Interview with Joloien Creighton & Patrick Brady

0:00 There we go. I'll just quickly say that it's nearly 5 o'clock in the evening of the 10th of February, 2009, speaking with Joanne Creighton. So what I gather is that a new software standard has been proposed by the LIGO lab that everyone should follow for data analysis structures and that right now people are engaged in writing new software and maybe rewriting old software to meet this standard. I was wondering kind of what the background, well, to start with, I guess I was wondering, is there a lot of software that needs to be rewritten to meet the standard? Well, let's go back to the beginning and say, well, why is the standard needed at this point? So I think the standard is needed because for the LIGO end of things, it's probably just because they're planning to run code on their systems. They have their own LDAS as their data acquisition software. and if they're going to run LSC-contributed code, then they just want it to be easy for them to use. The standard is nice in some ways because it means that if people in the international community want to contribute software then they do it in a somewhat formal way so that it's things tend to work together a little bit better so it has the advantage that it doesn't just make life somewhat easier for the lab but that makes it more likely

2:30 The input of the lab on the LSC software design has been rather minimal apart from specifying the document or doing some specification of the standard. But we haven't really heard much about how they're planning to do things. So I sort of think of this thing as just a useful way for everyone in the community to contribute software that they're, in fact, going to be using. I mean, people are building software for their own personal use, as well as to meet the milestones of the LSC. And they're going to be able to want to port the software onto their own machines. So I mentioned that people in Potsdam, or Goal or whatever, will be writing LAL code for their own data analysis system. They'll also contribute it to LIGO if LIGO wants to run it on their system. But they'll be able to use the software written by everyone in their system as well. So it's more, I think of it as more portable to everyone. Whereas I think the people at LIGO sort of have a notion that the ultimate goal is that it slots into their system, which is an ultimate goal. Right, but in addition, it's likely to be advantageous if everybody else can run everybody else's. Yeah, and I mean, not everyone's going to have their system. But when Joe Romano gets his Beowulf going, he's going to have a large computer in Brownsville that he can do a lot of data analysis.

5:00 The Beowulf here, well, we should be able to run things on. Just as an aside, I don't know if this is something you can answer any better than I can, but how far down the road is it likely to be before people not only have their own system that they can analyze data on, but, you know, can take the data and put it on their own system? I don't know. Sorry, Patrick. Yeah, you can. Right, so I think the, so you can at present do that to some extent, you can get tapes of the 40 meter data and look at them. You can probably also get tapes now of frame format Hanford 2 kilometer arm data, since that's been walking for a while. You, and presumably the current porting meter, when it's running, you can get data from that. But you can't get it at the level that people anticipate that they're thinking that there's going to be some sort of socket system where you can get data faster from the storage systems right are you taping? yep, I can turn it on you can turn it on so I guess I guess it's evolving I don't know when it's going to stop evolving, but at present, I've got a bit of hacked up code that just reads frame format data from the old 40 meter. So you mentioned that the LIGO lab had a certain amount of input into this new standard, but not necessarily all that much.

7:30 I was wondering who... So they sort of wrote the standard and just sort of dropped it out there. And part of the problem is they don't, I don't think they know exactly how things are going to be running at their end. I mean, Kent is busy so far as, from what I understood, he's rewriting MPI. He's writing his own MPI for their computers. And he's, there was some talk about how they might link the filtering routines into their code. But that kind of formal thing hasn't been filtered down to the code writers yet. so we're just we're just writing code that runs up to standard and runs but it's not yet clear that that will need to be completely rewritten in order to run on their system so I gathered that from listening in today degree, meaning that you're just beginning writing code that actually just conforms to the new standard, but doesn't have to particularly do anything yet. Yeah. And so, presumably a good amount of code has already been written under packages like RASP and that will, will there have to be a lot of rewriting done to... Yeah, so almost nothing can be taken exactly over. It doesn't take a lot of work to change it, but the standard does require adopting different types of conventions. and so you could write graphs too or you could think of this as some complete rewrite of all of grasp

10:00 but it might not do a whole lot more at present but it's more structured And are all of that, I suppose, are the various people in the LSE happy enough with this Andrew, given that it was just sort of dropped out? Well, so, I think they were pretty unhappy when they first read about it. I think that's because they didn't realize that it, in fact, wasn't so bad. They thought there would be a lot of overhead making it conform to the standard. And in fact, you can usually do that pretty easily. You have to rethink certain things. So then, I think Sathya, in particular, was unhappy with it. But then when he started actually coding with it, he realized that it wasn't so bad. I'm not sure if he's happy with it, but at least he, I mean, he's, David Church has produced a lot of Cove now for it. It seems to be, it seems that they're fairly comfortable with the standard now. And apparently they were talking to the people at AEI, and somewhat convinced people at AEI that the standard isn't so bad, that the people at AEI might consider writing their software to the standard, because they're all in, they're all part of the LFC as well as writing their own software, and so they've got commitments to provide it to the LFC, and they've got commitments to your GEOM. Commitments to GEOM.

12:30 So it would be useful if there were at least some level at which the code could be interchangeable. so that they might decide to write their geocode to the site. So my impression is that the archive that I'm maintaining is not, in fact, the LAL library. It is a repository where people put things, and I just keep it, and then at some point in time, LIGO will take bits of it and put it into their, into the real LAL after they've gone through it. Right, okay. So I'm not, so I guess it's going to be different, each different detector is going to take the bits of the code that they want and perhaps modify it a bit more to meet their own standards. So at the present, with the library or the repository that you have now, how much overhead does that have for you? I mean, when people put stuff in there, do you then have to sort of look at it to see how it conforms to the standard or whether it runs or not? Yeah, I'm not sure if I'm supposed to or not, but yeah, I look at it. and ask them to make it more like the other code that's in there, for example. The standard itself was not... There was a document written about it, but the document had a lot of internal inconsistencies, and you couldn't actually tell what the standard was, except for the fact that a few people had written code, supposedly to the standard, and if you looked at the code and read through all the code that had been written long enough, you sort of got an idea of what the standard really was, and it wasn't what was in the document. It wasn't what was in the document, just in that there was a lot that was unsaid in the document, or even in that some of the things didn't really apply. Some of the things that were said were wrong. or just it was it was the it was a document that had changed over time as

15:00 people had decided various things like what the type name should be at one point in time they all thought there should be underscores and then they decided that there wouldn't be but the document never gotten got changed so you had to actually look at the code to find out that in fact they didn't have Okay, so the actual standard is reflected in the examples more than in the... Yeah, yeah. So, that's interesting. And then does it evolve at all, or has it evolve at all, subsequent to the document or the examples? I mean, as people write more examples, is it more... Yeah, to some extent. Mainly, so I think the people here have been doing the most coding. with it. And we found that it needed a lot of extra stuff. We wrote stuff and then we wrote a whole bunch of macros that would make life easier. And then I've been encouraging people to use those macros every time they write a function. So it's more of a convention standard. And how, so, you were saying a lot of the stuff that's been written has been written here, how many, how many different individuals and groups have so far contributed into the library. Sort of very roughly, I was curious. Probably about ten individuals. nine. Nine people have actually written. That there's a software committee that's in charge of actually deciding what is within standard and what is not, and all this kind of stuff. But it moves pretty slowly, and I've been finding that issues arise faster than they can deal with them, at least at the beginning.

17:30 So I've been making a lot of on-the-fly decisions about how I think it ought to be. And then the committee comes along and it vetoes all of them, more or less. But it does it very slowly, and so I haven't actually... I mean, I don't think they've said anything that I've actually put into place yet because it's too much work. The other problem is that of the people in the committee, only two of them have written any software for this. And one of them is Alan and the other one is Sam. And Simon isn't writing any more software for this, as far as I can tell. So he's sort of out of touch. And Alan is writing software, and I think he appreciates the difficulties in writing software. No one else in the committee is actually writing LAL code. In fact, I don't even know if they've installed it on their own system, what's here. So, yeah, I can see that's a problem. So up to now, so as you say, in any case, whether they haven't, it takes them long enough to decide something that in the meantime you've already had to make some kind of a decision. Yeah. And does that sort of mean that ultimately, whenever it's convenient to get around to implementing their decisions that people will have to go back and rewrite code again that's already been written? So it could mean that people have to rewrite code that's already written which I think would be a waste of time for those people. It could be that the software committee just has to appoint somebody to go through and retrofit all the code to do it. So I could imagine doing retrofitting for small amounts of stuff.

20:00 But if they make big changes or things that would require big amounts of adjustment to the code, they'll have to find someone else to do it. They'll have to do it pretty quick. So one thing that took a very long time is they decided to change, so the one thing that they have decided on doing is changing the name from L-L-A-L to L-A-L, which meant that I had to change all of the code just so that they could have one letter less than the name. So that took a couple weeks. That's a lot of work. Well, I was... with one thing into the other, it took a couple weeks to get through. Yeah, but it does seem like it would be a real issue. Because obviously, in a sense, that's a fairly small change, changing the overall name. So you could see turn out to be a problem later depending on what kind of decisions they make. So is it another possibility that the committee would somehow in the end not be very relevant to the standard just because they wouldn't make the decisions until it was more or less I don't know I really don't know how this is how this thing is going to work out some of the people in the committee are in charge of the LIGO data analysis system so they're on there because they're the only ones who have a glimmering of the idea of how the interface is going to take place But, uh, but, I don't, um, and, and another person who's on the committee is on there because, because, for, for the interface with, with GEO.

