 Okay, great. We'll get started here. I'm sure a few people may filter in as the lunch hour. So it is our challenge, of course, to keep you awake and engaged after that really lovely meal, including some of us had to drag ourselves, I think, away from that sunshine. But it's really great to have all of you here today. This session is the privileged link, open access version of record, or let the user decide. And we look forward to your engagement both here in the discussion, as well as if you want to, on Twitter, we have a privileged link hashtag too, if you want to join in that conversation. So my name is Lisa Janaki-Hinschliff. I'm the coordinator for information literacy services and instruction at the University of Illinois at Urbana-Champaign. I am very honored to be able to have organized this session here today and invited my colleagues who are here to join in what has been a very interesting Twitter-based discussion primarily. And in fact, many of us had not actually ever met each other, even though we've had extensive conversations together in some cases, about this particular issue of as the open access content becomes more and more available, should libraries be doing things to surface those open access copies? And how should we think about surfacing those copies relative to the versions of records that we may pay for in our subscription databases? I must pause for a second and express my thanks to Ryan Rieger in particular for the Twitter conversations that we've had that in part motivated me to propose this session and through which the team here that is pretending came to be. Ryan says he's following along on Twitter and said, I hope people tweet, so we're sorry that he couldn't join us as well from Canada because that would have made us even more international. So I'm going to step back here and sort of, we work all the time with all these different copies of a single article, but we tend to use two terms, the preprint and the version of record. And often we use those terms to mean one is open and the other is not. That is of course not the case. And the other thing I particularly noticed recently, as I was looking at the NISO journal article versions standard and noticed that there's not a single version in the versioning process that is called preprint. And so a colleague and I were earlier today, we're like having this discussion like, well, what is a preprint? And I was like, no, that's a preprint. And she's like, no, no, it's this other thing. And I realized that we've got a term that we don't even use, that we haven't sort of like said, it always means X. So I particularly want to point out that there's a lot of different versions that could be coming through the pipeline. And in particular, any or all or none of these could be open access. And the question of equivalency is obviously one that is very daunting within this. I personally think we are going to be facing an increasing number of cases in which libraries are going to be confronting this question because increasingly open access content is being embedded within our tools. And in user workflow tools, there's increasing question of what should the user encounter first or maybe only. So as we think about what to privilege, we need to think about things like what to privilege when there is a difference between the open access and the version of record copy. Or in one way it's what we might want to say is an open on the web copy and an open access copy that is on the publisher's website. In both cases, those are also both open, but do we want to drive users to one place rather than the other? When there is a difference between the author manuscript, whatever version prior to the version of record, in what cases does that matter? Is it significant enough for all use cases to privilege only the version of record? And of course, we have the options here, which is whether the library should decide or we should empower the user to make a choice. But there are implications for user experience and confusion and education when we push that choice onto our users. Of course, if we decide for them, we might see that as somewhat paternalistic. We can also think about whether there should be unmediated access to all of these particular versions or should it be waterfalled, where certain copies are made available if they're available? Or even still, let's say they're all there, but what priority do we list them in? Because we know that users start with the first link when they see a table of contents, say, in the SFX resolver in many cases. We could also think about mediated access. Perhaps you have some sort of priority cascade there as well, and any user can get access to the open access copy, but you mediate whether a version of record will be made available to the user, particularly in cases related to interlibrary loan or document delivery. And maybe not all users or maybe not in all cases, will we provide the version of record? I think these are very real questions that will be facing us, particularly as I don't see anticipating a time in which libraries are just unaware of what they will possibly spend all the money that they have on. So that is not the scenario we're in. And so we're in some sort of rationing, prioritizing, scarcity, austerity, whatever word you like, of having to choose how we spend the monies that we have. The other thing that I'm particularly grateful to Ryan for helping me think through was the idea of how, depending what perspective you privilege as you make these decisions, you might come to a different decision. Ryan and I have actually often disagreed about what libraries should do, and he recently said to me, I think we disagree because we're prioritizing a different perspective. Some of these perspectives may be aligned in certain cases, and in other cases they'll be conflicting. So what is the priority for the reader? What is the priority for the