 get started. Welcome everybody. I'm Cliff Lynch. I'm the Director of the Coalition for Networked Information. And you've joined us today for the summary session for week three of the fall CNI 2020 virtual member meeting. Just to refresh your memory, week three was focused on standards, technology and infrastructure primarily. And it consisted both of real-time project briefings and a number of pre-recorded project briefings. Recordings for all of the live sessions should now be available for you. The purpose of this session, which we've held for week one and week two already, is really just for me to reflect a little bit on some of the themes I heard emerging out of the presentations and to hear from you about themes you heard emerging from the presentations or other reflections, thoughts or questions that you'd like to share. As is the case in the preceding two sessions, while I've allocated an hour for this, I really probably only have at most 20 minutes of comments. And after that, it's really up to the attendees how long they want to continue and how many thoughts and questions and reflections they have to share. For this session, I think that rather than trying to do a moderated Q&A or something, we'll just use the chat, although if you want to use the Q&A tool, you can use that too. And also, if you want to make comments by voice, just raise your hand. And I can turn your microphone on or Diane can and we'll go that way. So we will stay fairly informal about it. So let me dive in and let me stress going into this that it's not my intention here to summarize every session. This is not a replacement for attending the sessions that you're interested in. There are some sessions I'm not going to mention at all because they don't fit into any of the particular narratives or observations that I wanted to make. And that's in no way intended as a slight of those sessions. So please bear that in mind as we go through. So one of the first things that I noted that was interesting is that we're starting to talk more about digital preservation as a service or more accurately maybe as a bundle of different services that can be applied across a landscape in a relatively repository agnostic way and that really invites us to think much less about preservation as something that happens in a big monolithic repository somewhere and really something that's considerably more, I don't know how to say it, agile that can fold in services that are not necessarily services of your own making but you can build on services that other parts of the community make. We saw one presentation on that called PrezQT out of Notre Dame and we also saw a really exciting update from Yale about their emulation as a service. And interestingly enough, we saw at least the beginnings of a bridge between these in the sense that one of the things that PrezQT can do is actually invoke emulation as a service. I think there's a lot more to be done along these lines and I think that it really is perhaps a very helpful reminder of the importance of thinking architecturally about this. We see this same kind of thinking, interestingly, in a couple of the other presentations. There must be something in the water in Notre Dame or something because I was also looking at a pre-recorded presentation that they shared with us about microservices for operating across a whole pile of boutique repositories representing archives, museums, and special collections, which again had this kind of architectural character to it that I found quite provocative. It was not surprising but certainly worth noting that issues around how to codify and make real the fair principles pervaded a number of sessions. Herbert von Dessampel and Martin Klein gave a nice example of how those interact with ideas that they've been developing in recent years around signposting. Again, in another example of the composition of services, PrezQT allows data curators to connect up with a fair evaluation service that's been available on the network. Another noteworthy thing in a completely different direction was the continuing quest for appropriate platforms and strategies to do what I would characterize broadly speaking as controlled digital lending. Basically, the need to be able to loan documents in digital format to circulate material in the current environment where it's impractical or impossible to get to the physical material. We're seeing a lot of different variations on that. We had a very nice report from one of our institutions on that, which unfortunately I just want to note was not recorded. I'm sorry if you missed this presentation by Brown University, but they asked us not to record it so that they could talk a little bit more frankly about some of the design decisions they made. I'm sure if you're interested that the presenters would be happy to talk with you individually about that, though. It's worth noting that they drew significantly on some similar work that was done at Princeton. I'm not sure whether we've had a formal presentation on the Princeton work, although certainly I've heard about it in some detail. Also related to the pandemic, we saw a very interesting presentation that for me raised quite a few questions. This was from Clemson, and it was about tracking occupancy in various kinds of spaces and managing occupancy. This ranged all the way from ensuring that capacity constraints around density could be met to running what are essentially reservation systems for seats. I'll just share two reflections from this, and I'm aware of a number of other campuses who are working with this same kind of technology, although each campus has its own variations. The first is that I really wonder how much of this is going to continue after we come out of this pandemic. In order to avoid queuing, to avoid crowding, a tremendous number of activities have moved to appointments, to reservation systems essentially, and those systems are very attractive in some ways to patrons, whether they're patrons at the local Motor Vehicles Bureau or whether they're patrons trying to use a special collection or whatever, study space. I do wonder a lot, now that we've made the investment in these systems, how many of them we're going to retain as it no longer becomes essential to use them for public health reasons. You can ask the same kind of questions about a number of contactless systems or practices that are being put in place elsewhere in the society, everything from proximity swipes on credit cards all the way through contactless check-in in hotels. The other thing, though, that was very striking to me about the system, and I don't really intend this as a critique so much as an observation is that as institutions have had to move very fast to get these kinds of systems in place, they are creating a Pandora's box of privacy problems. These systems collect a lot of data and they've been put in place often with very little thinking about how do we anonymize the data, when do we anonymize it, how long do we keep it, who gets to look at it. I don't know what the answers to these are, but I think that if anything, they have really accelerated the urgency of grappling with a lot of issues around data governance and privacy in our institutions by raising the stakes substantially. Another very different project that I want to really make sure gets underscored is the work that the University of California at San Diego reported on about reconstructing the Temple of Bell, a destroyed temple that was part of the broader destruction of Palmyra in Syria. They used some very interesting and sophisticated techniques to move from photographs that had been taken of the temple before it was destroyed at various points during its evolution to reconstructing a three-dimensional model. This is actually, to be fair, part of a much broader set of efforts to reconstruct a lot of buildings and indeed even cities, archaeological sites that have been destroyed in places like the Middle East primarily. We're getting actually pretty good at this. We're getting quite good at documenting physical structures which which we can we can be served very well by that