22:30 Mm-hmm. Mm-hmm. And, uh... So there is a definite effort to keep Geo and LIGO coordinated? Yeah. Or a little bit. The... Just to go back to something that you mentioned earlier, that people were writing code obviously in the first instance for themselves how much how much does the various bits of code that people write have to fit together is it more that there's just a few pieces that'll be used an awful lot or are they all likely to be so I like them to fit together as much as possible And I think that's sort of the goal. I think the committee seems to feel that they've designed a certain set of basic structures and they want people to use those whenever possible. They've written the standard in such a way that you can't use those structures exclusively. You have to be able to make your own structures. structures, but they want their structures to be reused as much as possible. And so it's nice. For example, they have some notion of what a vector is. It's a type of structure they have. So that means that you can write a Fourier transform for a vector. And that That means that anyone else who wants to do a Fourier transform, you know, and they have a vector of the same type, they can just use the same routines. And so it means that people are less likely to move too far away from the structures because then they would have to write all the support utilities for their own things. so there's I think it's self-motivating in some ways

25:00 Right, as in the way once there's enough material there that conforms to the standard then it begins to impose itself But on the other hand the thing's organized into a whole directory of packages and each package is more or less contributed by one person or one group and so So, in that sense, they're somewhat independent, although they do call routines in other packages. So, right, so each group has their own set of packages, but because they all have access to each other's packages, by and large, there are plenty of cross-calls between them. So if one person decides to overhaul their package completely, it may mess up someone else. So there is some risk to having it too closely connected. So given that people had been making calls between the different packages, had there been a kind of an incipient standard previously that had arisen in a de facto way because people were making use of each other's... I mean, before, obviously there was a reasonable amount of code around before the new standard, the standard was imposed. And I was just wondering if the fact that some of that code from different groups may have already been intended to work together or at least call each other's routines, if there was a kind of a de facto standard or maybe even that had existed beforehand? Well, so to some extent, the code that was before was GRASP or people's own code that wasn't in GRASP. And then there were a few standard things in GRASP, a few standard structures in GRASP that would be used, but not many. And then the other thing that everyone used was numerical recipes.

27:30 And one of the things about the LAL is numerical recipes isn't used. So that produces a big void, you know, in a lot of the... Right. Basically, on the other hand, you only use, most of what people did only use numerical recipes in a few different ways. So there was only a few routines that needed to be recoded. So root-finding routines, and the Fourier transforms, and the big ones. okay so previously everybody just used that numerical recipes for FFTs or whatever you can't do that anymore because it just doesn't meet yeah so well for one thing numerical recipes is copyrighted so you can't distribute it with that permission and I guess the software committee decided at some point in time that they're just not rather than ask for permission they're just not going to do it there are some other reasons why you wouldn't want to use numerical recipes just because so one of the reasons is the functions aren't always re-entered numerical recipes And that can be a problem if you want to use functions recursively or in different threads, which are both conceivable within the standard. Another problem is that the numerical recipe's Fourier transform is not particularly fast. And there'll be a lot of query transforming in the data analysis. So... Alright, so it'd be better to have something more efficient anyway. Yeah. The... Well, I guess the other question I was interested in asking is,

30:00 that have been taken up by, you know, the kind of, you know, housekeeping chores of this kind of thing, you know, as opposed to, say, doing real science or... So, yeah, I've basically spent the last month doing this, trying to get something close to a release of this stuff. I hope that that's sort of a start-up problem and that you have to write a whole bunch of stuff in order to do anything, but then once you've got all that, it's fairly easy to go on. The problem is that people have to meet milestones within the LSC. code, and if it's not going to be some haphazardly designed thing, there has to be a set of basic routines already in place. Right. So, that's basically what I've been doing. Right. So, for the moment, you've had to do a lot, most of the writing to get the basic thing off the ground? Yeah. Yeah. I think it's basically, I mean, I don't think it's going to let up completely, but I think I'm going to be able to spend more time doing research and thus maintaining this from soon, unless the software committee throws a twist at me. Yeah, hopefully not. speaking of the milestones since obviously throwing out a standard requires everyone to attend to those kind of issues for a while so you mentioned dates for the milestones I mean when are they are they to sort of vary from group to group so it's kind of funny LSC has milestones and milestones and whatever and then and then each individual group also well

32:30 often has an MOU with LIGO which is which allows them to be in the LSC or to share data or various things like that and and it's a it's a weird stage where these things may not overlap. The MOUs that you have say, so all the MOUs say we will provide LAL software to do this itemizes everything. And then in the LSC you might have committed to do things that are not in your MOU or other people might have committed to do things that are in your MOU or various crossovers like this. So it's not it's not they're pretty close to being in sync but there's two reasons why you might want to produce a piece of LAL code it's either because it's on your LSC milestone or it's in your LIGO MOU right so the that keeps it really confused I was wondering, is there any kind of sense then with the people trying to write code that the standard is kind of too close to their milestones in that? I mean, obviously, I guess they knew they were going to have to produce it to a certain standard, but, I mean, is there plenty of time for people to now write the code according to the new standards being developed, or is it a case that you have people out there saying? Well, so the biggest problem, I think, is that there still isn't a document that really says what the standard is. The only thing out there that people can look at to see what the standard might be is something I threw together just before Christmas. just because I thought it was important that there be something out there that people could look at. And I want to get a release out now so that it's fairly up-to-date because that's all that people have to go by. But on the one hand,

35:00 on the one hand, the software community hasn't answered certain questions that need to be answered. to be able to definitively get the release right. So there's kind of a tension, yeah, between waiting for them to hand things down from on high and at the same time getting something available to people. Yeah, on the one hand, if they're going to make a big decision that impacts everyone, you don't want to. You want to wait for it, but you don't. On the other hand, if you don't think they're going to do that, then you want to get the release out fast so that people can see where it's going. And the other thing is, when code is contributed, it's best to get it out sooner rather than later, just so that people think that they're not throwing code somewhere and it's being ignored. somewhat guilty of, because I got code from Cardiff just after Christmas, and it hasn't really been included, basically because it was written too, or it wasn't quite up to standard. So in that case, do you just sort of send it back to them? So in this case, I decided to send it back because well, for one thing, file names were not done quite right. And if I put this into a CVS repository it's a little bit awkward to to change file names so if the file names are fixed and people just tweet the code and have it changed and that's fine the other problem is that they were using numerical recipes routines so it could put in And do you find that people are kind of chaffing at the bit to get a release, to see a release?

37:30 No, actually I hear very little from people. So I think most, I'm not sure who's supposed to have written code by when. I should pay more attention to that. I guess Maria Alessandra is going to start trying to organize as part of the software committee. But the only people outside of Milwaukee who have contributed real packages, so Sam contributed some basic routines, but then Joe Romano has contributed some stuff for Stocaster. backgrounds. And Sathya and David have contributed stuff for Inspiral Chirps, which I'm hoping to get in by the end of next week. And then the rest is open from Milwaukee. So it still mostly revolves around here. Do you think that's, I guess, obviously there's a good bit of manpower here. Do you think the fact that there's a good bit of manpower here, and people are all physically in the same place that they interact more often that helps a lot, yeah. But, but the, the other thing is that, you know, people, Sam is now writing software for, for the data monitoring tool, which is a completely different standard, different package for doing similar things. So it's not clear, uh, why they're There are so many standards being maintained in so many different ways and such. So it's actually a very different standard for that end of things? Yeah. It's in a different language. It's in C++, for example. And the language for LAL is C? It's just C. they wanted

40:00 I'm not entirely sure why I think they just felt that most people would be more comfortable writing in C and that they would be would find it easier importing C into their stuff C++ is a much more dangerous language So maybe in the case of the data monitoring tool, they just thought there might be a smaller group of programmers probably? Yeah, I think it was going to be more centrally organized. And so just on that note, does it seem as if, you know, obviously things are less, my impression the ACES and data analysis side that things aren't that centrally organized, so the paradigm, obviously, is having people out there doing their own thing and there's a standard, so does that seem to... It's not centrally organized in the sense that there is a laboratory making the software maintaining the software and doing all that kind of stuff. But it is kind of centrally organized in the sense that people do have task lists that they've agreed to do and milestones that they've set down and such things like that. Right. So, yeah. There's a kind of a collective level of central organization that's decided at the meetings where... Yeah. tasks are drawn up and people agree to take it. It's actually more organized than I thought it would be. Which is just sort of optimistic. You think that maybe people have gotten software written by the time it's needed. But it's not, I mean people, there hasn't been that much communication between the groups

42:30 in terms of what's been, what everyone's been working on. I mean, we've been working on the chirp filtering routines, so we've been talking to Cardiff a fair amount about, they're writing the Inspiral template code, so we've been talking a bit about how those two bits will fit together, but they're pretty well writing the code We're writing our code however we want. So it's not that necessary to coordinate between the two to make them fit together because there's a fairly clear division. Yeah, I mean, if in the end we assume we're going to get this and we get something slightly different, we can probably just accommodate. It's pretty modular, I think. Not that, not as much need for something. So do you think that, does it look reasonably helpful that all the software will be written for when it's needed? Well, I don't know when it's needed, what when it's needed refers to. I think I'm sort of optimistic that people will have software running on their own systems on engineering data, doing more or less what they want on their own machines with data that's sent to them on tape when the engineering data comes out or soon after. Whether or not there will be software at Caltech or on the sites that will be filtering data online when the first engineering data comes out, that's much less likely. when the first data run comes along that's a year later or something maybe they'll have gotten things up to speed by then presumably I mean something will be running

45:00 the thing is that in the best case scenario Run a filter, you know, in-spiral filter with a large range of parameters. In the worst case scenario, you'll just be able to run 50 templates near where you think things might be. It will be basically useless. But there will be at least something there online. And then the question is how much of it will need to be in follow-up, offline analysis, and how much of it will actually be on-site. That will be interesting to see. Thanks, Joel. Great, thanks. So it's 11 in the morning of the 12th of February, 2000, and I'm talking with Patrick Brady. And so, well, I don't know, I should start off asking how you like being one of the people responsible for one of the leading gravitational wave theory groups. Oh, it's brilliant. I don't really feel that way, actually, if you want to know the truth. I'm not sure I would do it. No, it's not I don't feel too responsible. I just don't quite feel like I'm sort of one of those leading lights. It's, I don't know, for me anyway, it's a big change, I must say. Yeah. I don't know. You'll notice it yourself when you get your job in the next year or so. Right. What's the big change? It's just a different set of responsibilities, right? You've just got a lot more different, a lot more things going on. So you've just got to keep yourself organized. You don't only have your own research, now you have multiple things. so it just becomes a new set of tasks that you have to take on board and make extra room for same amount of time new number of tasks I've already had that experience since I had a kid oh gosh yeah actually that's true I think that might be more difficult to work out but anyway Okay. Well, what are the main, my impression is that the main focus of the group is LSC-related issues.

