Interview with Sterl Phinney

0:00 It does actually work for the tape that I used to use. So anyway, it's ten past eleven on the 24th of March, and I'm talking with Stroll Finney. And it's actually working, so. Well, as I explained, the subject that I'm particularly interested in is the calculations of source frequencies for projects like LIGO. And I know when I was here as a graduate student, the canonical figure that we always used was derived from your 1991. 1991 paper of three per year out to 200 mega parsecs for neutron star black hole binary so I was kind of interested and one thing my impression is that while that was sort of all the canonical thing for years and that's the only largely the only paper that we quoted nowadays there's sort of a flurry of papers claiming different more optimistic theoretical results so I was kind of history of this, and in particular, just at the beginning, what was your original interest? Was it purely just gravitational wave stuff, or was it...? No, I guess two remarks. One is, the only thing people remember from the paper without ever actually reading it is the 3 per year to 200 megaparsecs, but the paper actually quoted conservative limit, which was 100 times less, and theoretical maximum optimal limit, which was 100 times more. And basically what's now happened is that other people have realized both of these things that I said are theoretical optimum rates that are much higher and observational pessimistic limits, which are lower. So now there's more papers arguing either for the the high end and the low end of the 303 options that I gave were basically the same reasons. So, I guess the reason that I got interested in the field was basically the discovery of Pulsar 1534, the second double neutron star binary system in the galaxy that would merge in a huffle time.

2:30 and the Holtz-Taylor pulsar is on the other end of our galaxy in an anomalously bright pulsar system so rates based on one object which is so far away and so anomalous are essentially meaningless and of course that didn't prevent people from having like Mandelweig having calculated rates based on that one object shortly after it was discovered I guess in the late 70s, early 80s but those rates were actually calculated wrong even though it's one object that sort of doesn't matter. But the mathematical way they had gone about computing the rates from the data was completely wrong. So when the second one was discovered, and it was already known then also that there was one system in the globular clusters, the 21, 27 plus 11 c, so that we had three objects in the galaxy galactic system that were going to merge and be candidates. I thought three objects was enough that it was worth writing down a paper that explained how you should actually go about calculating the rates given the pulsar observations. So there were two steps to it, which I think I did right. One was just pointing out that you should use a 1 over Vmax method to actually get the rate emerging in the galaxy in the popular cluster systems. And so you calculate the volume out of it. You can see the pulsar in the pulsar survey. and then the number of systems in the galaxy of that type is the effective volume of the galaxy divided by the volume that you've actually surveyed for pulsars of that type. And then, so the total number of the galaxy multiplied by this ratio is your best estimate of the total number of systems of that type of the galaxy. And then you divide by the total lifetime, and the point that I made is the total lifetime is not P over P dot of the pulsar, and it's not the gravitational wave lifetime, of the lifetime from now to the future, merger, plus the time from when the system was created to the present. And you can estimate the latter from the gravitational wave merger time in the former, the sort of previous history, from the spin-down age of the full swath. And you can add them together to get the total rate. So we're off again, and you were saying that... Yeah, so it should basically add up the total lifetime in the system and then divide the number in the galaxy by that lifetime to get the rate in each type of object.

5:00 And the reason you want to consider different types of objects is, of course, the pulsars have a luminosity function. And the thing that was new about 1534 was its luminosity was very much lower than that of 1913. And it was much nearer, but at that time estimated to be about 20 times closer. so the volume would be in a flat galaxy that would then be 400 times as many of those pulsars as of the 1913 type which is much brighter and that was sort of consistent with what we knew about the luminosity function of pulsars which is roughly a number brighter than L goes as L to the minus 1 and this was 100 times fainter than 1913 and we were estimating a few hundred times more in the galaxy so that basically boosted the observationally and determined merger level and was sort of starting to get close to the reasonable theoretical estimates of the rate. So for that reason I thought it was sort of interesting to write a paper just to point out that we now have observational evidence based on three objects for a rate that was sort of actually interesting for LIGO. So even if you didn't believe any theoretical stuck with this rate. Now, at the same time, it was very important to point out that there were a large number of uncertainties in this rate. First of all is the pulsar distances. They're usually determined from dispersion measures, which are notoriously bad. It's impossible to copy of Wiles' e-mail in the folder, so in case you have any questions. Oh yeah, the distances based on dispersion measures are notoriously bad, and easily a factor of two uncertainties in the distance. just translates into a factor of 4 in the area or 8 in the volume, depending on if there's a scale that is large or small. Right. And there are individual cases you can point out when the dispersion measure does much worse than that, even if it's wronged by a factor of 5, which would be 25 in the volume. On average, it probably does better than a factor of 2, but if you're interested in one