documentation if something happens. You may recall, for example, some years ago we had a very nice presentation, a keynote, at CNI about work that was ongoing to use laser technology to produce extremely detailed architectural models of French churches. I was put in mind of that work much more recently when they had that enormous fire in Notre Dame in France at the cathedral there. Now I understand they're using those very detailed kinds of architectural documentation to help with the reconstruction. I think we're going to be facing a lot of this in coming years. One of the things that was very evident when you looked at what UC San Diego was doing was that in the short term this was pretty human labor intensive, but that at the same time this was exactly the kind of work that looks extremely amenable to machine learning in the relatively near term. So I think there's a lot to think about here. Another aspect of this is essentially crowd sourcing reconstruction by using pictures that are part of an archival record that are in special collections and amassing a body of evidence to reconstruct with. A third dimension is how this helps to support a case for archives that are digitized and that can be readily used computationally. So I think there's a lot to think about carefully here. I would just note also that there are some substantial lines of work that have been blazed in this area over the last decade or more. I'm thinking back for example to a system that I believe came out of Microsoft research called Photosynth where you could get a whole lot of pictures that different people had taken of a scene from different points of view and stitched them together. That kind of technology again is potentially very relevant here. Another theme that we've been hearing about as this conference has progressed has been about how we take things that are very in person and move them virtual and in future how we maybe should think about putting the virtual and the real back together in new ways. One of the interesting examples of that this week was a presentation from Maggie Milo of the University of North Carolina at Chapel Hill who's looking at maker spaces and is looking at how as we reopen maker spaces we might make them more welcoming. She's going to build what's essentially a VR maker space in order to experiment in this area as she moves on from her initial data collection. I think that's a really interesting project to watch and we've already seen several other interesting examples over the course of the conference about how we take activities that we used to do in person, split them between being in person and being remote so that you only need minimal people there, how you project these spaces into virtual spaces. I think we'll see more of that in the coming weeks. I think it's a very useful discussion to continue thinking about. It will help inform not just how we operate in the virtual environment but perhaps how we can operate better in the mixed physical and virtual environment to come. Those are a few of the observations that I wanted to share. As I said, I am making no effort to systematically cover every session or every pre-recording video. That would take much more time than we have here and you've got all of the source evidence available except for the Brown presentation. But rather to just try and extract a few broader themes. So with that, I see we've got a question or comment and I'll respond to that in a minute. But let me invite others of you to raise your hand, put comments in the chat or put comments in the Q&A tool. This question is from Don Waters. He is interested about the value of architectural thinking in which related services are proactively designed to work together. Here is an evergreen question. How do we foster this kind of design? APIs is necessary clearly but not sufficient. They need to be in the right places with the right functionality. Service providers are not always interested in interoperability because they want to build monoliths or they are unaware of useful interrelations. I think all of those things are very accurate. I think that maybe one thing that is becoming clearer to me is that we have a tendency or have in the past had a tendency to think about these kind of monolithic systems for doing things like repositories to preserve things. And we need to back off from that kind of monolithic thinking. It is not just about APIs. It is maybe about taking problems in smaller pieces, not trying to do everything. It seems to me to be more than just APIs and and more about how you scope services and how you compose services together. And I wish I had a more articulate way of saying that. It is an observation that has been kind of creeping up on me as I have looked at the evolution of systems and linkages between systems over the past four or five years. Some of it maybe is hidden in some of the discussions that people have had about microservice architectures. But I also think we need to look realistically at how smoothly these things compose and fit together and how usable they are in collections. I am really going to be interested as PressQT matures a bit to play with that system and see how easy it is not just to integrate another service into it, but as an end user of the system to find my way among services and leverage across services. I think that is a really key question that hasn't always been kind of rigorously tracked. So those are a couple of reflections, but I am not sure it is a great answer. I wish I had a better one, Don. Other comments? What stood out to you all among the presentations for week three? Or what were you really surprised not to see anything on? That is always an interesting thing to consider, too. What did I expect to see and didn't see anything about no takers on this? I am very surprised, given how many very smart and very insightful people I see in the audience here. I am quite surprised that nobody wants to jump in with an observation. I will throw out one, which is that certainly in the broad area of technology, I would have expected perhaps to see a bit more about machine learning. On the other hand, I continue to try to understand exactly who is really doing what with machine learning inside of organizations. It is clear, for example, that there are a significant number of digital humanities projects that are incorporating various kinds of machine learning technologies. It is not clear to me, though, that our institutions are doing a lot of machine learning as opposed to trying to make, for example, collections that they steward friendlier to machine learning. I think that trying to get a somewhat better understanding of that may be an interesting issue in the months ahead. If anybody would like to make a verbal comment or set of observations, please just raise your hands and we will turn your microphone on and make that happen. Wow. I am hearing deafening silence out here. Perhaps if there are no further questions or additional observations, we will simply wind up a little early and you will get a little bit of time back today. I will offer a final call for comments and observations on week three. Hearing none, I will invite you to join us starting tomorrow for week four. We have got some really interesting things lined up for week four, including a couple that are really just lovely. I look forward to sharing them with you. I will note that I believe, as of today, have released the week four prerecorded videos. There are many fewer of those than for week three. For some reason, we got a tremendous number of sessions for week three relative to the other weeks. Many of them really were the kind of thing that was highly amenable to a prerecorded video. We had a lot of those for week three. We only, if I recall correctly, have something like three or four prerecorded videos for week four. I hope you will enjoy those. I hope you will join us for some of the week four sessions. Final call for observations or comments that anybody wants to make. Hearing none, I will thank you for joining us. I look forward to seeing you tomorrow or later in the week. Thanks for being with us today.