47:30 Is that almost exclusive right now? So it depends. It depends which part of the group you take in, right? Of course, I mean, it's a big group here, four professors. and two are well not exclusively but Bruce is 90% committed to LIGO by contract almost I'm about 60% committed to LIGO by contract of research time I would say between the two of us we spend most of our time doing LIGO related research then you've got of course Alan Warren, Jolian Tevye who's not here at the moment and one grad student at the present time who's seriously working on LIGO-related stuff. So that's seven people in the group seriously working on LIGO stuff. But there's still a large other group as well, which you know about, Leonard Parker, John. They both have at least one student. Actually, I think in John's case, I think he has two. three other students are doing non-LIGO related stuff, plus as a post-doc I've been non-LIGO related. So, it's certainly majority of people are doing LIGO research, but it's not everybody. And is there a definite split in the sense that the people who are not primarily involved in LIGO stuff really don't have much connection to them? On a day-to-day basis, that's true, but not if you average over time, I think. I think the point is that there's as much interaction on a sort of a week-to-week basis as there is anywhere else. You still see a lot of conversations between Leonard and John and then the LIGO people and also the other postdocs and stuff. of interest. I mean, not so much with the students that work for Leonard and John, but Alpan Raval is definitely interested in LIGO-related research, although he hasn't really moved that direction and probably won't since he's going to become a computational biologist. But, yeah, yeah, so, so it's unlikely that, well, he still interacts a lot with us and asks us lots of questions and

50:00 trying to find out what's going on and stuff. But as I say, on a day-to-day basis at the moment, at the moment we're just, it's grunge work. It's just coding and very little real development, I shouldn't say not development, but research, let's say. Most of the research has been done over the last two or three years, and so now it's a matter of implementation for a couple of years before we then get to the stage of more research. on a daily basis, it's kind of hard to have exciting interactions with the LIGO group. Sure. They probably don't want to hear about the latest coding issues. Yeah. Pointers to pointers to structures of pointers or something. That seems to be the conversation over lunch, so it's pretty bad. So, the way you see it, for the next couple of years until LIGO is sort of up and running, that's just going to be concentrated on getting the software written that's required yeah I'd say I'm not sure how the group in general is going to work here but I would say that for the next 12 months my time is almost solely going to be devoted to LIGO research development let's say software development and stuff and I'd say that's going to be true for most people that are currently here in the LIGO When we sat down and actually put together our list of tasks that needed to be done, given that LIGO had now told us that we were the people writing the software and it was our turn to take action, we put together the list of tasks that needed to be done. And I must say, after doing that, I became very aware of how manpower-limited we are. So there's a real lack of manpower. We really don't have enough time to do what we need to do. so I think it's going to require everybody to be fully committed to writing code for about a year I would say, that's my guess I think actually after a year this time next year people will start to have a lot more fun, I think there'll be data around we'll be sort of learning a little bit more about the instruments about how the data analysis works we'll actually be able to start thinking about doing science which is something that has not been thought about as yet like you were saying yesterday in your talk you were talking about harry's discussion

52:30 with people at lsc and then looking for a theory to shoot down if they don't see any gravitational waves well it seems to me that there's there's the entire you know the three categories of source that we have there's got to be science we can do even if it is only playing and placing upper limits. And I think as yet, people don't fully know what sort of constraints and what theories they're going to constrain and things like this. And I actually think in 12 months time when people start thinking about that a little more seriously, it'll be interesting. We'll actually be able to think about astrophysics, figure out things that we might want to do and how we might use gravitational waves even without detecting them to constrain models or at least tell us that we've now encroached on the regime where we might have optimistically thought we'd see gravitational waves, and so now we're starting to constrain theories a little bit. That's my hope. Right. It'll be at that stage that people will sort of consider more actively the kind of atrophysical things. At the moment, there's just enough to be done getting things ready. Absolutely. We're still, I think the only research that's being done, as far as I can tell at the moment, is really still associated with issues of algorithm development for gravitational wave data analysis. There is really, there's only a small number of people who are probably putting significant effort into thinking about astrophysical stuff. And that, yeah, I mean, that would probably be Tom Prince, Kip's group a little bit, although Kip seems to have pulled back and be more interested in just issues of, well, you know, he's definitely interested in gravitational wave generation issues, so he's interested in figuring out if there are other models we can expect waves from. But I think really Tom and him are the two people I know of that are seriously thinking about the astrophysics stuff. I'm sure that'll change, though, because there's too much exciting potential. Even if we don't see gravitational waves in the first two or three years, there's still reasonable potential to do good experiments. When you mentioned the Big Man Parliament, is that more just a case of the number of people available to this group or a general thing for the whole LSE?

55:00 I think it is general to the LSC. I was just reading an email this morning from Sam, which is public, so I don't mind talking about it, in which he was listing his duties. And I'm going to probably pull it up and just quote it to you. It's on the ASSIS webpage, so if you ever want to track it down, you can find it there. Indeed, my hands are currently overfull with activities in detector characterization, LIGO 2, the LIGO TAMA analysis, the construction of the LDAS data conditioning API, and participation in the Collaboration Software Coordination Committee. And that's a list of Sam's tasks. So it's just, I think overall, there is a huge, there is definitely a shortage of manpower. And as well, it's been, I think it's somewhat difficult because it's such a big collaboration. There are a lot of people who are unfamiliar with large collaborations, including myself. and figuring out how exactly you should fit yourself into it and get to do something that's useful, it tends to be a little tricky. Once you're around for a year or so, you very quickly figure out there's lots to be done and lots of places you can help people. But just finding the group or the people that you should slot in with and get things done takes a little while, I think. But overall, I still think there is a manpower shortage. UWM has two main tasks, one being the binary and spiral hierarchical search code, which I think is well in hand. It's not too big a task because we already had a search code developed, and the new code really is just going to be a cleaner version of that. the biggest problem is going is on the other hand is continuous wave sources where we don't have any code and we have to we have a commitment to develop that as well and that there were far in far worse shape and people are less enthusiastic about continuous wave sources

57:30 um you know binary and spiral is the nice glamour nobel prize winning stuff cw sources for potentially small returns. And so I think it's harder to get people enthusiastic about that one. But it is interesting to me that things are shaping up and there's a lot of people beginning to figure out how to get involved with, say, the actual data analysis codes or the detector characterization codes. the group in Michigan for example is a good example of that they're very familiar with detector characterization so Keith Riles is leading that group but they and they also have I can never remember the guy's name he's one of the experimenters he's been at Caltech basically for the last 10 years Michigan you'd know his name if you heard it So they've got good experimental background. They've got very good experience with detector characterization type issues. But now they're also starting to move into the data, the sort of astrophysical source type modeling and detection issues. So I think that's a good example. Keith has been on the committees for about a year and a half or two now. He's been doing what he knew very well and now is beginning to move his group towards involvement in the data analysis from the direction of the sources rather than from the characterization. So I just think it's... I think we're in a position where we're manpower limited and people are finding it a little tricky to see exactly where they are and know exactly where they should go to contribute the most to the project. So that's kind of my feeling on it. But we're definitely manpower shorted here. If NSF wants to give us more money, it'd be real nice. It's a shame they're not going to hear that. Exactly. No, I mean, it's hard enough. Although here, I think, if I'm realistic about it, given what I've seen so far, I'm not necessarily a good person to judge this, but our biggest problem is we have a lack of graduate students coming through. And so it's hard to find people

1:00:00 that kind of would like to spend a year doing something that maybe will get them a solid or involvement in LIGO and we don't have that type of junior person to do a lot of the legwork for us. You know, and that's, that I think is going to, is one of the biggest problems we have here. We've got a great top heavy group. I mean, we've got brilliant, brilliant postdocs and brilliant, you know, people. So it's, it's fabulous from that point of view, which we only have one grad student out of seven people working on LIGO. So it's pretty much. Just a little, yeah. And is that just a kind of a problem at UWM that there are not that many graduate students in physics? I think so. UWM, as you know, it has a great reputation in gravity, but I think as a physics department, it's not so well known around the country and certainly would not be the first place that people would think of applying for grad school. That, I not to have a high enough profile because we're a small department. So this is really an issue of the size of the department rather than the gravity group itself. So hopefully over the next couple of years it will change, but we'll have to wait and see. So it's one of those things. Well, does it, does it, do the, I'd say Cardiff has plenty of, quite a few graduate students. Do the other groups who are involved in data analysis typically have more graduate students? Well, that's a difficult question to answer because I don't actually know the other groups involved in data analysis. So if I think about it from the point of view of AEI, which, of course, is not a graduate school, I think they do have some connections to Potsdam or Berlin or something like this, so they do have a couple of potential graduate students. But basically, they have none. it has a lot. Penn State currently, as far as I know, has no graduate students that are working directly on LIGO-related, well, directly on data analysis-related stuff. They do have an experimental gravity group, which has four graduate students. And then this UWM, which currently has one PhD candidate and one master's candidate, but that really only counts as one

1:02:30 person, really. So it's hard for me to tell what the general situation is. Caltech, of course, always has lots of graduate, well, not always, but at the present time still has lots of graduate students, but they're not doing data analysis work. And the groups that are doing development work are somewhat limited. Aina Flanagan's doing some data analysis development work. He has a number of graduate students, but it seems that they're not working on the data analysis code development. That's hard for me to tell exactly, but just a little bit of recent interaction with them suggests that that's true. And then Joe Romano is another person who's involved in it, but they don't have any graduate students there at the present time. So they do have a flock of undergraduates who are working on LIGO data analysis issues, which is a good thing. But again, I don't know how up to speed they are as yet since their group is very, very new. So all in all, it's hard to tell. I think some people in the gravity community are a little uncertain about what LIGO, how involvement in LIGO is going to pan out. And as a result, that may actually be somewhat stifling the involvement of graduate students. I don't really know, but it might be. Well, speaking of the general attitude of the community, when last we spoke in front of the recorder, you mentioned that you'd just been at a PCGM, this was a couple of years ago. Right. And there have been some talk, voice there, people expressing resentment at LIGO-related issues, and so on. Is that something that you've seen more of since far? Is it not really that much of a mission? You know, I haven't seen any more of it since because that was probably the only open forum that I've been around that I saw a large discussion of this. I'm not a member of the topical group on gravity and I haven't been to any of the APS meetings so I haven't encountered it in that type of context. It's hard to tell. If you're interested in sort of getting a flavor for what the community might think, a A good place to look is, this is very formal, of course, but a good place to look is the National Academy of Science Gravitational Physics thing that was published.

