 Let me welcome you to the December 2017 CNI member meeting. I hope that people did not have too much trouble getting here. We did have a little bit of snow over the weekend. It was kind of picturesque, but I don't think it amounted to very much here. And I'm really glad you've made it. I'd like to extend a special welcome to our international guests. Some of you have come a very long way, and I know how difficult that can be. And I'd also like to extend a welcome to our new members. We actually have quite a number of new members, and I'll just read them out. The University of Illinois Springfield. I have a lot of Athenaeum, thank you. For some reason I have terrible trouble with that word. Athenaeum 21. The University of Northern Iowa. Northern Arizona University. Dominion University. Villanova University, the Fall v. Memorial Library there. Binghamton University has rejoined us. We have the Ontario Council of University Libraries, an important consortium. Smith College, and I'm very pleased finally to announce that Dora Space has become a new member. Please recognize and welcome our new members. You will find in the packet of stuff you picked up our new 2017-2018 program plan. And that if it's not on the web already will be on the web imminently. And I hope that you enjoy looking at that. I have a number of things to say about that in the course of my remarks. Finally in the meeting packet I just want to note that planning on, planning for the joint CNI JISC meeting to be held in Oxford in July is well advanced and we will be announcing speakers for that quite shortly. But there is a hold the date note in your packet that you picked up. A couple of session notes. First off, a very small number we think of the agenda programs, the physical program, have got the wrong tabs in them. They've got all the right pages, they seem to have been a dozen or so that have the wrong tabs. If you have a bad one traded in for a good one, but otherwise you've got SCED, if you're using that on the web, and that's fine, and it should all work out. We don't think there were very many of those, but I'll just alert you. We have a message board out there, and we will announce any agenda changes or cancellations. I do want to note that we know we have two cancellations. Joe Lambert from JISC was unable to be here. The session Tuesday at 10.15 on the JISC shared a research data service, had to be canceled. And Ken Klingenstein's session, scheduled for today at 3.45 on identity federations, also had to be canceled. I am very hopeful we will be able to bring both of those sessions to you probably in the spring. And I think that that's everything I want to say about program and logistics, other than to note that we have been tinkering with the time and the duration of some of the sessions, as I indicated in the roadmap. So the goal being to allow as many sessions as possible, and to also be more responsive to the fact that some sessions don't need a lot of times, other sessions need a good deal of time. With that, let me turn to talking through what's going on in the world, some of the things I have my eye on, and some of the components of our program plan. And I'm going to have to go very fast here, because unfortunately there are an awful lot of things going on, and there's some things I need to talk about that I really hoped I'd never have to talk about again. So basically I think we are dealing right now with a tremendously uncertain world, and this really challenges us at many levels. If you think about the shifting policy environment for a minute, I think that we've seen some really bad things, and we've seen some responses to it, and some other leadership that we should be very proud of. The entire data refuge movement, which was really cobbled together by a small number of leaders and a big grassroots effort on very, very short notice, did an enormous amount of good. It's unfortunate that we need this, but it's a great tribute that we got it in place. Some of the research funders still don't seem to be coping very well with the need to support infrastructure, and that that infrastructure needs to include adequate infrastructure to manage research data, and that this needs to be a more serious partnership between the research funders and the universities themselves. I think that sadly but importantly, we have been reminded that the federal government can be an unreliable steward, and I want to be clear what I'm saying here. We have a tremendous number of very dedicated people and good stewards scattered through the federal agencies, and we have some organizations like NARA, like the Library of Congress, like the National Library of Medicine, that are absolutely first class, dedicated to what they do and doing it very well, albeit generally with inadequate resources. But we've been reminded of two things, that all of this is unfortunately at the mercy of politics and funding, and that indeed both memory and science are becoming increasingly politicized in various ways. And I think that that reminds us of the importance of resilience and diversity of institutional participation as we think about how we're going to manage our scholarly and our cultural record. We really need to be careful wherever possible to minimize single points of failure of all sorts, technical, financial, political. I think that this is an important reminder and what happened with the data refuge movement should be thought about beyond just that movement and really taken to heart as we think about the overall structure of the stewardship provisions that we are evolving. Finally, on the policy front, I never quite thought it would come to this, but it appears there's a very good possibility