge what you're saying correctly, just say email contact or that is probably not really a substitute. Now, I don't think so, and writing papers in some ways is not really a substitute either. It's hard to sit down. stand-up but you don't get the uh you don't get the same out of it as you would get out of it if somebody sat and talked to you about it and and in fact it's even better to talk to somebody than it is to hear a talk yeah hearing a talk is still pretty close to reading a paper if you can you can sit down and somebody can start talking about the problem usually they're immediately into details and what you want to know isn't details at all you want to know the big picture what you you know, what are you doing this for? What good's it for? And usually what you get in the talk are all those details, and what you want to hear is, again, the overall idea behind it. And that's really where I think the sort of face-to-face interaction with people is really good, to really get at what, again, the big picture is going to be. And that's always more important than the details.

57:30 Although on a day-to-day basis, we're all busy solving these little problems, and those are the details, and those are things that have to be done. And they're always, when I'm working on something, it's that I've got a detail of the day that has all my attention, and it seems to me that if I can only do this, I'll have everything. I'll finally have everything in control, and then I'll get that done, and, oh, something else will pop up, and I'll have to go focus on a different detail. So I think on a day-to-day basis, we're focusing so narrowly that we then miss the big picture and so getting together to talk to people at meetings is really it makes you it pulls you back from those small details and gives you a much better overview of what's going on and I think it's really important to try to keep that in perspective Interesting Yeah, that's very interesting before we have to get back to the session. Okay. But that's really very interesting, thank you very much.