1:05:00 It's actually on the web, although you can order the book as well. I think it's pretty cheap. And this is the Hartle Report, essentially. They definitely lay emphasis on gravitational waves, but naturally enough, it's a decadal report in which they clearly don't want to just say, is the way to go, they have a reasonably balanced presentation in which they do suggest that gravitational waves should be emphasized, but not to the detriment of the other major things like quantum gravity or numerical simulations of gravitational collapse and other things. So it's, actually I think it was a fairly balanced report when I look at it, and reasonably enthusiastic about gravitational waves. despite the fact that the community may have been less pleased a year or so ago. I don't know. Well, that's a good suggestion. Actually, funny enough, it hasn't occurred to me because this is the end of the decade to look for that because recently I have been looking at the reports from the previous three decades going back to 1970 and actually this is something I might try introducing into a version of the talk that I gave yesterday one from the beginning of the 70s, is they're aware of Weber's results. And they say, well, we don't know what to make of this yet, so gravitational waves might come to be an issue. But then they actually, you know, just go on and say, but, you know, of course, on the other hand, we actually don't know if we have any reason to believe in the existence of gravitational waves other than, you know, the obvious analogy that we think they ought to be there. And it sort of has a couple of paragraphs about this sort of philosophical issue. So in a way, it really emphasizes, if you look at the each decade report, It really emphasizes the change. Obviously, different people write the reports, but... Do you know who was... I don't know who wrote that one. I guess I'll try to find out. Yeah, well, I think the committee, at least in this, I have a copy. It's at home, unfortunately. The most recent one does have a list of the people who were on the committees, although I don't know if it indicates who was chairing it. Yeah, they do have lists, but I don't know who did the writing. It looks like it certainly was one person. But nevertheless, it's, I think, probably a reasonably reliable historical guide to the changing attitudes and how quickly, in a number of decades, it's changed from that.

1:07:30 How about over the other, so then there were two more between the then and now? The 80s and the 90s, yeah. Yeah, and did they have much mention of gravitational waves? Yeah, increasingly, each one much more. So the 90s one, by that time, it was, you know, talking a lot about LIGO and so on. well, so the 70s one is saying, well, obviously this is going to come up more, surprisingly. And the 80s one was saying, well, you know, this could be a burgeoning field, which there's some looking at. This is now my recollection from a few months ago. And then the 90s one, you know, obviously this is going to be a big deal for this decade. Yeah. So it's definitely that. I mean, it's definitely flagged as one of the biggest deals, It's one of the main areas in gravitational physics, so no question about that. What I am familiar with talking to Santi is that in Britain, which of course where they've always been much more reluctant to put money into it than here, they've now decided that this should be one of their main areas of focus. Cool. So, in their equivalent of this type of rapport. So, obviously, there is a trend towards seeing it now as a big avenue. It's an interesting sociological trend. But considering we haven't had any major successes yet, it's interesting that there's so much money being invested around the world in the projects. That's something that seems very, I don't want to say surprising, but definitely interesting to me. It's an interesting point that people around the world are so keen on, which probably points to the potential that's there because people really believe we can do what they claim. So I think it's pretty cool from that point of view. But interestingly enough, your impression is that it's not like you're getting swamped with enthusiastic recruits who figure they'd better catch the wave. Nope. which I thoroughly don't understand because I was certainly very aware of it when I was a graduate student that if I wanted to do gravity well I should say coming to the end of my graduate career I was very aware that if I wanted to continue to do gravity that getting into gravitational waves in some aspect was pretty important so I am a little surprised we haven't seen a bit more enthusiasm for it. But again, it is probably, I don't know

1:10:00 how it is in high energy physics, and to some extent the model for involvement in LIGO is somewhat similar to the high energy physics model. And so it's from that point of view I don't know how kind of people who do, grad students who do high energy physics, how they feel about getting involved in it and things like this because there is a lot of startup and you're a real small fish in a real big pond initially and if if that and traditionally i guess that never didn't suit gravitational physicists right and so if if that if you kind of go with that attitude as as what people are coming to gravity with the attitude of maybe that's why it's just a personality issue, but I don't know. It'd be interesting to know, but I'm not sure. When you mention the difference in the collaborational style that entails LIGO-related work, how constricting is it on the way that you approach the research? I mean, obviously within a given group that's collaboration in any way that people are more used to, even if it might be, you know, more like four to five people instead of two, two or three or one in previous times. But especially I'm thinking of between coordinating of effort between different groups and so on. Does that mean that a lot, it happens pretty regularly that you have a way that you feel the problem ought to be tackled and they're just, you have to sort of decide to agree to do it another Are the tasks sufficiently divided up that you can sort of do your own thing still at the research room? Yeah, at the present time, the thing is, and this comes back to the manpower limitation, at the present time the task is sufficiently divided. Basically, as I mentioned already, UWM has commitments to two particular tasks. And realistically, although there was some interest expressed by other people, they haven't been banging down the door to help out. So there hasn't been an issue about trying to coordinate from that point of view. I presume it will change. We get together often enough and talk about the stuff often enough that people know roughly what each group is doing, which is good.

1:12:30 But I think eventually it's got to change. I don't know on what timeline, but people who are building the data analysis hardware out in Hanford or down in Livingston have got to start becoming aware of what's being developed in terms of software for actual data analysis purposes. And while they are to some extent, because LIGO has some internal people working on tools tools for data characterization for detector characterization and other things they're not i think quite as close to the development that's being done by say the astrophysical source identification and signatures group um and that i don't know i'm not sure if that if that's a bad or a good thing at this stage i don't really have a feel for it but it certainly hasn't been something that's caused us any problems yet and eventually it may but right now it hasn't and there's enough it's interesting there is enough to do that at the present time that people are not encroached on each other's territory to be honest i'm sure to some extent nsf is concerned about that within the united states because they have 200 whatever people working on the LIGO project from outside of LIGO itself. And so they have to be aware that they're not double funding certain development issues. But I think as of this time, that hasn't really been an issue. It may start to be now. Again, this year was the first time that... So the collaboration has been coming to understand itself a little better over the last year or so. And this time was the first time that the NSF had insisted that people proposing to do LIGO-related research gave a proposal to the Program Advisory Committee or gave a presentation to the Program Advisory Committee so that at least there would be some feedback from the PAC about what the various proposals seemed to be doing and how it fitted into the overall LIGO project. So I think basically we've been learning a lot over the last two years or so to figure out how the collaboration is going to work.

1:15:00 There's lots of formal documents being put together, but realistically for people from the gravity community who have no experience of this type of collaboration, those documents don't mean a lot. They're more frightening when you first see them than anything else. It's like, oh, my God, all this formal palaver about how we're going to collaborate with each other and how we're going to publish papers and how we're going to do research. And you realize it's necessary in one sense, but on the other hand, you get very scared of it because you're not sure exactly how you're going to fit in. But my impression looking on has been that over the last three or four years, we've seen everybody become slightly more comfortable with what it means to be involved with LIGO. And the laboratory has clarified their position on what it means for other people to be involved in LIGO. And this has really helped, I think. It certainly has made me feel more comfortable. I don't feel that I'm sort of... I feel like I can actually spend some time where I don't achieve what's normally regarded as the achievements you're like publications, et cetera, and still feel that that is serious scientific work that needs to be done. And before, when people were unclear about how they were tied to LIGO, it was harder, I think, for people to commit themselves fully to something that would basically cut off their other research during that period of time. So I think over the last few years, this has been really clarified, and most groups seem to be happy enough with it. LIGO's becoming a bit more sticky on certain aspects of what people say they're going to do and how they actually do it and things like this. So the MOUs have been a bit more formalized this year. But the basic thing, I think, has become more comfortable. But it is interesting. There's all kinds of interesting sociological things. fascinated by personally you know i mean there's all you know particular you know people that don't seem to quite get on with each other or have particular attitudes about how things should be done and they're very different than the attitudes that other people have and it's it's interesting to watch that develop i imagine i my suspicion is at the moment it's fairly open i think i think

1:17:30 anyone who happens to be around or be on the emailing lists which are essentially available to everybody at the present time um would get a sense of certain things that go on but but in the next i would say in the next year or so those types of issues are going to become less public i think ligo is going to tighten up its reign on who's allowed to attend meetings who's supposed to be involved in certain types of interactions, and as we get closer to data, that's going to be a little less open, I suspect. I don't know if that's...we'll see if it actually comes to pass, but that's kind of the feel I have from it at the present time. So on the one hand, LIGO has moved to reassure people that their contributions will be recognized and that their contributions clarify what way their contributions will help and so on. But on the other hand, they're taking a more active role in trying to mold what it is that people are doing. I think they're not actually taking that active role in terms of the molding. But they are taking a far more active role in assessing whether or not people are choosing to do the right thing, in a sense. They're just looking a little more closely at the MOUs and the types of research problems that people are saying they're going to do and how that relates to the aspect of the project they're involved in. And again, this is most serious, I think, for people involved in the LIGO-1 collaboration because it's a short timeline. The gravitational physicists are theorists. They're not building the instrument. They're not building parts of the data analysis hardware. And so the question is, you know, you're a member of the LIGO-1 collaboration. What is it? What's the producible you're giving us? or what is it that you're actually going to supply to LIGO that they're going to use. And that's what they're watching a little more now. So it's moved away from the gravitational physicists being able to say, well, we're thinking about sources and how we might detect this type of source or that type of source. Now it's, okay, where's the code? And LIGO has, from that point of view,