author, particularly in this age of analytics? It might be important to the author that people be getting the version of record, even if that's not the most convenient for the reader. What is the perspective of the library? And I'd like to particularly at least problematize our library perspective here in one way. I think it's very common for us to think about driving down cost per use for something that we pay for. But I'm particularly indebted to Roger Schoenfeld's work, Red Light, Green Light, for giving me a new perspective on that, and helping me ask the question, is it driving down the cost per use if there's an alternative where the cost is zero? Would we instead be actually over accessing, if you will, our version of records that we're paying for, making that payment look more valuable than it actually is? Should we be driving up the cost per use of the version of record in order to position ourselves better for negotiations by being able to say, we only need this many accesses to that content, and this cost per use is too high? There's two ways to adjust your cost per use. One is to drive up use, the other is to drive down cost. Now, obviously, I think we all know what the publisher perspective is probably as far as which version they would like users to be getting access to. And then finally, I think this is a hard role to talk about, but this sort of political stance, and what I finally settled on here was the open access advocate perspective, that it is always better to push open access in sort of a political or theoretical kind of commitment sort of way. So depending which perspective you approach this question with, you might come up with a different answer. So what I'm going to ask you to think about today is if you were facing like, you've gotten all kinds of information, you know about libraries, if you had decided today, and like you had my little Twitter poll last week, what would you like the user to be presented with? Just the open access options, just the version of record, all of the options and let the user decide, or some other configuration? We're not going to come to any conclusions in this session today about which one we should all do, and it's even possible that we shouldn't all do the same thing. What I'm going to do is now invite my colleagues to explore some of the different aspects of this, and each of them will introduce themselves as they come to the podium. Hi, my name is Aaron Tay. I'm from the Singapore Management University. I'm the library analytics manager. Today I'm here to talk about the rise of open access and how it has impacted not just the way our researchers do their work, but also the way our libraries will do their operations. So one of the things that I'm going to say is that I'm going to argue that we are at a point in history where roughly half of what our users want to access is either open access or at the very least free to read. In 2014, I came across an early study that said that 50% of items that people were looking for were free to download. At the time, I thought this was just an overestimate because earlier studies were showing lower figures. But as time went by, more and more studies started to appear. For example, lazy scholar extension, which is an extension that searches for free open access items. The author actually found that for articles that had a DOI, roughly half of them were already available for free. Unpaywall, which I guess most of you are familiar with, did a similar analysis, and again roughly half of it was available. And the most recent preprint where the researcher actually painstakingly scraped Google scholar results, and again they were finding over 50% of the links were free articles. And they even commented that their results were very consistent with the earlier studies that I've shown. And of course, with the rise of preprint servers, we expect the level of open access to rise even further. So as we transition from a world where most articles were behind paywalls to a world where a significant portion of articles are free, there are many, many implications to libraries. I'm just going to touch on three of them. Firstly, there's a rise in the use of tools that help users find open access articles. Secondly, libraries have to think about integrating such functionality into their discovery access services. And lastly and probably most controversially, the idea that you might even consider the level of open access for journal titles as a factor for subscription and renewal decisions. So let me just first talk about all these open access finding tools. If you're not familiar with these tools, they are basically browser extensions, typically Chrome extensions or occasionally Firefox extensions that you install on your browser. And when you visit any page with an article, the extension will automatically search for a free copy and display the free link to the article for you. The very first Chrome extension that I came across that did this was something called lazy scholar in 2014. Some of you may not be as familiar with lazy scholar basically because this was the extension created by a PhD student at Purdue, Kobe Voland. And he is a PhD student in nutrition and he has no ties whatsoever with the library or publishing industries. So I've talked to him and he created this extension for himself and what he does is that he basically scrapes links from Google Scholar. So it's pretty clever. When you go to any page, it will scrape the free PDF as well as the library links resolver link. So it allows users to instantly see what free PDFs there are and what subscribe versions are available to him. Since then, Google Scholar also came out with a similar extension. It does exactly the same thing in 2015. And besides these two extensions, probably the most famous ones besides Google Scholar extension would be the Unpaywall extension. The Unpaywall extension together with the open