that they are going to do away with network neutrality. And one of the things I have been puzzling about is what that means to our community. And it's really complicated. I mean it's obvious to me at least that this is going to open the consumer market to untold bad things. I mean the opportunities for mischief are endless. But in terms of the research and education community, our institutions mostly obtain their network service wholesale and are much less subject to this kind of problem. However, so many of the resources that we depend upon are out there on the other end of networks and so many of our community work not just from our campuses and labs and things like that, but also from home or from afar in various ways. There is plenty of risk to go around. Very difficult to quantify, but something to be very mindful of. And I think one of the things that we may see is that this will make it important to make many of our key content sources hosted much closer to networks that we can control end to end. And of course the bigger content sources, much like our institutions themselves by their bandwidth wholesale, some of the smaller ones are probably much more at risk of problems here. So unfortunately this is a whole new area that we need to be very mindful of. And it's one that we can no longer entirely disentangle from access to content and scholarly resources, because we have taken that move to the network very deeply into the patterns of what we do. We also face a lot of problems right now because of an overall distrust of education, of scholarship and science, of journalism. We face an avalanche of propaganda, unverified facts, all kinds of material. And I'm afraid this is going to get much worse before it gets better. Obviously we face major problems in how to even preserve this environment so that we can analyze it for a later study. And I'll have a little more to say about that later. But one of the things that I see coming very imminently now is a lot of manufactured stuff up till now is words. And there is still a sort of seeing is believing, hearing is believing mentality at work in the world. And I think that that is very rapidly going out the window. If you've seen some of the systems that have been generated, been built lately, so pick your favorite public figure. Basically anybody where there's a good deal of audio and video around. And give this thing a script and it will manufacture that person saying those words quite convincingly. One can only imagine the mischief that's going to cause. Here's another interesting development. And speaking as a computer scientist, it's brilliant, but it has some very scary consequences. How many of you know about something called generative adversarial networks? Okay, I'll just tell you briefly the germ of the idea and it will become obvious what's going on. So you're all familiar with systems that use machine learning to build classifiers to recognize certain people or objects in photos and things of that nature. And that stuff has come a very long way. There is also work on how to detect fake images and things like that by looking for various inconsistencies in them. And there's a whole little branch of forensic science about how do you tell whether an image is real or not. But it's very specialized, a small world. Well, someone came up with the brilliant idea a couple of years ago where you take two machine learning systems that kind of work. One that tries to recognize things and the other that tries to generate fake instances of the thing. And you let them talk to each other and run. So you have a system that generates fake images and also has an enormous store of real images. And it hands the other system that wants to recognize fake images, examples of each and then tells it after each one, did I get it right or did I get it wrong? And the machine learning thing grinds away. And you know what happens? Both systems get better. They learn from each other and both of them improve steadily. And after you let this go for a while with big enough training sets, you have one system that gets pretty good at recognizing things and another one that gets pretty good at generating them. Whether they're fakes or what have you. See where this is going? Sounds a little like an arms race, doesn't it? So we can look forward to lots of interesting applications of technology like that coming in the near future. I think that one of the things that this does, one of the issues it creates is it reminds us that in an age of perfect replication on one side and increasingly good generation of things that never happened on the other, trails of provenance become hugely important. I mean the only way we'll have any authenticity anymore is in a very real sense in conjunction with provenance. We have pervasive personalization now in everything. Experiences of all kinds are customized and tuned to individuals. And I think that this creates an enormous set of challenges as we seek to preserve the cultural and indeed even the scholarly record. Particularly to the extent that we want to capture not just the static of what was presented but the dynamic of how information is spread, who knew what when, how widely was something known. And I'll have a little more to say about that in a few minutes. I think that some of these developments also, at least for me, reinforce that open access is not just important for scholarship. It's important for society. Basically, access to quality vetted information very broadly is, I think, increasingly an imperative for society as a whole, not narrowly for the research and education community. I also see a lot of interest and a lot of progress towards increasingly replicable and