1:20:00 I don't think they've become so much more involved in handing out tasks, but they certainly have become more involved in asking people about the deliverable from that point of view. So they're more interested in having something concrete, as you say, in the form of code, and also presumably in having it by a certain time. Yeah, exactly. So when you commit to do something, they expect you to do it unless there's a major reason why it got held up, So, which we'll see how that holds up in the next round of MOUs, but that certainly is the way they're moving. And I guess, presumably, it's correct to say that the chief carrot then, in return for a slightly more activist role, is access to data? That, you know, that's still open for discussion. That is not clear as yet. And an interesting little sociological event, again, happened last week. There was an email sent around, indicated by one of the data analysis working group leaders, saying that there was now data coming from the interferometer. Actually, I don't know if it was the interferometer, but certainly it was data coming from Hanford. It may not have been interferometer data, but there was real data from some of the channels. And it was now available, and Jim Brow in Oregon had agreed to put it on data tapes for people, so if people wanted it, they should let them know, and this, that, and the other. And then yesterday, there was another email sent by Keith, which said, And sorry about that, guys. I went off half-cocked. The data analysis, the distribution of data has not been decided yet. That's going to be discussed at the next LSC meeting. I apologize to everybody after discussions with the laboratory and the LSC. I'm aware that I made a mistake and this kind of thing. So I was very impressed by how this went. I mean, he just sent out his email as he thought it was assessed. and once i guess he got a talking to i don't know i like i don't know exactly what happened he probably got a talking to and then he just but he just wrote a very straightforward email in which he just clarified the situation and apologized and that was it and i i thought it was very nice very clean you know no no issues no bad feelings no nothing you know so i thought

1:22:30 that was pretty cool but the data is going to be an issue and i would doubt if it'll be decided in March, even though that's the LSE meeting, I'd say it'll take another year before that gets decided. That's my guess, again. Just, things tend, my impression, as I was saying this to you yesterday in the corridor, my impression is that people with involvement in LICO, people have been concerned over the years about this or that or the other or whatever. And in general, the laboratory has been quite reasonable in how they've resolved issues or how they've, maybe not resolved, but how they've moved forward. So my expectation is that whatever's decided with the data will be reasonable. I don't think it's acting as the carrot at the present time. I think the real carrot is just to be involved in the collaboration and the publications. because it's just not clear how they're going to analyze the data, who's going to get it at their own home institutions, and who isn't, and things like this. And it's a tough one as well because, of course, the instruments are omnidirectional and sensitive to every single source. So when you send it off to a given institution, you can't say, okay, you're only allowed to look in the opposite direction for boson stars. You can't do that. I mean, people well, you can, but people in principle could do anything they wanted with it. And so it's going to be a little bit, I'd say there'll be fairly definite rules laid down about how people interact with the data. And before we see that, people won't know how much of the data they're going to get. So I don't know. One of the issues that one could see potentially arising there is thinking that wasn't at the GW Dalmady in Paris, but one of the things I'm told happened there was that Bernie Schutz gave a report on some of the Cardiff AI analysis of geoprototype data, and he was talking about removing line spikes, and Ray Weiss gave him a hard time that that theory shouldn't be involved in that kind of thing. And obviously, there's a natural issue regarding the split between having both theorists and, you know, experimenters potentially involved in data analysis when, okay, the theorists know all about the sources and so on, but it's the experimenters who know about the instrument.

1:25:00 And I suppose one could see the experimenters being unwilling to let the data out of their hands until they... Right. I think that's going to be one of the things, to some extent, us theorists are not familiar with how much needs to be done with data before it gets to the real analysis stage. And that's something that I'm very unclear on, and I think the experimenters have a much better feel for it. the the the incident you refer to actually was was even somewhat somewhat worse in a way um it was so what had actually happened was alicia sintez who was on your postdoc at the time actually still is i think had was giving a presentation about line removal and ray stood up and gave her an awful talking to about that this was not what the theorists should be doing So this was a real example of where, and Ray, of course, is brisk, as you know. So he came across quite brisk in his comments to her, and he really did sort of say, you know, I don't know why you goddamn theorists are wasting your time on this kind of thing. This is for the instrument builders. We should build a better instrument. He's sort of right. There are certain things, there are artifacts in the prototype interferometers which presumably won't be there in the full-scale ones. Until we see the full-scale data, it might be a little bit premature to be thinking about removing certain features. On the other hand, we know there's going to be wire resonances. We know there's going to be 60 Hz line spikes. No matter how hard the experimenters try, they're going to be there. So I think that was kind of like just a reaction to the fact that But his being familiar with different instruments, he's aware that the character of them changes quite a lot depending on what the exact construction of the instrument is. And so building techniques which were very specialized to particular instruments might be a bad thing, I think. But I may also be softening what he said because it didn't come across very soft. there is definitely

1:27:30 going to be an issue from that point of view I think the theorists involved in LIGO are going to have to become a little more at least engineering oriented if not experimental oriented because there was another occasion at last year's LSC meeting when Ray turned around and basically I wish I could remember the quote it may even be on the minutes on the web it was something like, if you think I'm going to service you goddamn theorists. And don't quote that as Ray Wise. But it was interesting. I think he was really just trying to make a point at that stage. This was not a definite statement. This was more half-joking, whole-in-earnest. Beware of our position as experimenters. It is going to be an issue. I'm not sure how it's going to raise its head in the end, but I think it will be an issue. From your point of view as a theorist, and one who's had experience at this point with actual data from an actual instrument, what level of experience do you think the theorists need to have? Obviously, it's necessary to have previous experience with some actual data. Presumably, binars and theorists aren't going to have actual experience with the detector. much sort of interaction with the experimenters or so on, do you think? Realistically, my feeling on it is that the optimal situation would be to have spent several months or spend several months a year at the sites where you, you know, okay, you may not be out, you may not be down in the laser room tweaking the laser or turning knobs somewhere else, but you might be in the control room or you might be crunching data in the back room or whatever I actually think that's going to be very important certainly if I could I would do it other constraints start to factor into those things I really do think that's an important thing to do but I don't know how many of the theorists are real how many of the more junior theorists are going to have a chance to do that before the instrument comes online I don't know And anyone who's basically, the field is fairly small still from the theoretical point of view.

1:30:00 We've got, I mean, here at UWM, we have probably half the postdocs in the field. And that's a slight exaggeration, but it's not far off. And most of those are going to be moving on to junior faculty positions in the next year or two. they're all going to have additional commitments to teaching, etc., and not have time to go to the sites. So all in all, I'm not sure how the theorists are going to, especially the junior ones, are going to get that experience out of the sites. It's a little unclear unless they spend their summers there, which is an option. But of course that might be also a little bit more difficult to get funding for. I'm not sure. My feeling on it is I certainly will be encouraging people around here, the postdocs that have a little more time to take a month or so each year and go out and try and talk to the guys who are working the instruments see how it's going, find out what's going on, help out but I don't know I don't know how everybody is going to be able to do that because of other commitments and I'm pretty sure that they won't want a deluge of several hundred people to have to go to the sites during the summer months you know so summer camp yeah exactly summer camp in Livingston Louisiana I'm not sure it'll I don't know how that one's going to pan out realistically but I I'm certainly going to be trying to get out to the sites and see things and try and I don't know hold the toolbox while the other guys do the work or something you know just just to get a bit a bit of an idea of what's going on how many troops have you made I've only been there for the LSE meetings and the inauguration so what's that one actually I think that's only two trips to the sites themselves no three I mean it's really is necessary to go and see the instrument everyone should do that in my opinion because it's just fabulous have you seen it? if you get a chance do go it's just again, especially as a theorist who isn't used to seeing big instruments for gravitational wave experiments this is fascinating to see

1:32:30 something that's three times the height of a man which holds the test masses and four kilometers of beam tubes stretching off over the horizon this is an impressive sight and well worth seeing but realistically really want to do is stand around the lab and help them build stuff so I don't know and I don't know how that's going to pan out but is your impression that you know people at the sites in the lab are pretty open to the idea yeah I think I think they're their point is if you if they if they feel like you're you're in any if they feel basically like you're meeting your what you to do to them they're completely satisfied um as a matter of fact probably even delighted and and they're perfectly willing to to help help out i was just trying to contact ligo about going there during the summer just after kid fest and whatever and i was i was actually surprised i mean very now you know okay so it doesn't cost them anything except an office but still i was still impressed they they were really quick to get back to me and said yep no problems just let us know contact so and so and we're looking forward to having you so they are keen to get people down there to get involved and have contact with the people at the laboratories and you know at the observatories and at the laboratory as well so i think i actually find i've been very impressed with the leigo management and coming down like from the very top and down everybody's been is really good I think about having people come visit and help out so personally I think it's really good yeah well as you said one can certainly imagine it from a sociological view that it will be really important to have theorists who are maybe looking at that and also see be there at the instrument and talk in the experiment oh I think I think it's going to be extremely important. Because we have a different way of doing things. The experimenters very much, we learned this from working with the 40-meter data. The experimenters, so we went away, we did all the analysis, we had answers. We said, right, here's what we've got, and

1:35:00 showed them the crunched numbers in graphs and plots and whatever and this of course wasn't all that interesting to them they were like okay well so explain this step oh so you throw away things that do this can i see all the things you throw away and this was while we had looked at these things we hadn't looked at all of them they literally wanted to see them all and and so this mentality their attitude was well if we look at the things that are not gravitational waves we'll learn something whereas we were going in going well we're looking for things that look like gravitational waves it's a very it's a different it is a different community I definitely I found it interesting to see the reaction to what we did and how the experimenters at what questions and what was important to them, which was very different than what was important to us. So I think it's going to be extremely important for us to interact closely with them, or else you're just going to find that they're going to tell us to go to hell, or vice versa. Well, actually, they can tell us to go to hell. We can't tell them to go to hell. I can't have a further access to the period. That's right, exactly. Exactly. So it was your impression that they found it an interesting exercise looking at the results of the data analysis and what bits that were thrown away or the bits that popped up. They were, as I say, they were more interested, I think, in the things that kind of, while they didn't look like gravitational wave signals, chirp gravitational wave signals, they were there in the instrument, and they knew about some of them. They'd listened to the data and could hear them happening and things like this. But they just wanted to see the structure of them and compare them to chirp waveforms and see what they looked like and just get a feel for how they were. something which, as I say, we definitely had some feel for this, but more based on a small sample. They were kind of like, you know, show me a bunch of these things. Let me look at the graphs of the data and see what these things really look like and what's going on in a couple of the other channels and things like this.