access button only finds free articles, not the subscribe articles that you may have access to. And the other thing that's special about these two extensions is that they have an API that can be used. And some of them are actually used by some of the other extensions that I really show on this slide. The last two extensions that I'm going to talk about first is a library access by Lean Library. This is actually, my co-panel is actually the CEO of the company. And basically I understand it targets libraries. So the libraries will actually subscribe to this extension and they can provide their users access. And lastly, a co-piano which was in the news lately because it was actually acquired by Caravad. But the thing is if you look at the slides, you look at the numbers, it's relatively small. Even for Unpaywall, which is probably the highest profile, there are only 128,000 users. But the thing is this open access finding functionality is starting to appear in places such as databases. So for example, the highest profile integration is in Caravad's web of science where they use the Unpaywall API to actually surface articles that are free. Not just Unpaywall, but things like digital sciences dimensions, science open, Europe PMC, these are all databases and indexes that now use the Unpaywall API to actually surface free articles. Even library link resolvers are starting to do that. If your library is using SFX or using 360 link, you can actually turn on a functionality where it will display free articles from using Unpaywall API if there are no subscribe articles available. And I understand that X Libris is going to add this to the Elma U resolver. It's on the roadmap. So the thing is because of all this integration, the founder of Impact Story, Jason Primm, he mentioned on Twitter that in fact the number of API calls for Unpaywall actually exceeds the number of downloads in Syhub. So this shows the impact of the ability to find open access articles. Of course, not just daily basis, but web skill discovery services are also starting to pay attention to the significant amount of open access items out there. So Summon now has this open access indicator. Primo will be following suit soon. Not just that, but they're also adding more open access sources. For example, Core, which is by GIST. Basically, it's a big global aggregator of open access resources. It will be soon be available in the index of Primo and Summon. The thing is about web skill discovery services is that they have always had the possibility of turning on open access sources in the index. But traditionally, I've found that librarians are reluctant to turn them on basically because the moment you turn on open access source, you will be flooded with articles that were open access articles that were not. For example, if you turn on an institutional repository to another university, you get some results that had full tax and you get some results that had metadata only. The same thing came with identifying hybrid articles. So this is an area that I believe X Libris is now taking it more seriously. I think last year they set up a base camp to discuss such issues to better identify open access articles inside such sources. The last thing I'm going to talk about is the idea that you could actually substitute paywall articles with open access articles. I believe GIST had a pilot project on integrating such functionality into ILL. But the one I'm going to talk about is this controversial idea that you would take the level of open access of a publication and use this as a factor to decide whether you want to renew or cancel. So this librarian last year actually talked about this. She came out with this formula where you actually discounted the cost per use based on the level of OA available. So for example, if you counter stats for JR5, you would just discount it by 50% if say 50% of the articles were free for that journal title for that publication year. It's a lot more complicated than that so I suggest you read the paper to get a sense of what's going on. Of course there are a lot of reasons why people don't do this. But I think one of the practical reasons why open access levels are usually not used is that up to recently it was very hard to determine the level of open access in the journal title. But recently in the last six months it has become easier. I know of three ways. One way is to pay for a service by one sign. They'll give you a list of journals that you subscribe to and the levels of open access. The second way is to get your hands dirty use the Unpaywall API. Ryan has also created this Python script that automatically does it for you. You just type in the ISSN and it pulls out the open access percentage. And lastly the easiest method that I recently discovered was that you could go to dimensions which is free filtered down to the journal you want and just look at the open access articles inside. So the thing is all I've mentioned are things that will impact the way the library will work. But the underlying assumption for all this is this. All the articles equal value. The different versions are of different value. So it will be interesting to see how the library will react to this for link resolvers. How are they going to show it for deciding cost per use, discounted cost per use? Do you discount more if the articles are pre-prints or if they are final versions? Thank you. Just this way, right? Thanks. I'm very pleased to be here. Lisa, thank you first and for getting this all together. I'm a bit intimate with these heavyweight colleagues. So I'll try to do my best. I'll give you a quick intro of myself. My name is Jo Wontills. I'm not a librarian by profession. And before I found the Lean Library I did work at the University Library. I worked at the Innovation and Development Department over there. And I was a program manager for a program that dealt with what we saw as a change in