reproducible scholarship in the cases where that makes sense. And I think one qualification that can't be said too frequently is that it doesn't make any sense to demand that all science and all scholarship be reproducible. There are endless numbers of exploratory kinds of things that basically seek to develop hypotheses or insight that future hopefully reproducible research can then advance or put to rest. There are vast amounts of scholarship that are matters of interpretation rather than, you know, deterministic outcome. And so I think we need to recognize that replicable scholarship is an important goal, but it's also not a meat ax to simply be swung around to declare large amounts of important work without value. Finally, as I look at this unstable world out there, this broader world, I think that we are reminded very strongly of the need for public outreach and communication by scholars and scientists and by educators. And I think that we should be able to make a significant contribution to facilitating that just as we all here make significant contributions to scholarly communication. This is going to become an increasingly important part of the scholarly communication agenda, I believe, in these days. Let me just say a couple of words about open access and where that stands. And I constrain these remarks to the United States. I am struck by how different the open access picture looks from nation to nation now. There is no question that we have made significant progress in opening up the scholarly record. It is also unquestionable that a lot of it still hasn't been opened up. I do think that it is a very appropriate time for our institutions to take a deep breath to step back and to recalibrate or reaffirm their goals, expectations, and the level of importance and commitment they assign to open access goals. I see sort of a set of decision points coming up and some of these decision points surround funding and where funding is going to be expended and committed. Some of them are around policy sorts of matters. We see very large discussions happening in other countries. Think about what is going on in Germany or the proposals that come out of the Max Planck Institute and CERN. Or the proposal that some of our colleagues here have been advancing and I believe we will be discussing as part of this meeting, the so-called 2.5% commitment. I think that this is a time where institutions really need to be clear about what they are trying to do and about what they hope to get from it and what they really are realistically prepared to get from it. I don't think we have been appropriately critical in our thinking about that and I think some of this reappraisal has started in the last couple of years as people have said, well, we have coupled institutional repositories in open access and it doesn't seem to be working very well. Maybe we better think about that connection and we also ought to think about, well, why do we have an institutional repository and what do we hope to achieve by it? We had some fascinating discussions of that at last fall's meeting. I think it continues to be an important discussion but it's also important to take the open access question to the next level. I'm not going to give you the answer to that but I think that institutions are going to need to figure out one by one where they think the answers lie to those questions. Let me turn to something that I'm increasingly seeing as a really important programmatic component for us to be paying attention to. I believe that for the good of both scholarship and society, we have got to address what I would characterize as a crisis in the preservation and stewardship of evidence. Now I want to be clear what I'm talking about here. I am not talking about the scholarly record. I think that we are doing much better than we were doing 10 years ago on the scholarly record. There is still plenty to do but we at least are starting to put institutions in place, practices in place and I think we've established a very healthy trajectory there. Although in particular research data and ancillary research objects remain exceedingly challenging. What I'm talking about here is the much more ephemeral in some ways and much more volatile in many ways, broader cultural record which becomes important evidence for scholars in the future. It's not material that's produced primarily for scholarly purposes but it becomes the object of study by an immense range of disciplines. It's really hard to know what to preserve here, what the right objectives are. One of the core problems is you're essentially making largely speculative judgments about what you think is likely to be important in future. It's at best an orc. Now we have somebody of what I'd characterize as the broad evidence base where we already have some good reason to believe it's important and that's because it's been cited in the scholarly record. If you look at references in scholarly work, many of those aren't to other scholarly works. They're to pieces of evidence and in fact we see a number of systems and practices showing up that basically call for the systematic archiving of referenced components of the cultural record that appear in the scholarly work. Here in the scholarly record there's a very interesting initiative called Permacy C which had its roots in dealing with Linkrod in the legal literature which as it happens makes a lot of cross-referencing to the broader popular literature of various sorts, the popular cultural record. It's now spreading out into other disciplines. We have a series of archiving on demand kind of things where scholars archive references as they cite them. And we have systems like Momento that facilitate this