1:37:30 And so, yeah, that may also have been associated with the fact that it was a prototype instrument. So in a sense, they were kind of curious to learn things. think in general we'll find that the experimentalists at the sites and also in the laboratories are going to be interested to see the spurious things that come out of the instrument, at least for the first couple of years. I guess to some extent we all are, but the experimentalists have the most to learn from it, since they might actually be able to figure out what's going wrong or what's happening in the instrument. you mentioned them listening to the is that quite common it was at the 40 meter prototype they had a set of earphones set up and to get it back in lock they'd literally put on the earphones and start tuning it I presume at the sites they'll have something like that I don't, I'm not sure to really be sort of down at the instrument where they're doing things, but presumably they'll have an audio stream that if people want, they can listen to and see what's coming out. It would certainly make sense to me. I mean, it's very clear when you listen to the data that there's a lot of extra structure in there that you might not notice if you just saw the raw output. And so it seems to me that it's a good thing to do. I was talking to Rick Genay, who you probably know, Caltech. Maybe you didn't. He was Tom Prince's graduate student. He's a radio observer. He was involved with a few of the projects that Tom was doing with high-speed data taking and things like this. And he was down in Arecibo just after the upgrade there last year and was taking data and was looking at the output that he was getting and it was like this is just noise, I'm not getting anything useful and he was supposed to be trained exactly on a pulsar and it was only really a calibration run at this point and he said well so what he did was he plugged in the earphones piped it through sound put it on and sure enough he could hear as the pulsar was ringing off in the background. So I think that's a very good example of where the audio

1:40:00 stream is important. You look at the raw output, there's nothing there, but you put it to the ear, and the ear can hear the structure. So I'd be surprised if they don't have some facility for that up at the sites, but we'll see when we get there. They definitely have all the other high-tech gadgets. So when you guys get data, though, you get it on a tape, and do you spend much time listening to it? I know I've heard something myself. I guess, actually, I haven't spent all that much time listening to it, to be honest. and at first yeah it was very exciting at first so when we got data four years ago or whatever and the 40 meter data was sort of more available and easily accessible it was kind of cool to listen to stretches of it but we certainly didn't listen to the whole 20 hours so it's it's one of those things where i think it could drive you mad after a while but But I still, I do suspect that it will be used as a diagnostic tool. And even so, I could very well imagine that if we were doing data analysis, we'd probably had an event or something, that one of the first things we might do is listen to a few of the channels to try and identify things that are going on and just see if there's anything similar. I mean, there will be automated correlation detection methods and stuff like this, but still. to the channels to see if anything might happen that would indicate that what you're seeing is noise. Exactly. And I do think that'll happen. And then that would explain why experimenters are going to be more likely to listen because what you're going to hear when you listen is, say, noise, especially now in Gaussian noise. But in this field, unlike in the case of the radio astronomy, you're not going to hear a signal, of course. So that would explain why the theorists aren't going to be listening to. That's right. Although, again, I think it will be a nice diagnostic tool as a first cut for even the theorists. I think you'll find we will be pulling it out and comparing it to the 40-meter output and trying to figure out what the different little noises are,

1:42:30 things that sound like scrapers or whatever, trying to figure them out. So, yeah, I think everyone will be doing that initially. Eventually, who knows, but initially I'd say it will be done quite commonly. Do you find that you can identify sources of noise from a really limited experience? The actual sources, no. I have very little knowledge. But if you listen to another channel that's not the data channel and hear something funny there, that can tell you, okay, that's something. That would be the way I would imagine that you could do it if you had the ability to sort of quickly scan through several channels, have them up on screen, and just say, yeah, play that one, play that one, play that one, listen to them, see if there's similarities at the time. I could imagine that that would work, but as far as figuring out what's actually happening, the instrumentalists had some idea when we were doing the 40-meter data analysis. They sort of listen to it, and they go, yeah, you know, that's probably a cable click. This is detaching, and just detaching and reattaching. It's a dirty connection or something. Or it's an off-axis mode in the interferometer, this, that, and the other. But it was a little hard to, for the theorists, it was impossible, I think, just to listen to the data and say, oh, that's probably just not knowing the instrument. There's just no way you could say anything. That's noise. That was about as much as you could say. Interesting. So an example of the channels that you've listened to would be the size monitors and... I'll bet it'll be actually the microphone channels. We really only listen to the gravitational wave channel with a 40 meter. Bruce may have listened to some of the others, but realistically that we only listen to the gravitational wave. I'd say seismometers, yeah. Magnetometers, probably. But I can imagine microphones and things like that might actually be quite used as well, the actual audio channels that are going to be recorded, just to know if there was any unusual activity going on around the test masses at the time that certain pieces of data were taken and things like that. because I think that's I think that is going to be

1:45:00 well actually it's hard I don't know, there's so many channels going to be taken I don't think anyone really knows what's going to be the important ones and what isn't The microphones are just in the building next to the chambers or something like that I'm not sure how many are there but they have a bunch of them around the building around the end masses center stations they may even have a couple in the beam tube know, I actually don't know if that's the fact. That's interesting. I don't talk about anything confidential. Yeah. Exactly. I'm not sure exactly how sensitive these microphones are going to be, but you'd be a little concerned. Big Brother is watching you. It's interesting. This, of course, is the example of the kind of thing I was too dumb to ask about when I was at Faltech. I mean, they occasionally give us a little tour that it never occurred to me to ask about whether they listen to the data. So apparently they actually would listen while getting in lock. Yeah, yeah. It was, as I understand it, because they were, well, as I understand it, they're still a little uncertain about how the interferometers, sort of what the theory of the locking of the interferometers is. And so back then when they were very unfamiliar with it, that the only way they could really tell that things were going well and that their adjustments were basically helping was to listen and wait and see when things smooth out and they knew they were in luck and could go from there. That's my understanding anyway. Interesting. Because it's an interferometer, I sort of naively imagine people looking at fringes, but instead you're listening for fringes. Yep, yep. It's pretty amazing, I think. But that apparently was how they did it, yeah. Well, I mean, from the sociological point of view, this is an example of the, you know, interesting ways in which science is obviously a craft. And so the sociologists call it tacit knowledge. presumably an experimenter by spending a lot of time listening to something like that will have a possibility of figuring out what it is and then obviously as you say you know the theorist might be able to say well there's something there but he's not going to be able to say more than that and that's obviously exactly where you would expect that it's only by the theorist going down to the place and seeing what goes on that he's ever going to find out these things because probably nobody's going to think to tell him otherwise because it's just so natural.

1:47:30 I think that's absolutely true. The experimenters just sort of have a feel for how things behave. And I can't imagine, like, I'm sure we all have certain things that we're so familiar with that it's hard to imagine trying to communicate to somebody else exactly how to perform the task or figure something out. It's just one of those things that you know, you learn, but there's no way to teach. So I think that's exactly what this is. Without playing with the instrument, you're just not going to know what's going on. Yeah. Well, that's really fascinating. But let me go to something really dull, which I thought I should ask just to get an idea in my own head. So I guess when you want to do research for the OSC, write code for LIGO, you have to typically reach a memorandum of understanding with the lab. You're going to also, as I understand it now, have to write a proposal to the OSC, and in In addition, you're presumably going to write a proposal to the NSF for money to pay for the people who are going to help you do the work. So that's quite a lot of proposal writing. Yeah. What's the kind of, is there kind of a thing that you typically do first? Yeah, write your grant proposal to the NSF. That's kind of my feeling, but I don't know. So this is a new game, I think, for almost everybody. The MOUs are fairly well set up, and I wasn't involved in direct negotiations since Bruce already had negotiated an MOU. So they're pretty well set up, and just each time, based on our experience with the different groups we're involved in, we say we're going to undertake this task. group and come up as something that's high priority and needs to be done or whatever

1:50:00 is prioritized in a given way and so you say you're going to spend so much time on these things. So as far as the MOUs go, that happens very naturally. That just comes through out of requirements that come up within the different subgroup meetings. The proposal writing end of things, I have no idea how that one's going to shape up. the first proposal as I'm sure you heard from Alan it was submitted about six months ago it has been with the LSE and a variety of different people within LIGO and finally was approved yesterday I think for it to be worked on and there's just In a sense, the formal statement of what this proposal writing process within the LSC is supposed to be is very straightforward. And the statement that was made, it's not in writing, but the statement that was made in words about it was that it was supposed to be a one-page proposal. So you literally give a title, you know, it's like, here's a title, here's half a page on what we want to do, here's the manpower requirements for the project and here's the breakdown of those requirements in terms of LIGO staff and LSC scientists also there's a few other little things that have to be laid out within it which are actually quite important it seems to me part of the process I think is very important it's kind of a scary one because in a sense it's not this is going for for people who are not involved in big projects all the time it's very unclear how this is going to allow you to do your own research or figure out problems because you know how it is within research you're doing a problem you're not always sure exactly what the question is until you're sort of halfway through and really it's changed quite a bit from what you asked initially yeah so i'm not sure how that's going to how that's going to work in this case in a sense Their proposals almost sound like you've got to be already at that point. But the important thing that's in them is the statement about LIGO manpower requirements for the projects and also protection of graduate students.