user behavior where our patrons were not by default starting at the library anymore. Bypassing the library, bypassing the library website, diving into databases directly. We were trying to come up with ways on how to deal with that change that we saw. One of the things that we thought we could do was just keep pushing people to the library website but we tried to come up with alternatives. U-Turtle, of course, is quite well known for trying to do away with the local discovery tool. So if your users are not starting at the library website anymore do you even need a discovery tool or is Google Scholar or any of the other discovery tools out there is that sufficient for your end users? I did something else myself as well. I started experimenting with browser extensions and as an idea to put those library services into the user's workflow. There's a famous quote from Lord Condemnsee of course about libraries needing to be in the user's workflow, hence also the title of my presentation right now. And we got a lot of positive feedback from end users about having those library services available in their browser and that's what propelled me to actually found the library and develop a professional fully supported version of that browser extension that's available and in use with quite a few libraries worldwide right now. And just real quick, what it looks like because I found that it's actually quite useful to give a short just a few seconds video of what it is, what it might look like. Oh, it's two. There we are. So what I've set up here is I've typed in an URL web address of one of the databases that's licensed by the library that I'm affiliated with which is in this case the Dutch National Library. And if I just go there, what our browser extension does is get this to run, yes. What it does is it recognizes that domain as one of those domains that can be opened by the proxy of my library does so automatically for me and afterwards also informs me. So as an end user, I'm helped with getting access to that resource and I'm also informed that it's my library doing this for me. And as a plus side, I don't have to know about this domain, I don't have to recognize it, I don't have to head over to the library website, find a list of databases, try to locate this. I don't have to do that all. That's just one of the things that our browser extension does and one of the ways that we put those library services into the user's workflow. And I'm not here about doing a sales pitch, I'm here to give you my perspective on that privileged link and so this is not really a rehearsed presentation, the sales pitch come out way more smooth right now. So you'll have to bear with me. And again, of course, if you're even saying any really stupid things, I'd love to hear that afterwards as well. So yeah, just one perspective on the library collection, real nice link with the plan re-opening session from Joan, of course. When I, again, I'm not a librarian in bright profession so I really needed to dive into the subject and I'm just trying to figure out, so what does it do? What is it, that library collection? And I came up with this diagram where you can see what's out there, all that scholarly material that's available worldwide or maybe it's not available, but it is there. And then there's your library collection, a subset of what's out there. And it's necessarily a subset, of course, because now library can offer access to everything and anything. And it's a curated collection. It's a selection made by your librarian, the best possible as they see fit, of course. That's probably not important. And the size of that collection also depends on your own institute, of course, if you're affiliated with a really rich library then that size will be bigger than otherwise. And the thing is, of course, libraries did maybe still do present that library collection as a starting point. This is where you have to go. This is where you go when you start with the local discovery tool. And it's also sometimes presented as more or less of a walled garden. If you stay within that library collection, then you're okay. And the versions of the stuff you'll find, they're good enough for you. That's what you can use. And with that limitation, the factors that come into play, of course, are costs but also convenience. So if you stay within that library collection, then it's easily accessible, all that stuff to you. And having convenience being a factor or in what's available to you, that kind of strikes me as odd. If you're a serious researcher, you wouldn't care about something, about the convenience factor coming into play that much. I mean, if you really, if you just find an article and you need it, you don't care about how difficult it's going to be to get it. You just want to see it. And again, also from a library perspective, if you go outside of that collection, then you're sometimes a little bit on your own. There might be predatory journals out there and questionable versions, maybe illegal versions of articles. Who knows, but you're a bit on your own. And it sounds a bit ridiculous, but then I found actually this one, and this is actually what it's a screenshot from, it's just a recent screenshot from Prima, of course, and it actually does show what I was just telling you. So if you see this checkbox over here, expand the old library collections, that checkbox is by default, it's off. So by default, if you go to Prima, you're searching within that wall of garden, within that collection that has convenience, that's quite a big determining factor. And if you really want to step out of that wall garden, then you need to take deliberate action. You need to say, I want to expand my search beyond what's easily available through my library. So what happens then, of course, is tools like Unpayable, what's easily available to me becomes bigger. So now it's just not my library collection anymore, but