sort of referencing and versioning. I think that extending these processes and strengthening them is going to become an increasingly important thing. This is the one part of the evidence base that's a fairly sure bet in terms of if we looked at, if we wove it into the scholarly record once, we're very likely going to end up wanting to have another look at it at some point in the future. But we need to go far beyond this and I think we have huge problems in how we set priorities. Indeed, one of the things I've been looking at a bit and I wish I could tell you there's great news about it is how are we doing? Do we have any idea how we're doing? In actual fact, it turns out, and again I'm limiting my remarks here to the United States, our measurements of rates of production of various kinds of components of the cultural record are pitiful as in often just one step shy of non-existent. If we can't even figure out how much is being produced, we definitely are going to have trouble with, yeah, and how much of it are we saving? Or even where are the hotspots that particularly merit effort? I mentioned the whole question of machine learning, personalization and those sorts of developments earlier. This last year I've been thinking a lot about what does this mean for our ability to preserve the cultural record? And a number of other people have been thinking about aspects of this too. David Rosenthal has done a wonderful series of articles about it called the Amnesic Society, which I commend to you on his blog. I recently summarized a lot of my thinking over the past year in an article that appeared in first Monday this month. And briefly, you know, one of the things that I came to recognize is that today's stewardship of the cultural record is about much more than the sort of preservation or archiving of objects, be they physical or digital. It's about fundamentally almost like documenting performances. And obviously we can only do this on some sort of a sampled basis, given that there are an almost infinite number of personalized performances happening on the web every day. But I think we can't just say, oh, it's too hard, we can't do anything about this. Let's just stick with the digital and physical objects that we're comfortable with. And one of the things that says to me is that the enterprise of stewardship of the cultural record broadly is going to need to weave in a lot of practices from ethnography, from journalism, from documentary filmmaking, and videography, and similar sorts of things. And this is a whole relatively unconsidered set of issues, I think, but it's going to be crucial for dealing with the news. I can't leave the subject of archiving without talking about two developments. One of them was just a genuinely wonderful thing that happened last week. And many of you will probably be hearing about it for the first time, because the public announcements just came today. Last week, I was contacted by some folks with a heads up saying that a number of organizations committed to preservation had prepared a joint statement, a declaration of values, and that they wanted to announce it today here. And several breakouts will, among other things, include some aspects of this. And they just wanted to give me a heads up and an opportunity to read it in advance of its publication. I can't take any credit for this other than being fortunate to be able to recognize, support, and publicize a very, very good thing. But I immediately said, would you mind terribly if CNI signed on to this as well? And if I actually took a moment to direct everybody's attention to it at the plenary. There are already links from a number of the partners to this document. I'm not going to read all the partners, but we will have a link to it on our web page sometime tonight. So basically this is up as a Google Doc at present, and it will remain open until March for comment, at which point it will get finalized. So, I commend that to you. The other thing that I just wanted to say that's showing up to me as an issue under many guises now, as part of the broad conversation about digital preservation and digital resilience, is the way in which this is colliding with the move to the cloud and with storage services available in the cloud. We are going to need to sort out how much we trust the cloud and how much we want local copies of things under our control. And right now there are very interesting arguments happening around the place at institutions where people are saying from the IT side, we want to do a cloud strategy and just put all our storage in the cloud and rely on guarantees from the cloud vendors. And the stewardship community is saying but, but, but, but, we need to do fixity checks which don't work very well in the cloud on a regular basis, and we want a copy under our control if something happens. We don't want it all centralized in one place. And I think that navigating and understanding that challenge and explaining the position and sometimes good reasoning for it to the IT community is going to be a very significant discussion over the next year or two. Okay, so I've talked about policy uncertainty, uncertainty of the stewardship of the evidence, and I've talked about the shifting government policy environment. I'll leave you finally with a discussion of some of the technical uncertainty that I'm tracking. And I've put that into a couple of buckets. There's one bucket that is about scale up, about whether we can successfully move from prototypes and things to a genuine sort of social adoption and infrastructural integration of things. And in that area, I'm looking very hard at annotation. We have built some very good annotation technology, although it's