7:30 particular case, you'd better be prepared to accept big mistakes. so I pointed out we should put that in we don't know the pulsar breaking index very well the statistics of two objects of course there's just cosmic variance or whatever you want to call it being accidentally unlucky exceedingly lucky of having something unusually close to you so for all those reasons if unwanted I sort of put in limit, which was if you sort of said that, well, the dispersion measure distance was off by a factor of two, and we were unlucky at the 10% level, then the, I forget, I put it in various other number of things, anyway, sort of multiplied them all together in the worst case scenario to produce what I thought was sort of a real rigorous lower limit to the actual rate you could sort of possibly imagine having with this. And I think nobody has ever even now suggested a rate that was lower than that. And some of the rates that have been, I mean, recently, basically what's happened is 1534 has now had its distance determined by measuring the dipole formula, orbital decay, whose rate is limited by the galactic acceleration, which then determines the distance to the pulsar. So if you assume that relativity is correct, is calibrated from the original Holtz-Taylor pulsar, had a distance to 1534, and that distance is twice the distance that I assumed in my paper, and actually three times the distance the four recent papers that assumed for the 1534, basically telling me that indeed we were unlucky in the dispersion measure distance and the object is twice as far away as nominal dispersion measures would have led you to suspect. Right. So that factor of four then is part of my conservative estimate has been validated. We don't know yet about the unluckiness portion, but that was sort of one step in that direction, which is how we actually know properly the dispersion major distance was wrong. Now, the other thing that I pointed out in the paper was that you could also have grounds for optimism in the other direction. And there were sort of two things that I pointed out in there. One is you could have an upper limit to the rate, if you imagine that every neutron star

10:00 ended up in a binary, then the rate at which these form would basically be the rate of type 1b and type 1c supernova, so you know there's that rate in the universe or in fact a few from supernova searches in galaxies, and that's an absolute upper limit if you think that all neutron stars come from supernovae, and all supernovae and binaries are type 1b loss of hydrogen envelopes and binary transfer, then this is an upper limit to the rate, and basically any pulsar kicks could unbind a lot of the binaries, and the evidence from looking at pulsars is that most of the binaries do get unbound from kicks, but the actual rate of supernovae in close binaries of the type that could make these is certainly less than the type 1b and 1c supernova rate, so I pointed that out. And the other thing that I realized in that paper, which I believe was for the first time, is something that people should have noticed a long time before, but hadn't somehow, which is that if you have low mass helium stars as the progenitors of the neutron star, so stars which were not very high mass and have now had their hydrogen envelope stripped in the binary evolution. They lose their envelope during the spiral-in phase when the companion star spirals into the progenitor of the neutron star and leaves behind the helium star. And if the helium star is less than about three solar masses, it turns out that in its late evolution, I guess, the carbon burning, it expands like a miniature red giant and will cause a second spiral in of the other Nichron star into a very short orbital period, maybe 15 minutes orbital period. And since there are more low mass stars than high mass stars, it seems likely that for every system with a high mass helium core, which might produce 1913 or 1534 type objects, there ought to be many more low mass systems that would produce these very short orbital period systems and so I pointed that out in there and basically the trouble is those very short orbital period systems would have such short lifetimes you wouldn't have a hope of seeing any of the galaxy as pulsars at any given time but they might actually be dominating the rate because there might be ten times more of