it's also all that stuff that's available as open access. So Unpayable just simply plugs that into your browser, and if you land on one of those articles, and it's available in open access, it will just tell you automatically, really nice, really nice. And then there's a question of course, what drives this change for end users, why they expand their searches a bit? Maybe also Google Scholar, because we see from a lot of libraries that they recognize that end users are going to Google Scholar, why is that? Is that because those end users are unsatisfied with that wall garden, or is it just simply the ease of use of the user interface? And as one of those consequences that if you expand your searches, is that you get confronted with all these different versions of those materials, all those different versions of open access, and do you know how to deal with that? Take a look at this screenshot from Unpayable. It says, it tells you about the various versions of open access, green, gold, bronze, but please do note the wording. If you really want to see, if you want to get more information on those different versions of open access articles, you need to select the OA nerd mode. What's that all about? So really, you need to be a nerd to actually understand all those levels. You need to be a self-assigned nerd. That's kind of strange, of course. So there's little assistance on those various different versions. Maybe just a question for the public. Who would say that they know about all these different versions, green, gold, bronze, gold? Who would say they're confident in not many people? Heather, of course. Yeah, you just studied work that I was talking about. I had to work on it. I mean, I'm not even going to ask how many of you think that all your patrons know about all these different versions, of course. And then, again, this one, the same screenshot as Aaron was using, green accepted open access, green published open access, gold, west, bronze. I mean, how are you supposed to know? How are you supposed to know if the version of the article that you find actually connects with your context? So maybe you're a freshman and your paper needs to be in this afternoon. So which version can I use? Maybe you're a researcher. You're simply skimming topics. Which version is okay for you? Maybe you're a professor finishing a manuscript. Which version is okay for you? And is this sufficient to inform those patrons on which version they can use? So what this all could end up, I think, is maybe a bit different perspective on the library collection and also a bit different perspective on the role of the librarians. Think about this. So what if you see what's out there as where you start as an end user? So you don't start within that wall of garden of the library collection. If you take this perspective, then the library collection becomes merely a convenience tool. So you're looking for something. You find something that's interesting to you. And if it's available through your library, nice. Fine. So your library is helping you to get it onto your screen in an easy way. So what if it's available in open access? Really nice as well. Just boom there on your screen. But there's a role to be played by the librarian. I think in helping that patron out with determining which version to use and in which situation, in which context. And also if you find something that's out there and it's not within your library collection because your library might be small, because it's just not in there. And there's also no open access version to be found. So what are you going to do? I think there's another role for the librarian there as well in helping you get a version somehow, maybe through interlibrary loan or maybe where the librarian can help you with simply sending an email to the author. So for me, from my perspective, I think there's a role for the librarian to be filled here in assisting in determining what version to use. And I think a librarian should step up and claim that role. Thank you. My name is Maria Agazarian. I'm from Swathmark College where I work with digital resources and scholarly communications primarily. And today I'm going to talk a little bit about how these decisions of privileging links affect user needs and user experiences. So when considering whether or not we should be privileging links, we need to focus on a few things from the user's perspective. Is it easy to access? Is it accessible? And if multiple versions are available, are the differences between those versions meaningful to the user? I work primarily with undergraduate students so that influences a lot of my perspectives and a lot of the language that I will use during this talk, but for the record, I think that there are more similarities than differences when it comes to working with graduate students or faculty. So the first thing that I'd like to talk about is ease of access. When someone is looking for access to an article through library resources, they might be faced with a screen that looks like this. There might be multiple links from multiple providers, or if we don't have access to it, there would be a link to request it through interlibrary loan. What we currently hear from students is that when they're presented with multiple links, they always click on the first one. Why? Because for most users, there's not really a meaningful difference between these links. For users who are a little bit more experienced, they might go after a name that they recognize. So for undergraduates, a lot of the time, if JSTOR is at the bottom, they'll go for JSTOR because they recognize it. But generally speaking, they can't really tell the difference between what a publisher is and what an aggregator is because these differences are not meaningful to them. They don't care how they're getting access to the content as long as they're getting access. And I don't think this priority is unique to students. I