1:52:30 And that, I think, is probably the most significant thing in the written statement about the proposals. It's the fact that the idea here, one of the major ideas here, is to make sure that you protect the graduate student's research projects. If they're heading them towards PhDs, you protect it from the collaboration on a bigger scale where people could just go, that's a really interesting idea, yeah, we should work on that, let's get it done. so I think from that point of view that this thing is actually very important and I also my feeling on it at the moment is this is not a process that's going to be a big deal for the next 12 months or so moving into 2001 and coming towards the end of it I think we're going to see a lot of people starting to kind of assemble teams potential projects this is wild speculation it's two years away who knows what will happen but I think that's what we'll see it's just because it is the type of thing by yourself you're not going to figure out what the interesting questions are it's going to be kind of groups someone asks a question it's going to be people, teams to get together and figure out what are the real issues that need to be addressed and how to go about it and I'd say on that timeline going in. Between now and then, I'd be surprised if we see more than maybe two or three. But once the real data starts to come along, I'd say there will be at least five or six a year, is my guess. At least. I mean, probably even more than that. So, but, yeah, so that one, I don't think there's, the overhead is actually not that great now. They only require as to have reports and statements about research every six months. So that's reasonable. And, well, we'll see a little bit more about how streamlined things like the proposal process within the LSC becomes, which is one of the major concerns of the moment. The biggest problem, I think, the biggest thing I see with all of these things, I've become somewhat more philosophical about them over the last four years,

1:55:00 as I've already mentioned, is that when you first see them, they can seem unreasonable. But when you sit down and think about them for a while and there's enough open discussion of them, the unreasonable aspects generally get, they're usually pretty small and they get thrown away pretty quickly. The biggest problem is too much administrative overhead. And with these proposals, that's going to be the biggest concern, I suspect, put in a proposal on how long is it going to take to get it reviewed, how many people are going to be involved, how often are you going to have to convene one of these committees to do this type of process, and there's a bunch of things like that. But I think that that will become more and more streamlined as time goes by, and people will become more familiar with how it goes. But the most interesting piece from the sociological point of view on that is not going to be the proposal writing and stuff, but it's going to be the publication issue. I mean, I think that's going to be the most significant question over the next few years. I don't know when the debate will begin on that one in earnest, but I do have a feeling, as I say, there's been indications already that the LSC is going to be closing off its meetings a little more. I'm sure there will still be people that will be invited or can ask to attend and things like this. But as data becomes available, I think there's a big concern that the meetings should be purely LSC people or people that have a certain responsibility not to talk about what went on because of the data being there and maybe having gravitational waves or maybe not having are gravitational waves in them and things like this. But I think that will mean that data analysis subgroups in particular become a little more private in terms of their affairs. That's a guess again. It's just based on indications I sort of have heard but I don't know when it will happen. I do think that primarily would be because it sort of makes it more clear a demarcation between who's in and who's out to avoid gray areas I think that's exactly, I mean, they've already said that the ASIS subgroup is not supposed to have people attending the meetings that are not members of the LSC.

1:57:30 It's a reasonable statement. I mean, the point is that after a while, this subgroup is going to be responsible for certain tasks associated with the data. And the bottom line is, if you're in and contributing, you deserve to be there. If you're not in and just hanging around, the question is, why exactly are you hanging around? And so I'm not sure how that one's really going to pan out or what it's going to develop into. But for our, for ASIS, it certainly has been a bit of a concern because there's a broad community of people who've contributed to understanding sources. pretty beneficial to draw on that community, but on the other hand, if they don't have the interest to actually become directly involved in LIGO, then you have to ask whether or not they should be at these meetings or just drawn upon in other open meetings or something like this. But I'm not sure exactly on what level this type of thing is being discussed or if it is being discussed at the present time. It's just there was an indicator that we should I see members at these museums. Well, obviously, the issue of the publication and who will be on them, it must be regarded as one of the figures to address because I guess there already have been efforts to address the policy laid down. The policy is laid out and it has been approved by the council, so that is in existence and I presume would need some act of Congress or other thing to change. it's there and it's very clear, what types of publications and what the author offerings are and things like this so from that point of view that stuff is already there but still I think when it comes down to really writing papers the policy document is probably the most minor part of the task so we'll see what really happens and I gather that I gather that the trend of the policy that's been made down so far is to be, in a broad sense, inclusive to avoid problems. Yeah, it is largely.

2:00:00 I haven't read it closely for a couple of years now, but the publication document was fairly much inclusive, which was a cause for concern for some people because it was hard to know what exactly was delineated as something that was a direct LIGO piece of research I guess the real emphasis was to make sure that anybody within the collaboration who wanted to contribute and had something to contribute could be involved in the publication and I think that's perfectly reasonable and that actually is the same emphasis on this proposal policy People should put in proposals and let other people within the LSC know about it. And if other people want to join, find it an interesting topic and want to join in, they can say, look, I think I can help you with this. And if they do, they're supposed to, within reason, they're supposed to be included in the project. But as I say, at the present time, first of all, it's not going to be that many proposals, I think. But second of all, people are so busy with other things that it really doesn't matter. So I don't know. Going back to this business of the different people you've got to make proposals to, in the sense of the NSF and the LSC and LIGO itself, So do they, are they all pretty much independent or, I mean, for instance, with the NSF, before they actually give you any money, will they want to know that you've got the MOU, for instance? Yeah, so there's one thing that's for certain, and that is that the NSF is, in terms of its actual standard grants and LIGO-related funds, They are insisting that people have, well, they'll definitely want people to have an MOU with LIGO if it's real LIGO research rather than something that's peripherally connected. They have had this policy or program advisory committee meeting. The NSF had all of the proposers present at that so that they could get definite feedback

2:02:30 from people who were essentially guiding the project. So I think that's going to... I'm pretty sure... This year was the first year that was done. I'm pretty sure that's going to stay in place. I would not see that changing at all. And it's fairly sensible. These things are not... Provided you're within the working groups and listening to what's going on, these things I don't think are going to make a big effect on whether or not your proposal gets funded because you should be proposing... proposing to do tasks that need to be done and so from that point of view even though they seem like a whole series of new steps that are in between or in the process I think the bottom line is that if you're actively involved with with the collaboration and going to the meetings and going to the different subgroup meetings that you're involved in that you just you quickly learn what are the what are the open tasks that need people working on them and so I would be be issues of how they get prioritized for funding, but I'd be very surprised if these particular extra steps really cause significant problems. That would be mine. So the real research direction, in some sense, are still decided at the meetings of people involved in the OSCE? Yeah, that has certainly been true, Rakesh. Yes, you can come in, but I... Oh, I was asking if, you know, it was still within the LSE groups that the research directors were really being decided, and that was the important issue. Yeah, I really think it is. You know, like, this is where I think the LIGO management have been superb in terms of realizing things. I mean, they realize their skills. They know what their skill set is and where they are. They've been brilliant at the engineering tasks. they've got very good scientists working for the project as well but the trouble is that they're overburdened with getting things sort of up and running so they've been very good at realizing that the LSC is an important well, I mean I think the management has a great deal of experience with this type of thing where a collaboration develops with a fairly wide variety of people in it

2:05:00 and they've been very quick and very good to realize that So the sea is really one of those places where the research gets discussed, bashed around. People figure out what needs to be done and get in there and do it. They have to be a bit careful that what's being done really does mean something to them in the end, and they are very much aware of that. But still, the experience so far, although it was like, so the ASIS group was formed in 1998 in March. So this is basically coming up on its second year. And it has very quickly evolved and adapted to how the management of the laboratory has sort of seen the role of the subgroup and has quick, you know, on six-month timelines, it seems to me, we've changed the approach, we've changed the attitude to what sort of pieces of the research we need to do, how we need to emphasize things, how we need to move ahead. And it's meant that, in a sense, a lot of times we try and put together a plan, we get started on it, and the plan has to get updated fairly soon afterwards. But still, all it has meant is that the plan has become a little more precise, I would say, rather than anything else. It hasn't curtailed the research in any way. So from the experience with the analysis, which it is hard to be involved with more than one of the subgroups because they tend to meet roughly at the same times, LSE meetings, and so while you can be involved in the teleconferences at different times, it's a little difficult to be very well tied to each of them. So my experience with ASIS, I found that we've been very much self-directing in terms of what we regard as the important research and what needs to be done. And as soon as it's become clear that this is a task that's been laid out to us by the laboratory or by the collaboration, we've been fairly quick to figure out how we should get involved in it, at least as quick as we've been able to be, which, you know, when there's large amounts of policy involved, it takes a while to figure out exactly how you fit into it, but the groups are, as far as I can see, pretty much self-directed, and that really is where the decisions on a lot of the research issues get made. Again, that might change once the scientists within LIGO

2:07:30 become a little freer and pass the building stage where they've got the instrument in place and start to really play with it. They may start to indicate different directions that need to be taken, you know, and so we'll see. Yeah, that's interesting. I was going to ask, this is something that I've heard and I don't know, I'll try them later. Obviously with the whole, given the whole question of data analysis, besides all of the interactions with the experimenters in LIGO that we've been mostly discussing, there is the question of the other bits of the theoretical community who are still involved in coming up, more directly involved in coming up with actual tempus. and how much interaction is there in that case and also is there any remaining sense on stat analysis people that not enough is being done for any reason I would say at this point interaction is somewhat more limited than it was, say, five years ago when there was this extreme involvement and a very tight group of people. That's not... Coalescing binaries are clearly still the one case where we have a clean system where calculations can be done and in some sense probably still need to be done a little bit. But for detection purposes, the statement has always been that we're pretty much there already. So we have most of the details for non-spinning binaries at this stage. The question does then after that, you wonder whether or not that's true, if the binaries are spinning or processing or anything. And it's just not a not known, I think, at that level. it's kind of hard it's a little hard for me to judge that one I think Alan might have a better feel for it but certainly the emphasis on waveform calculations in the post-Newtonian sense has certainly died down and a lot more emphasis has moved towards issues in numerical relativity issues associated with the finer points

2:10:00 of gravitational radiation reaction or things like this But I think for most, I'm just trying to think, is there any other place where we really would like waveform calculations? There really isn't. I mean, every other possible source, it seems that the darn things are so complicated, there's so much complicated physics in it, that there's no way you're going to parameterize the models by a small number of parameters. And so to go looking for very precise waveforms that somebody happens to give you, it seems a little unrealistic. It seems very unlikely that those waveforms are going to be close enough to make your detection. So I think in general we've moved away from worrying about waveforms themselves in pretty much every respect except for the binary black hole case, which is one that we'd like to see solved but as far as I can tell progress is both difficult and slow I certainly would not be laying bets with Kip about theorists calculating waveforms using numerical relativity before they're detected Kip is likely to win that bet I would say I would say the last grand challenge meaning the numericists didn't look terribly happy about being obliged because I'm professionally obliged to take this pen right I know it's one of those unfortunate situations I'm not sure it's hard to tell from the outside why what's holding things up I mean it's clearly a difficult problem there's no question about that but it's somewhat hard to tell from the outside whether or not it's just purely computer power or manpower or ideas or all of the above I always get the impression it's all of the above in the end but given the great hopes that they expressed six years ago or whatever it was when they started Grand Challenge it was it's kind of a little disappointing that we don't have at least a better feel for what's going on and why we're not able to do what we can do or what we want to do I think there's one I think there's one obvious sociological reason which is that that's one reason that sticks out and which is that unlike in the data analysis case where as we discussed there's more collaboration needed but the problem is better understood and can be divided up neatly into packages which are then far and down and people can work rather than be independent but in the case of the finer black hole problem there was clearly the need for collaboration to get the efforts of lots of people