12:30 these being formed per bubble time than the longer period systems but it's just that lifetimes would likely in a single galaxy to have very many of them on at a given time. So I pointed out that if you included that, that might be an additional component to the rate. And Ed Danden Heuvel kicked himself for having failed to note this before, and then proceeded to publish five papers on how we were just seeing the tip of the iceberg, and there ought to be lots more of the short period systems. And that, I think, has sort of continued to be some sort of industry with people going of stellar, binary stellar evolution models with parametrized forms of the spiraling energy efficiencies to calculate what exactly the orbital periods would be under these particular model assumptions and seeing how large a population of these short period binaries would be formed. So I guess the most interesting thing I think that sort of happened since that paper, I mean most of the other things I would say They were sort of bells and thistles which were particularly surprising and sort of burying the paper somewhere. I guess the most interesting thing that I think has happened is that the precision mass measurements of 1534 show that the two neutron stars have masses that are equal to within a few parts in 10 to the 4. in 1913 they're equal to about a percent in the 1534s of few parts and 10 to the 4 and that seems to me sort of interesting and a little surprising it makes me take more seriously the models that Gary Brown has been proposing that the way you make these systems is not in the way that Van den Eubel has been proposing that you actually have to have two stars with nearly equal masses which actually both go into common envelope phase at the same time. So rather than having sort of first one star evolves and becomes a neutron star, and the neutron star spirals in the envelope of the other one when it goes, the idea is that they have to be sufficiently equal that they both swell up at the same time and actually become a sort of double common envelope system at almost the same time. So that then the core masses are very nearly equal, and it makes them, if final neutron star mass is correlated with initial core mass and the initial mass of the stars, a little bit less surprising in those types of models,

15:00 but the masses of the neutron stars are so nearly equal. Yeah. Because in 1534, they're more nearly equal than the total dispersion we know of allowed neutron star masses from all the systems, which range from 1.34 to 1.41. So there's at least sort of a several percent range in here in the system that are equal to a few parts in 10 to the 4. So I guess I would believe that that system would actually form or something like that, or nearly equal mass scenario. The reason he has to do that is he wants to, he argues that in the other systems, you make neutron star, low mass black hole binaries, which I think that particular version of their story I think has a lot of trouble because they would predict that we already ought to have discovered among the other star population 10 black hole neutron star binaries and we know zero. so that's either if we're very unlucky or there's something quantitatively if not qualitatively I guess the other I guess the second half of what I thought I contributed in the 1991 paper was once you'd established the rate of merger in the galaxy and in the globular cluster systems the question is how to extrapolate to the rest to do a fairly careful job sort of illustrating how you would do that by actually adding up the blue light density in the universe and the assumption there is that the properties of basically we know that the star formation rate scales quite accurately with the blue luminosity of this for fairly massive stars which produce most of the blue light in the universe so the two assumptions you're making in that scaling are that the initial mass function between 3 and 15 solar masses doesn't vary wildly from galaxy to galaxy and the properties of binary star orbits don't vary wildly from galaxy to galaxy. So given those two assumptions then if we know if you add up the blue light density in the universe you basically get star formation rate in formation rate, since we know from our galaxy, the ratio of these pulsars formation rates figured out in some way eventually, to the star formation rate in the galaxy, the blue

17:30 luminosity of our galaxy, just scale the rest of the universe using the blue light. And, I mean, a lot of sort of gravity wave theorists tend to say, well, our galaxy is one galaxy, we know how many galaxies there are per unit volume, but that's a terrible thing to do because they're an infinite number of galaxies per unit volume because they're essentially infinite in many very faint galaxies and so what you really have to do is integrate over the whole galaxy luminosity function to get the total star formation rate in all kinds of galaxies and you can easily make mistakes of factors of five to thirty or something by just pretending that all galaxies are like the milky way in fact most of the light is from ones that are much bigger or much brighter than the milky way and starburst galaxies are sort of anomalous and contributors in their infrared, you know, like, put in a 30% correction for the dust-shrouded starburst galaxies, which, of course, 30% of the star formation rate is occurring, but you don't see them optically at all, because there's some covered dust. So 30% corrections in this game are unimportant, but at least in principle, I think, sort of, the roadmap for how one would do it when the statistics of the number of Polsars discovered warrant a bit, because it was arguably premature to put in some subtleties. Well, as you say, one particular figure tended to be picked out from the paper. I mean, I can remember in at least one paper making use of maybe the more optimistic estimates. But apart from that, there was a kind of canonical figure picked out. How soon was that picked up? Presumably the influence there is LIGO. Yeah, I think basically what happened is I think that maybe even before the paper came out, I don't remember, a line appeared on Kip's diagrams, which were shown everywhere. They use 203 per year at 200 megaparsecs of LIGO advanced and regular detectors or something. And I think that sort of, I think people were basically quoting the paper off Kipp's line or something like that. It was all that was remembered, even though I tried, even in the abstract, I think it fairly carefully gave the ultra-conservative and ultra-optimistic limits, as well as the sort of best-guess version. I guess that Neutron Star Binders had already been, or Neutron Star Mergers, had already been a kind of favorite candidate for LIGO for some time at that point.