still a little fragile. Look at the work that Hypothesis has done, look at the work that's coming out of the World Wide Web Consortium. And I believe there will be an update session on the Web Consortium's work, the W3C work at this session, at this meeting. But we have all these unresolved kind of uncomfortable questions. Who gets to see what annotations? Who gets to annotate? Where are the annotations going to be stored? Who's going to run the annotation servers? Are the people who make content available really comfortable with it being annotated? I know a lot of authors have a very visceral negative reaction to this. Should they be told too bad, you know, live with it? Should their wishes be respected? This is all the social and adoption contract. Then the other one that I think is potentially very important and very unpredictable is containerization as a means of preserving and sharing software. This is one of these things that is starting to work pretty well technically, but we have a whole lot of problems about standard configurations, how many versions they're going to be, proliferation of containers. There's a whole practice side of this that we don't understand. And I'll also note there's a whole software licensing side of this that we also don't understand very well. In terms of media and things, it's getting really hard to tell media from technology. I want to note three things. One is three-dimensional stuff. Now, something very striking has happened here in the last couple of years, which is that we have managed now to get the full life cycle from capture to reproduction of a very significant range of 3D objects with reproduction occurring at unreasonable quality and at reasonable cost in many cases. And I think that this is going to have a huge impact on education as people actually can handle things, a physical copy of it as opposed to looking at pictures and trying to imagine what that must be like. We have a whole library apparatus that needs to be put in place here. Libraries with 3D objects, good standards for storing them and describing them, good provenance to tell us where they really came from and whether they're real or designed on a computer. I think that's going to be a very significant and important challenge to step up to. Right now, I would say that the reproduction of existing objects that have been 3D imaged is probably farther along than the computer-generated ones, which tend to be much more sensitive to the generation environment and to certain classes of stuff like architectural models or their own planet. And CAD-CAM is a whole other planet and a lot of these are very stove-piped. Coming up behind that, we have augmented reality. That's certainly getting a lot of hype and a lot of attention. It's getting attention, though, importantly from beyond the academy as a way of annotating places, architectural things, allowing annotation as you're experiencing things. The tourism industry is very interested in it. Again, we need to figure out standards for storing this, for describing it, for taking the associated geospatial reference frames that go with it and also storing those. I think there's plenty of work to do there. Always looming behind it is various kinds of VR objects, which depending on who you ask are always just a couple years away. It's clear there's a trajectory where the devices are getting a bit cheaper and a bit better for experiencing this. But when they really cross that critical uptake line beyond super-expensive, immersive caves and things of that nature, I don't know. There's a lot of apparatus that I think is rife with various forms of uncertainty. I'll just mention a few of them that I'm watching very carefully. Linked data and how linked data genuinely is going to work at scale for cultural memory institutions is really unclear. There have been pilot projects. There are lots of problems that nobody's really talking much about. There is digital biography, which is kind of related to this, and the convergence of biography and bibliography, especially in the contexts that include chris, they include encyclopedias, they include documentary editing. Snack is a hugely important development, and if you've not followed what's going on there, you really should because I think that it holds the promise of really massively changing the way scholars discover and use archival and special collections material. And then there's Momento, which I mentioned before, which I suspect Herbert may have a couple of things to say about tomorrow since he was one of the pioneers of it. That's both a technology and an infrastructure. You need Momento-capable servers. And one of the big questions, again, is we have the core technology. How much infrastructure is going to be out there? And that's an adoption rate question. Then, just to round out my pack of jokers, I have a few basic technology things I'm looking at. One is quantum computing, which it's incredibly hard to figure out how fast that's going to move, although there is some evidence it's moving a lot faster than people, let's say, ten years ago would have predicted. It's important to understand the first stake there. What quantum computing is going to do is invalidate a lot of cryptographic infrastructure, which includes authentication, various kinds of code signing and public key signing things. And I guess the short story is we are not real agile about switching out large-scale cryptographic infrastructure. So we could suddenly, as developments happen there, find ourselves in a very unstable situation. The last thing I'll just put in the deck of jokers is blockchain technology. Now, I'm sure you've all heard much tedious speculation about bitcoin and