think that it kind of extends to wider populations as well. And I think that's one of the reasons why SciHub is so popular. It's not really about how they're getting access. It's not really about the paywalled content that they're looking for, that they can't access it in any other way, but the fact that going to SciHub is one click and they're getting access to what they're looking for. Anything that interrupts that ease of access then becomes a problem. So when we have resources in this menu that don't link at the article level, but dump people on the journal page, we often get reports from users saying the linking doesn't work. You know, I can't find what I'm looking for. And sometimes I have to write back very politely worded emails. Hello. You actually have to search for the article again once you get on the platform. Or actually, this is an open-access article and I understand that the linking doesn't really work well, but if you had gone to Google instead of the library catalog, you would have found it in five seconds. And that's what I did. And here it is attached to this email just so you don't have to worry about it. So this is difficult. Whether it's troubleshooting an actual linking problem, whether it's providing user education in those brief moments, or filling an interlibrary loan request, any step of staff mediation, in my opinion, is going to interrupt that user's ease of access. We can't be responding to emails 24-7, I think as much as our users think that we do live in the library and work on the weekends. However, instant availability is not the only issue here. No matter how many links are available, if none of them are accessible, it's not going to make a difference for our users who use assistive technology. PDFs are a barrier for most folks who use assistive technology, such as screen readers. In general, PDFs are the least accessible version to read an article in, and they're the most effortful to remediate for accessibility. In order to be accessible, PDFs need to be well tagged and well structured. That means that the content in the PDF needs to be tagged so that the screen reader understands the order to read it in. They need to understand the difference between headings and text. If there are images present, there needs to be alternative text that describe those images for the user, and there also needs to be OCR run on the entire document so that it understands it as readable text rather than just an image of text. When we are assisting students who we know need accommodations, we don't assume that any PDFs will be accessible at all. We direct students to full text HTML when possible and we ask faculty to do the same when they're working with assigning course readings. So one issue here is that we cannot currently privilege links based on accessibility because there are no tools or link resolvers that can identify the accessibility of an article. And if you know of any, please let me know during Q&A. I would be very happy to hear about that. So if we begin streamlining the process for our users by removing links, we may actually be unintentionally eliminating links that lead to more accessible versions, which complicates our assumptions of what is best for the user or what the user actually needs or wants. Another issue is that as the culture and practice of sharing preprints and postprints continues to change, different article versions is not just an open access issue. Publishers are increasingly making accepted manuscripts online before the version of record. And while it might be a short amount of time before it's replaced with the final version, users are going to be encountering these versions whether or not we promote them actively. So if students are encountering an article that looks like the image on the left on a publisher website, are they going to be able to understand maybe some of it different from the article on the right? Even if students do understand that there's a difference, maybe they recognize that the formatting is a little bit different, they don't care. I'm betting you that as long as they have access to one, they're going to be happy. And especially if it's one with simplified formatting, it might be easier for their screen reader to read that simplified formatting. So they're not going to be picky about which version of the article they end up with if they can use it. So at this point you might ask, well, what about the faculty? Wouldn't their professors care if students were citing a version other than the version of record? So we would assume that as authors, faculty would be a little bit more attuned to differences between article versions. But in my experience, it's really difficult to communicate these differences even to faculty. That's why when I write to faculty for preprints or postprints, I often end up with the publisher's version in my inbox saying, here you go. You could put this in the IR. No, I actually cannot legally do that. So I don't think it's necessarily because they're just not reading my emails. Maybe that's it, that they don't read my emails where I say, this is what this is, this is what this is. They might not be distinguishing between a postprint and a draft, but they might consider a draft of an article that they don't actually want to share publicly. So if we avoid linking to preprints because we think that faculty only want students to be reading peer reviewed articles, the question comes up, are we using peer reviewed as an indicator of scholarly source or measure of quality? Academic level and discipline obviously play a role here and it comes back to the discussion of whether or not it is acceptable to cite preprints. Many folks argue yes, though it should be recognized as a preprint in that citation so it's clear which version the researcher is citing. But if users genuinely can't tell the difference between article versions, how are we supposed to