20:00 I mean, you were watching that with the motivation. Actually, no. I think, I mean, basically when there was just one object in 1913, there had been a paper calculating the rate by Vanden Heuvel and others, as I said, quite early on. But, I mean, that paper was sort of, was so obviously wrong that I think people had discounted it. And also, when there's just one, I mean, quite rightly, when there's one object, you don't really trust the statistics of event rates. And so I think it, you know, it had been something that people realized could be a source of gravity waves, But it's sort of as something to which you would actually take seriously as something that LIGO was guaranteed to discover. It was a goal for sensitivity of LIGO to guarantee. I think it wasn't, so far as I could tell, really, in the culture in that way. Yeah, you couldn't. sort of when you had three of the objects and you could sort of start, you know, to make some, and then a factor of 10 anyway, sort of reasonable estimates of the rates, and I think sort of, and the fact that sort of accidentally the curve ended up, my best estimate rate ended up somewhere between LIGO early detectors and LIGO advanced detectors, sort of made it a useful thing to draw on the line to argue for the need for advanced detectors or something. And so it just became sort of the standard source, for better or worse, even though I think we probably should have emphasized that one should make it a very thick red band covering the optimistic, the conservative, rather than the single narrow line, the best he has. So, you know, it was a possible source in people's minds, but it was only at that point that it was possible. I mean, my involvement with LIGO was sort of peripheral. That was sort of my impression, and obviously the people who were actually in it would know how seriously they were taking it before and after. Sure, yeah, I'd be interested to ask them. But still, my impression also is that while the idea of the neutron star mergers had been around for a while, it was only relatively vague in the game that it became such, as you say, a kind of crucial, crucial test.

22:30 So I think it was really the discovery of Bolchan's discovery in 1534 that catalyzed all that. I mean, it motivated both me and Narayan and Shemey and Braun in order to write these papers. In fact, it's sort of funny, actually, that it was kind of a divorce or something. When I started, I mean, I did a sort of back-of-the-envelope version of that, and then I, because Ramesh had the best sort of code for calculating the volumes surveyed in various surveys, I wrote to him and asked him if he wanted to write a paper together about this, and he said, oh, well, Farhan and I are already writing lines. And then there was actually some sort of bad feeling, because I sent him a draft of my paper and somehow he kind of lost an email or he lost it or something or other and I don't know what happened and then when I sent him the final draft of my paper he was deeply upset because he thought I hadn't been writing one or something even though I had sent him the first draft two months before but eventually it also got sorted out and then it was a long-term illustration of the dangers of sending papers by email and assuming that if somebody doesn't respond So it was so the discovery of new objects more or less sparked a good deal of interest. Yeah, the discovery of that new object basically, as I said, 1913 was, I think it's still probably the second brightest recyclable Tsar or something like that, so it's way out of the luminosity function tail. So, you know, as the luminosity function of Millis, when it was discovered before any millisecond, so people had no idea what the monosity function was, or even that it was anomalously bright, but as time had gone on it became clear that it was anomalously bright, so I don't know if other people had thought about it, but if they had, they should have realized that probably there were many more finger ones, or when the first of the hopefully many more finger ones was discovered, then sort of immediately it became clear that this was part of a much bigger population. You know, the usual problem. Well, I guess it's the usual problem if the theorists were confidently went around saying, well, we know the luminosity function, and this object is bright, and there ought to be many more of the fainter ones the observers would move all over them and say, well, we haven't