related cryptocurrencies. The thing that may not be clear to many of you is that blockchain is actually a set of technologies, which, although it has some very real limitations still, is much broader than cryptocurrencies and has a lot of potential applications for attestation and attribution in a kind of a verifiable public way. And exactly how that plays out and the extent to which that may supplant or supplement a number of current practices I think is very important to keep an eye on. So just to conclude, I think we're dealing with massive uncertainty in almost every sphere at this point. And in looking at this, I really come away with just a few perhaps trite observations. The first is that this is a time of all times to be keeping a very close watch on developments and be prepared to respond quickly. That in trying to navigate these developments, we need to have a pretty firm grasp on our values and goals and a pretty realistic set of expectations. And the last is that effective collaborations are going to be especially important now. I feel like resources are getting spread very thin and things like this digital preservation declaration of shared values and the thinking behind it is so important because it basically says that we're making a commitment to work together and to try and maximize resources by coordinating rather than squandering them on redundant and uncoordinated efforts. And I think that there's a terrifically important theme there that I hope CNI can advance and contribute to and that's going to be very important for navigating these uncertainties in the coming years. And that's what I wanted to say about how I'm viewing what's going on right now and some of the developments, a number of the things that we're watching carefully and we'll try our best to keep you informed about developments on and a few programmatic initiatives that I think are particularly significant, many of which as I say have said before, really address not just today and tomorrow morning but a few years out and I hope are going to be very valuable to you in strategic planning. Thank you very much and I've actually finished in enough time that if I can see questions with these lights in my eyes I would be delighted to field a question or three. Thank you. We've got mics in both aisles, all three aisles actually and you can also just yell and I'll repeat questions and I'm sure we've got some. What? Nobody's going to take the vape? Deal. Who gets to decide what? Well, you know, ultimately, provenance is no better than the initial metadata that's assigned to a thing. At best after that you can construct chains of custody and you're going to decide those based on which organizations you trust. You know, you will believe some people are, some organizations are trustworthy stewards and others aren't and that's typically the way this works. I think one of our real challenges is how to deal with that, you know, kind of initial creation event, how you attach the metadata to it that you can trust or quickly attach some kind of attestations to things that help to generate some trust in it and I think that's something we need to be thinking about very carefully. You know, when you consider, for example, how heavily people are relying on crowdsourced photographic evidence of events now, I'm starting to feel like there's a desperate need to somehow be able to weave in some, you know, hard to defeat way to increase confidence that what you've got is something that really is what it claims to be and I don't think there are easy answers to that but I think these are hard problems we have to be thinking about. Ultimately, though, a lot of it is going to be subjective trust in various people and institutions. I wish I had a easier answer for that. Yeah. You spoke at the beginning about the likely end of net neutrality and can you say more about the implications of that for those of us who are relying ever more on cloud computing? Can you make a link there or give us some questions we should be asking ourselves? So, in terms of cloud computing specifically and in particular some of the big cloud platforms, in a certain sense, I don't see this as being a big issue because your typical institution is doing cloud computing on Amazon or Google or IBM or Azure from Microsoft or something like that. If you look at the pads there, you have one contract between the R&E institution and a major ISP typically. Then you have another contract between the cloud service provider and a major ISP directly. So those two contracts can prevent funny business with traffic prioritization. Typically those two primary ISPs are directly peered or there is a very well trusted third-party peering point in the middle. So the opportunities for problems there are fairly limited. On the other hand, if you look at content resources that we might license or rely on, those are scattered all over the place with potentially complicated and largely unknown paths between the consumer of the material and the provider of the material. So I would worry much more about that than your sort of more routine cloud computing. Even if you're trying to do cloud computing from home, in most cases you're not moving that much data back and forth to the cloud computation and storage platform. The stuff's already up there and you're just doing control, which doesn't take a lot of bandwidth typically. We have another question there. I'll get over there next. Thank you for that overview. Can everybody hear me? So I noticed that a lot of the trends that you're identifying and the future directions including things like blockchain technology as well as the adversarial algorithm kind of conversation, a lot of those are very