communicate this expectation, especially when they're writing that paper the night before? So if we consider instead the accepted manuscript which has been through peer review, we come to another point of tension. Who is adding value to the article at this stage? Is it the peer reviewers or is it the publisher? That is to say are the resulting published papers significantly different from their prior versions? In content, accuracy, formatting? Some people say no. One 2016 study of Works and Archives found that there were no significant differences in aggregate between preprints and their published versions. And when that study was published in 2018, they also looked at bio-archive and they found that there may be differences in article versions that vary by discipline, particularly in regards to formatting and copy editing. While formatting and copy editing may seem like trivial differences if the content is largely the same, as I've already explained, formatting is a big deal when it comes to accessibility. By making full text of articles available as HTML, publishers have an amazing opportunity to add significant value to the work in terms of accessibility. This is a service that institutional repositories and author websites don't typically provide. However, this is not to say that the publisher's version PDF will be more accessible than the prior version, per se. So I put this up here if you are not familiar with how the accessibility of an article may differ across versions. This is just kind of a quick example based on a recently published article by one of our Meritai faculty in biology. So they accepted manuscript of the article, which was available on the publisher's site for about a week before it was replaced with the final published version. They did an accessibility check. There were all different kinds of errors, and about 3,000, one of those kinds of errors is content error, tagged content error. There were 3,000 tagged content errors in that one document, which is almost the entirety of the document, and that's probably because the text of the article wasn't actually tagged as text. So by comparison, a week later, you have the publisher's version PDF, still has lots of accessibility errors, lots of different kinds of errors, but the tagged content errors seem so bad. But when I tested each of them using a screen reader, the accepted manuscript was actually a lot easier to read than the publisher's version PDF. And even easier to read than both of them was the publisher's version HTML. This seems to suggest that maybe prior versions with simplified formatting would be okay to be prioritized. We could justify prioritizing them because they might be naturally more accessible than their version of record counterparts. But unless an organization has made it clear that they are providing a remediated PDF, we cannot assume that institutional repositories, for example, will be providing the more accessible version either. IRs that use cover pages are a good reason for this. So librarians such as Matt Ruin at Grand Valley State University have made it well known that the process that Digital Commons, for example, uses for repository works that are uploaded when someone uploads a work to a Digital Commons repository and chooses to use cover pages in that structure. The process that generates the cover page with the link back to the repository, a link back to the article, citation page, a link to the author's other works, that process actually strips the accessibility tagging out of the PDF. So if you had a perfectly accessible document, upload it, nice cover page, accessibility is gone. So if we privilege open access links in our link resolver or elsewhere that would lead to an accepted manuscript version in a Digital Commons repository, we can almost guarantee that these works will not be accessible because of those cover pages. Authors and IRs, I think, have the opportunity to change the landscape here, though. One of our faculty, for example, has made several of his pre-prints available online as HTML pages. These versions are infinitely more accessible than the publisher's versions, because the publisher is actually only providing these as PDFs. If institutional repositories provided HTML full text that included the DOI, the version of record, this may be able to help tools like Unpaywall and Open Access button with the more accessible version. But if our tools and technology remain PDF-focused, then we risk alienating users who use excessive technology. So we're faced with this conflict between what we believe is a good user experience and what we know is necessary for accessibility. Do we provide a single link for easy access, even though we may cut off paths to accessible articles, or do we present the user with everything, even if it may be overwhelming, because one of those links just might be accessible. As we are faced with decisions with whether or not to privilege links, we must consider who or what we are prioritizing with these decisions. As Joan said in yesterday's plenary, are we centering the needs of our users, or are we privileging the needs of some users over others? Thank you. So, even having heard all of this already once before, my brain is swimming again with all of the different ways. So I want to present to us just again. We're not going to do like a poll, but whatever you had in your brain 32 minutes ago, as the way you answered what I asked you on the left, which was before this presentation, how would you want to or would you want to privilege any particular kinds of links or versions? I'd like to offer you the opportunity to think about whether you would answer that differently now and open it up for a discussion, dialogue and questions. Thank you. Well, we promised an issue-oriented briefing, with Mary-Lise's comment, because I think having heard said that we achieved that, that's a good place to end. Thank you.