25:00 discovered any yet, so there must be something wrong. But as soon as many observers discover one, then of course it's much easier to take the case. Well, as I say, my impression working in the kids' group, of course, was that for many years we more or less stuck religiously to this canonical estimate of the source we can see drawn from your paper with very few exceptions, probably and so, but perhaps because of that it's maybe a completely biased impression, I have the idea that that it's only in more recent times that there have been a lot more papers on revised estimates, as you say, many of them going back over old ground ramps. So I was wondering if that's a reasonable impression and if so, why there would be more interest now, what would have sparked that? Let's see. Yeah, I don't know what's... the distressing thing is that there haven't been any more observations. In fact, the Australian All-Sky Survey for a short period pulsars discovered at least a dozen neutron star point dwarf binaries, but not a single new double neutron star binary. Bad luck. The southern hemisphere doesn't have any, who knows. There really haven't been any new objects. It's a little bit surprising, given that the number of other millisecond pulsars has sort of tripled and quadrupled since that time. It's hard to know. I mean, I guess what's happened, what's been happening, I guess the thing that has sort of sparked it, as the population of millisecond pulsars has grown, some pulsars with B-star companions and pulsars with low-mass helium companions high-mass helium companions, the numbers of all of those systems have increased enormously since those days, basically because of this Southern Hemisphere Park survey. There are now about 35 of these systems. And because there's now such a big, you know, there's still only two of the double neutron star systems in short periods. There's now 35 other objects and maybe 20 neutron star, low-mass helium star systems.

27:30 So people have gotten interested in actually population synthesis of actually, suppose you lay down a certain number of types of binaries and you have a model for their interactions and mass loss and spiral in, then the field decay of the pulsars and everything. What do you predict for the population? And can you understand the relative numbers of all these different types and the number of systems with a function of orbital separation? And so there started to be an industry of writing little codes to put in prescriptions for all of this stuff, and then synthesizing the population. And one of the byproducts of that is a prediction for the number of double neutron star systems. At the moment, the observational motivation is more all the other channels of objects, but one of the channels leads to double neutron stars. And because these people are sort of aware that it's an important thing for LIGO, there's always a section of the paper which says double neutron star systems. I guess I've always been sort of skeptical it's interesting from the point of view of reverse engineering what it tells you about these complicated messy hydrodynamic interactions and binary stars I guess I've always been a little skeptical of sort of taking it too seriously going in the forward direction of actually imagining that whatever prescriptions they put in for this modeling actually gives you something that you would really take seriously as a prediction. And I think it's interesting to do it and then compare to the observations to see what you can say about the various models you put into the computer code. That part seems to be very valuable whether you get all excited because my code produces twice as many double neutron stars as your code. Well, the only way we're going to know which code is right until we actually discover enough of them to really know what the ray is. So I still think ultimately either LIGO will solve the problem by determining what the ray is and then we'll know the murderer for that channel or else. There are several new much bigger Pulsar surveys coming online with multi-beam telescopes in the Indian giant meter wave radio telescope. These are a third the size of our sea boat but steerable. And And I think the number of these kinds of systems is going to go up by another factor of five or something over the next few years, and we'll get better statistics and learn something from that. But I guess the other thing that sort of is, there have been lots of papers that people's

30:00 interest in the problem of the errors and dispersion major distances, again, that's the statistics of the number of pulsars has improved. People have gotten more concerned with that. So I think after my paper appeared, the rate actually went up from what it was in my 1991 paper for a while, because in 1993, Taylor and Corb just published their reanalysis of the dispersion-measure distance relationship, and that moved the predicted distance of 1534 closer. I think I used 600 parsecs, and the preferred number for a while in the late 1990s was 400 parsecs. Of course, now it's gotten shoved out to 1.1 kilobarsecs on the actual measurement. It just gives you an idea of a sort of level of uncertainty in all of this. Of course, then people are calculating statistics on these synthesized models with various models with a disparate measured distance relationship is sort of another industry that keeps people busy. The, well, do these estimates based on kind of population, extrapolations, are they by and large more optimistic, less optimistic than estimates based on? Well, I mean, again, I think, I mean, I would say there, the uncertainties in the models uncertainties in the dispersion measure distances were, so at the same level of factors of 10 or 30, I think one doesn't take them too seriously. I think, you know, probably, I mean, a lot of them are more optimistic, basically for the reason that I pointed out, that there is this very short orbital period channel, which a lot of the papers reproduce in their models as they ought to, and so there's a substantial body of things going through this than that has to the LIGO rate without changing the pulsar detection rate. So that I think is the reason why I don't really know that there are any other channels that have been discovered in these binary synthesis models. I mean basically the point is that if they predicted a rate that was a factor of 10 higher or lower in the easily detectable wider orbital period of Pulsar or traditional Pulsar channel, I