2:12:30 but since it wasn't so clearly understood how to solve the problem it didn't prove possible to just divide it up neatly between the different people so you have a whole different order of sociology involved than actually trying to get everybody to work together yeah I think that's right and it's one of those actually I'm sure that's a fascinating sociological study I'm not sure they'd want you to study that one but I'm sure it is a fascinating one yeah that's my impression the collaboration was hard to form and again, in that particular case it was a bunch of people that were not familiar with a big collaboration without anybody to really help them figure out how to form it and with a nil-defined mechanism to do it so, yeah, I don't know but all in all, I think we still stay fairly closely in contact with the theorists who are doing waveform calculations. But there's a certain level of belief that we have enough in some ways, and also there's even been some skepticism about certain, because these are, as you know yourself, these are getting down to the nitty-gritty level that you're still, you're kind of like, this is awfully precise considering you're basing it on post-Newtonian or some variant thereof. And so I think there's a little bit of skepticism about exactly what information is being provided, say, with binary and spirals, when you step to the next level. And in particular, that would not be, for example, to stick with post-Newtonian and just do the next post-Newtonian calculation, but with the Pade approximants. It's a little unclear exactly how much they're capturing the essence of what's going on of the Inspiral, and certainly it's leading to, well, I wouldn't say open debate, but

2:15:00 I think that there's a certain amount of, I don't even know what to call it, people are just wary, people, like I don't think anyone's going to go out and swear blind that either way is absolutely right and gets the right essence of the last ten cycles, but On the other hand, they're also a little bit, I don't know, there's just a little bit of scepticism about certain claims that are being made, I think. What do you mean about whether, say, patty approximates or whatever? Revolutionize the way we do things. I really don't know. I mean, I don't have a feel for it. I just, and I've seen both Thibaut, who's very involved with that, and also Satya, give good talks on it. But even at the end of those talks, I still come away a little uneasy with the reasons they're telling me that these things must be better. and it's just so actually to be honest it's more an unease than any sort of statement of being incorrect or dumb or anything I certainly don't feel that I just feel a certain amount of unease or maybe lack of understanding on my part about what exactly these things do and how they are so much better than what we have somebody who doesn't know much about it might help if you go and read numerical recipes and what they say is well okay it seems to do some great things, but of course you don't know for sure what exactly it is doing unless you already know what you're approximating to and get the answer. And that's exactly my concern with them. It's not obvious to me that they're converging to the right answer. I mean, I guess it's pretty likely that they are since they can show it's convergent but on the other hand I'm still a little uneasy about it so but again it's not important it seems that it's unlikely to be important even based on what Satya and Thibault and Bala have been saying it seems unlikely to be important for detection purposes so I'm less concerned I would you know I'd be worried about it if it was a detection issue but even they tell

2:17:30 that it shouldn't really be any important for detection. It really is an issue for parameter extraction and trying to understand what the nature of the systems are. So you feel that the current just templates from third post-intelium approximations without any fancy extrapolation or further approximation would probably suffice for detection? They seem to be good enough from what I've read, and I haven't actually checked this myself. good enough. Especially for the lower mass stuff. The higher mass stuff, all bets are off. Even Pade is like, it's not you really couldn't lay a bet on it because the last few cycles are lying right in the middle of the band. My guess in that respect actually is that pretty much anything is as good as anything else because there's going to be enough degeneracy across the things that you'll just make the detection by matching, sort of, on some other waveform you've got in your template banking. Because there are relatively few cycles, it's not going to be so critical there. You don't gain so much by matching. Exactly. That's kind of my feeling. My feeling is, as long as you get where the loudest pieces are, since you've cut off at low frequencies, you should be doing okay. I guess you'll do that, but it could be way wrong as well. You know, when we see the waveform, we'll know. So your feeling is, for instance, that the chances of detection are likely to be crippled by, say, favor to have full binary black hole? Well, actually, in that respect, I don't know. I think the final wave form is the final event of merger for two black holes. Neutron starts, it seems to me, is far less uncertain, but two black holes is extremely uncertain. What exactly happens if you've got too much angular momentum in the system when it hits the unstable orbit? Does something sort of weird happen that it hangs up in some sort of weird rugby-shaped ball, shaped horizon or something? i really don't i don't have a good feel for what's going to happen there and i think actually to be really honest even a single simulation for a kind of a situation which would should hang up

2:20:00 or should should produce a bit of something that would be a black hole too much angular momentum to be occur it'd be nice to know what happens during that just after it becomes dynamically So from that point of view, if you push into that regime of the black holes, I think that our current state of knowledge is likely to be... it's certainly going to damage our chances. If there are many cycles, we're in real trouble. If for some reason the darn thing decides to hang up in this peculiar configuration. I mean, it's hard to imagine how it does it, but if it does decide to hang up in this peculiar configuration, we could have a lot of waves coming off and give it a very special frequency structure, and if we don't know about it, we could be in real trouble. So I'm not sure that one, but as far as the lower mass objects go, I think we're in fine shape. I mean, they don't merge until the frequencies sway high, so we're well covered exactly, we're well covered yeah I guess the only scare in that respect having been Matthews in there that one seems to have resolved itself at this point I don't think yeah I think that one's pretty much resolved itself Yeah, as far as it goes with the waveforms, the one other thing that I think is kind of important, and I don't know how much, I mean, we haven't given it much thought yet. We're always interested, but, for example, Lars Bilson's mechanism that he had a few years ago for low-mass X-ray binaries emitting gravitational waves, anything of that is always exciting to the people in LIGO. Any news that might actually give us sources is a good thing.

2:22:30 So from that point of view, we do like to keep our ears to the ground with anyone who's doing gravitational wave source modeling. But unfortunately, I think at the present time, we're not seeing, there hasn't been a real, since the R-modes and the LMXBs, there hasn't been a real burst of excitement about a new source that we had not considered previously. and I guess it's unlikely that we're going to see it until we actually detect something that we haven't figured out it's kind of hard to imagine at least I'm not as smart as Dyson so I can't imagine these random things that might produce gravitational waves so it is difficult as you said it's always possible that something actually happens that will spark off I mean, you know, it would be fascinating to actually... Because actually in the case of Dyson, what, he was just part, it would be interesting to try, maybe he was just more willing to throw these ideas into print, you know. Right. Because that, I think, was also part of a general outburst of new ideas sparked by the discovery of quasars, and the idea that the gravitational class was somehow involved in And that really just led to a flow of new ideas at that time. But up to that point, nobody had ever given a matter of seconds thought about it. Right. And I think that's going to be true here as well. I think, you know, if we ever do see something that we didn't plan to detect, and we can get any knowledge about its structure, like, you know, what sort of frequency it's at, cycles it has, et cetera, et cetera, I think that's going to lead to a flood of different possible things. And I guess as well, realistically, the one thing that we have at the present time, which is the big question, is the binary black hole case. And if we did see something that could possibly be binary black holes, after all, I mean, binary black holes will be the most likely source if we see something that's a big burst. I think that might actually help in terms of trying to do the modeling of it. I find is that the thing that surprises me a little bit is that the numerical relativists don't always choose to solve problems that are astrophysical.

2:25:00 Now, clearly with the binary black hole case, they really are. But you, and there are two philosophies in numerical communities as well. One is that you should just get the equations, figure out how to make them work, put your boundary conditions on and see what you get. You should know something about the physical scenario you're trying to model and adapt your method to that physical scenario. There are dangers with the second one. There's dangers with the first one as well. You won't be able to do it. But the second one there are dangers with, clearly, because you might put in what you get out. But I think that in order to solve the problem of binary black holes, there has to be a lot more physical input. I think that's what's really holding up. Well, it's one of the things that's holding up advancements. in the field just because if you don't put a lot more thought into the physical questions you want to get answers about you end up just basically simulating something that doesn't necessarily give you the information you're looking for that is somewhat unfair said just that way, I mean because these guys clearly are physicists and actually do know about the physical scenarios at least some of them do but in recent years it seems to me a lot of the co-development for numerical relativity has been in a sense untargeted that's the tension that's always been there in general the physics approach and the formal math approach exactly, I guess that's true that's been around for a long long time yeah, it's not surprising it's still there I guess the only new thing is what you guys are doing where you actually have to listen to the experimentalists but this is not really relativity at the present time it's more C code development so I don't know it is actually I must say for me it's a very nice aspect of this particular have to think about a bunch of different physical stuff, or even mathematical, I mean, in terms of data analysis issues, you've seen I forget if you've actually worked directly on data analysis issues, but you've seen people talking about it and stuff

2:27:30 it's just it's a different field, you know, it's nice to be kind of trying to apply or learn stuff about a new field and figure out how it applies to gravity or whatever, but which is really cool and something that something that's a little difficult to do when you're just doing pure general relativity or stuff like that. I guess I can claim to have actually been one of the first people to come up with a problem. Well, it would probably be a reasonably good template, except that it's for something that nobody's going to see. Which one? Oh, right. Oh, that's right. That's right. That's the... I forget. Is that more detectable by Lisa? Yeah. It is. Okay. But it's still pretty ragged edge or well yeah I looked at Lisa a bit and the funny story was that when I was working on it I started thinking how about it more detectable by a space-based detector and so at group meeting I said to Kip about this and he said oh yeah I think he obviously must have completely misunderstood what I was talking about anyway but whatever it was he just said oh, yeah, I wouldn't bother to admit that, you know. So, stupidly, I didn't. Oh, God. And about a month or two later, I was busy writing the paper, or, you know, writing the paper. He suddenly rings me up all excited and says, what you really ought to look at is space-based detectors. Good idea, kid. Why didn't I think that? So, when you're young, you're stupid. You're getting wiser most of the time, anyway. But, anyway, so I did eventually look at it. Thanks.