computationally intensive and I wonder if you could speak to kind of where you see environmental impact playing into C&I and your sense of new directions for the following year. I have to say that we don't spend a lot of time right now on these environmental issues. There are some organizations that have started looking at this carefully, particularly with regard to cloud computing and there are some folks that are really interested in how to minimize the environmental footprint in various ways, by various metrics of those facilities. I know that some of the campuses that have invested in high-end research computing facilities have been very thoughtful about this. For example, there is a fabulous newish facility at Princeton that I had an opportunity to tour about a year or so ago that has won a number of awards for green computing. Also, the commercial players are often particularly sensitive to power consumption and for both economic and environmental reasons are siting centers, very near sources of inexpensive sources of renewable energy, solar, hydro, things like that. But that stuff tends to be much closer to the engineering of high-performance computing than we tend to get. This is an area with quite a bit of specialized expertise that needs brought to bear on it. I hope that helps a little bit. Yeah. You spoke about cloud storage versus local storage, and I'm very curious. You spoke about it, and are you comfortable expressing your either personal or professional opinion on that matter? Yes. I am in a very general way, and if there's one thing that we've learned over and over and over again, it is that things fail. They fail for a lot of reasons, especially the ones that never fail. They fail for economic reasons. They fail because of human error. They fail because the technology doesn't work. They fail for political reasons sometimes. Natural disasters, all kinds of stuff. And my bias is that for things where it's important that those things survive, a multiplicity of copies under stewardship that is as independent as possible with regard to all of those factors I just named, which, by the way, also implies a certain amount of international redundancy of material is the path for most likely success in being able to keep things alive. The other thing I'd say is that I think we often confuse reliability with resilience. There is a very large amount of material that I believe it is important to have survive, but if it goes offline for a little bit and it takes us a week to recover it, we'll be okay. It's much more expensive to increase sort of operational resilience than it is to increase resilience if you're willing to eat some availability problems on rare occasions. Thank you. We could probably squeeze in just one more question if there is one. Oh. Okay. We'll squeeze in both and then I will be terse. Yes. You're up. Ed Van Gemmer from Wisconsin. Thank you for your presentation. So you didn't mention anything about issues around privacy. And I suppose we could spend days talking about that, you know, the university risk involved in the privacy of various records and about individuals. Yet it seems like every time I call my pharmacy or clinic or whatever, they have more data about me and no more about me than I know about me. And so I wonder what insights, if any, looking forward with regards to the management of issues around privacy, I guess. Okay. So I'll just say three brief things about this because I could go on for an hour about this at the drop of a hat. The commercial stuff is ghastly and getting worse every day. And not only are they collecting data like mad and squirreling it away, whether it makes sense or not, but they're not doing a very good job of protecting it. And that is a bigger problem than we as a community can take on. That is a society-level problem. And some of it is a regulatory problem that we seem to have no will to deal with in this country. In terms of things where we do have some control, I think that I am happy to say that after a, you know, intoxication with analytics of various kinds at the institutional level, not the library level, which included a great deal of fairly thoughtless application of these things. We are now seeing a second round of investment in this that is characterized by some very thoughtful policymaking. I invite you to have a look at the talk that Jen Springer did, I believe, last meeting from Berkeley about their work on what principles that should guide institutional policy and the way that's been brought forward and codified by the IMS consortium. In terms of libraries specifically, I did a fairly deep study of issues around reader privacy, which I finally summarized in a publication that came out in, I don't know, February, I think. It's quite a long publication. I think that we have a problem in this area and I think that I would like to see and again, this is something that's in our control largely, I would like to see a lot more attention paid to those kind of issues, vis-à-vis contracts with content suppliers at our institutions. That's the very, very short version. Last question. Hi, Keith Webster, Carnegie Mellon. Not a question, but hopefully I'm not breaching an embargo by congratulating you, Cliff. Today, the ACM announced their new fellows for this year. You were one of the people honored. Fewer than 1% of ACM members and I thought we should congratulate you. What he says is true, but I did not put him up to that. I promise. The announcement came out today. I was really flattered. It's kind of you to say that, although I'm a little embarrassed. With that, I think I should send us on our way. We are five minutes late. Thank you so much for coming. I hope this was helpful.