32:30 think it could easily be accommodated by the observations or you could say it was mistakes in the particular models of the spiral and the code and you can't distinguish at the moment. So a factor of 10 level, I think, you know, you can't really say one way or the other, whether the models or the small number of statistics of the observations plus the spur of the major distances or the weak causes of any discrepancy that might be present. And the short period channel, I think, seems to me fairly compelling that one ought to have grounds for optimism, but again, it does depend on, you wouldn't have it if you took Jerry Brown's scenario really very seriously that all the other systems could make black hole binaries if the stars are not very nearly the mass or the 12 solar masses or something, if all the others make black holes, then you're missing a big population of double neutron star binaries. Of course, it's still wonderful for LIGO. You just now replace all the double neutron stars with a neutron star 1.5 solar mass black hole, and it's great for LIGO, and it's so good for both stars. But that series of models, I think, are the only ones that are sort of really qualitatively new, I think. I guess I personally don't believe the neutron star equation will stay as soft as he needs for that series of models. But it is interesting, I think, that that's the only model I've seen which actually rationalizes this very nearly quality of the neutron star masses. So would there be anything that's happened in the intervening seven years or so that would make you inclined to feel that the best estimate would be revised up or down much? Well, I think to the extent that one takes the new orbital p-dot derivative for 1534 4 is a good estimate for the distance, which seems reasonable to me, then I think the natural thing would be to adopt that distance rather than the parsecs or whatever I did, and that would push the rate down by a factor of about 4 or 5. So I think if I were redoing it now, the best estimate rate would be 4 or 5 times lower

35:00 based on the observed systems and probably have more confidence in the short period channel, so I might move that up in prominence and mention it in the abstract instead of just in the discussion section. So maybe the best estimate rate is about the same as it was before. You mentioned black hole, neutron star, I guess there, well, in that instance, obviously, one doesn't have the... We haven't seen it yet, so you only have these theoretical estimates. My impression is, but at the moment, I haven't had a chance to do more than a glance at a bunch of abstracts, is that papers estimating frequencies of black binders containing black holes very widely. Well, I guess, first of all, is that pretty much your impression? Yeah, I think that's right. Of course, that freedom is allowed from the fact that we haven't discovered any yet. It's also because I think we have a much less clear idea on what sorts of stars actually make black holes. There are basically two sorts of channels. One are failed supernovae, or the star which tries to make a neutron star, but then extra material falls back and it finds itself over the maximum mass of a neutron star and it becomes a black hole after trying for a few seconds or something to make a neutron star. And depending on how much falls back, that could produce, you know, if it's just the core that falls back, you get black holes as low masses, 1.5 solar masses, that would be the maximum mass for a neutron star. But probably more naturally, if it fails early on, if it failed sort of after it had sent off a supernova shock, but the additional fallback just tipped it over the edge. If it actually failed in the first eight seconds, so it never launched a supernova shock at all, then you would expect that a large fraction of the whole star would fall back

37:30 and you would end up with five to ten solar mass black holes, which are the only black holes we know about in the universe at the moment. We know about eight of those in some systems like Cygnus X1, where there's a very high mass companion in the black hole and others in these more interesting systems where you know the masses much more accurately there's a relatively low mass or 0.5 to 2 solar mass star 16 something or other and A0620 and all the other low mass black hole mine areas and those are all sort of in the range of 4 to 10 solar masses and those Those must have formed by fallback of a large fraction of the helium core mass of the star on the central object. So we know that those kind of objects form. We don't know yet of any of the 1.5 solar mass type, which Terry Brown has helped be more numerous than neutron stars. And the trouble is basically that we don't really know what the conditions are if you look at the history of computer codes of simulated supernovae, there have probably been more failures than successes in simulating a supernova, and most of the time the explosion doesn't work. And so if you argue that, well, perhaps the nature follows the computer codes, then most supernovae fail. On the other hand, maybe the codes are failing because they're not putting in the right physics, and even nature succeeds all the time. And then it's a question of whether it's very high mass cores, whether stars above 30 or 40 solar masses actually sort of systematically produce black holes, whether that could be reasonable, whether the same is true for, if they've had their envelope stripped in a binary, whether that affects the core evolution, it all gets very murky. I mean, the sort of late stages of the evolution of high-mass stars are very rapid and quite sensitive to multi-dimensional things and how you treat convection and reaction rates even that aren't understood very well. So I think it's really hard. One can make very nice stories that all stars above 25 square masses make high-mass black poles, and it works the same in a binary, and you get a reasonable rate. But you can imagine making quite different stories that have to be equally...

40:00 reasonable and consistent with what we know, although less simply attractive. So those two channels, basically, if you just, I mean, the way I guess the argument would go is if you sort of adopted the view that stars are more massive than 10 solar masses make neutron stars and above 25 they make black holes. if those black holes fail early they get smaller kick velocities than the neutron stars do. The ones that fail late probably get about the same kicks as neutron stars. So then you would preferentially retain them in binaries, so even though there are fewer of them, the actual number of binaries could be about the same as the number of neutron star binaries. And then depending on the details of the spiral-in involving more massive stars, you could quite reasonably arrange that the merger rate would be as comparable to that of the double neutron star systems. But you could equally plausibly make it higher or lower by changing the spiral end parameters. And you might have a huge population of them, which Kerry Brown is writing most of the time. The supernova does fail in the collapse of the supernova 1987A failed, and that's the reason there's no central source. There's actually a 1.5 solar mass black hole there, even though it was a supernova, even though it had the right number of neutrinos for making a neutron star and everything else. Maybe after 20 seconds it failed and said, oops, the neutron star collapsed to a black hole. So is interest in compact binaries containing black holes pretty much solely confined to possible sources for LIGO, or is there more in that general? Oh, no, no. I think there's two. I mean, Jerry Brown was interested in it because it would prove that his, and actually it started with Mark Law, he's a professor here, theory of the nuclear equation of state is correct. He'd like to know what the maximum mass of a neutron star is so it can find 1.5 solar mass black holes. That's another way of getting a neutron star mass. So I think, I mean, that's sort of an interest and I think the same people who are interested in making supernovae if you knew about a big population

42:30 of black holes it would tell you interesting things about failure of supernovae in nature as well as failure of supernovae in computer code so supernova people are quite interested in what the population rate is it has a big effect on things like galactic nucleosynthesis heavy elements of the galaxy one of the arguments for making all the stars the masses in the galaxy go into black holes, is if you allow them to go supernova, you actually overproduce some of the elements in the galaxy. So you'd like them to all fall back into the black holes that are being sprayed out in the interstellar medium. And again, there's uncertainties in the wiggle room, but that's been one of the traditional arguments for making the masses into black holes is to avoid trouble with overproduction of our process elements, basically. So I think, I mean, there are a lot of reasons to be, it's not just like a sort of peripheral interest of the young astrophysicist and now are more sufficiently aware of it than when they write a paper that they mention as a prediction, but I think the driving interest is not actually that. The, well, I suppose an obvious follow-up question to that would be, Do astrophysicists anticipate much from live? I mean, as the project kind of gets, you know, moves along, are people thinking that interesting things are going to come out or are they not really holding their breath? I guess it depends how traditional... I mean, I think the overall astronomy community, as you doubtless to know, was violently anti-LIGO. I mean, some of it was just bad feelings and historical accidents, I think. I mean, one of the crucial, I mean, one of the things that I think was crucial about the double neutron stars for LIGO in some sense was that for a long, a long time, the advertised source for LIGO was supernovae. And that had a very bad history because early estimates, sort of the natural estimate of a supernova explosion, if it had a quadrupole of order of unity, it should be about the same amount of gravity waves

45:00 as an emerging neutron star system. But when axisymmetric supernova calculations were done, they remained, because the supernova collapse basically is hydrostatic, the core collapses almost hydrostatically. they produce almost no gravitational radiation, sort of 10 to the minus 5 of what you get from a double neutron star. And so, in fact, if you think that there aren't any sort of... Subsequent fallback doesn't make a practically rotating core that develops a bar and produces a lot of gravitational waves. If it's just all of them all just collapse, LIGO will probably never see a supernova. Yeah. And yet that was one of the early estimates. So people, I think sort of the older generation of astronomers like Jerry Ostreicher and so on, sort of remembered this history of supernovae. And as soon as the axisymmetric calculations were done, they just thought LIGO was totally hopeless and ludicrous. And the double neutron star systems are sort of a guaranteed source, which, where you can imagine that the graph, because it's all in the two-point masses orbiting each other, But as long as you got the rate right, it's hard to argue with the amplitude of the signal. Whereas in the supernovae, there are factors of 10 to the 5 arguments about the amplitude of the signal. Or is this one there isn't? So I think that's one of the things that sort of made it why it was so attractive for the sellers of selling and propagandizing the LIGOs. At least to get the signal right, if you had a rate within a factor of 10, it was much better than the supernovae in that sense. So it was a much more attractive source. But on the other hand, I don't think there's, you know, the other reason that I think the astrophysicists sort of objected to LIGO was the fact that it was called an observatory, where's the actual number of sources it's guaranteed to detect, astronomers being very conservative and nothing exists unless we've seen it yet, might actually only be a few per year. And if you had, if you build a telescope and you could only detect three stars per year, you'd be out of business in three months, basically. So I think that was, you know, they thought that this should be called physics, and they were very worried that the fact that it was called observatory, it was going to slip into the astronomy budget, and then LIGO costs more than the two eight-meter telescopes that were being built for the nation, and

47:30 maybe that would take all that money away and stop it, stop it, I think that was it. You know, on the other hand, I think sort of more reasonable people who were interested, particularly one who were sort of interested in these post-binary systems actually are sort of looking forward to finding out what LIGO sees. Right. sees a rate that's most optimistic levels, and no need to be dominated by this channel, which is basically a so short log, you don't see it electromagnetically, you only see the final states in other galaxies, but the initial state in nearby galaxies. You know, if it really just sees, or if it sees lots of neutron stars going into 10 solar mass black holes and you get the A over M's, I mean, that would be really, really exciting, I think, if you want to go forward to that. But is it going to have a big impact on astronomy as a whole? I think the answer is probably no. There's only a very small community that's actually interested in these sorts of binary stars, as a sort of pulsar community, which is a pretty tiny one compared to the microwave background and the higher-edged-shift galaxies and large-scale structure or even interstellar medium communities. I mean, it's a tiny little group of people. And then if the other, I mean, it really won't have very much effect on what anybody else does. Because it's like, except maybe if there are a lot of 10 solar mass black holes which I don't see out sort of the board of one, redshift one or so, it's not going to have a big impact on cosmology. It's all a local universe, and it's basically telling you about binary stars and local galaxies, and it's not that interesting in the Milky Way. I'm sure they're not that interesting in local galaxies, so it's going to be over there. So unless you're going to see... So unless you start to see some dramatic evolution with redshift and lots of merging 10 to 20 solar mass black holes with redshifts of one and a half or something, then I think people in cosmology would start to get interested in telling you something about star formation, right? and massive stars in their own starburst galaxies or something. That must make me a little bit of a long shot. I mean, I think the real hope is if I can discover something and we've all sort of failed to suspect the sources or some population of supernovae who had a fallback and made some go far unstable and produced fantastic amounts of stuff and then another group of supernova people

50:00 that get interested in the subject. I guess Lee Lindblom and Ben Owen have been working just recently on instability, where you have a radiation reaction dominated instability, so I suppose that would be a possible news source, in fact, I suppose maybe the first one that people have come up with. Yeah, though I guess that, I mean, I guess it's not clear whether that would be good as a gravity wave. I mean, a lot of it sort of comes out over a fairly long, long time scale because it's limited by the sort of viscous amplitude, so whether it would be the impulsive source. I mean, it's actually quite interesting as a, potentially as a limit on the initial spin period of neutron stars. So I think that that paper, it would have had even more impact if I hadn't pointed out that shortly before they wrote it, a pulsar with half the spin, a young pulsar supernova remnant with half the spin period of the crab had been discovered and sort of ruled out draft one of their paper already. So otherwise they could have predicted that it shouldn't have existed, but it unfortunately existed, so they had to modify the paper a little bit to incorporate it. But still, the fact that they had to modify it telling you they were sort of getting close, you know, it was quite close to being quite interesting. whether either we sort of failed to understand something about the viscosity or temperature history of the neutron stars or you're really pushing up against these limits of the observed sources. I'm not sure that that will ultimately have too much impact on my gut. Right. In any case, that would still remain something of interest within sort of the pulsar community as opposed to the broader. Well that's great, thank you very much. Thanks a lot.