Some thoughts about (authentication) discovery aimed at librarians

So I’ve been having an interesting conversation with that Dave Pattern about authentication this morning and why librarians like EZProxy so much. I’m not going to get in to the whole discussion about the problems of proxies, the issues with faking an IP authentication and the lack of personalisation. I want to talk about why we don’t just do EZProxy. It’s about user behaviour.

I’m going to make a probably quite unfair statement, that is highly generalised, but might help explain the problem.

Librarians think that user behaviour can be taught and controlled until the users learn to do things ‘correctly’. Developers realise you cannot control user behaviour and try to design for that.

What do I mean by that? Well the main problem with EZProxy is that using it is counter-intuitive to general discovery. In order to use a resource, I don’t need to know where that resource is, I need to know where my institutional EZProxy login is. So I have to search for something other than I want. It isn’t that easy to find. A Google search for ‘EZProxy Huddersfield University’ brings up this result.

Not useful. So OK, a student might not actually type EZProxy. They might just go to the library. So I try this. After 4 clicks, I can type ‘ScienceDirect’ in to a thing called Summon and it comes up with this. Hmmm, but I just wanted to search ScienceDirect. However if I click on one of those results, I then get pushed out to another strange URL ( – oh god, where am I going, I don’t know….and then there is another screen – do I hit login? Do I hit article? What do I do?

Or we could just use Google. Just put ScienceDirect in to Google. In one click, I’m searching ScienceDirect. Yes at some point I am going to have to log in and that has to be as easy as possible (it certainly isn’t right now) – but don’t be so sure that your portal approach is any easier. There are just as many clicks, just as many breaks in user experiences, just as many things to learn. That’s the important thing – being forced through the library is an entirely LEARNED behaviour.

So these are some of the arguments Dave put to me.

  • EZProxy is seamless when used with a discovery service. Well I think I have just shown it isn’t that simple.
  • We tell our users to configure their google scholar with our link resolver. Well again that’s something you have to learn, and think to do. That takes a fairly sophisticated user. I’m sorry, I’m not a sophisticated user. I just type ‘ScienceDirect’ in to Google.

So give us a break please librarians. There is a reason why we are doing what we are doing and I think it makes good sense.

So yes we know the authentication discovery area is horrible at the moment. It’s terrible! It’s horrible! We are working hard on it. I’d urge you to look at Rod Widdowson’s presentation from the REFEDS meeting (coming soon) and the work Andreas Solberg has done on DiscoJuice. It can look so much better.

We also need your help. We can’t design the publisher’s interfaces for them. Us techie people have a horrible time trying to explain to publishers that we want their interfaces to be better. Only their customers can do that, and guess what – that’s you! So don’t complain to us that the ‘authentication’ system is broken. It isn’t. Get out there and tell the publishers their interfaces are broken and there is something they can do about it. There is something YOU can do about it. Anyone willing to take up that challenge?

13 thoughts on “Some thoughts about (authentication) discovery aimed at librarians

  1. Clare

    I’m sorry, that is a generalisation too far. I’m a librarian and I really tried to make Athens work before we moved to EZProxy. Yes, we contacted lots of publishers and I lobbied reps when they came, to try and get them to automatically pick up the Athens details saved via DA. Yes, I tried to get students to go to that tiny ‘Alternative login’ link rather than putting in their username and passwords in the main Athens boxes and getting error messages (even on the main Athens pages…). However, we have moved to EZProxy now and I can say that problems with off campus access have reduced to almost nothing. We used to get at least 5 queries a day when we used Athens.

  2. Chris Keene

    I agree with much of this.

    I think too many people in this sector think the answer to everything is to run induction/training sessions to tell people how to do things, the vast majority of people of course will never have attended, or probably heard of the training.

    We need to accomodate how people work. Sussex has just published a library e-strategy (on our homepage), about half of the basic concepts are about being open and working with the wider web, not just our site.

    So yes, ezproxy is flawed in the sense you need to either start from the library (or uni) site, or use a tool (such as libx etc) to make use of the url.

    For this reason ezproxy can not really be the only tool in the toolbox.

    But it is a damn useful tool to have. When people do start from a library site it makes linking to e-content behind pay walls a simple process. It’s simple technical design means it tends to just work. No maintenance, regular jobs etc, and it’s easy to troubleshoot when things go wrong.

    Compare this to when we had Athens DA. We had a lot of people with problems. Maybe it was cookies on the users computer (but how to check), maybe it was the central Athens system, maybe it was a problem at the publisher end, maybe it was some deep bug in the code setup in our IT Services script? It wasn’t helped that to comply with publisher requirements, our Athens DA authentication (hmm, that was a ‘pin number’ wasn’t it!) only worked with Sussex usernames which were the primary username for a student or member of staff, so we couldn’t setup a test account. Turned out one of the issues was a system clock on one of our web servers was slightly adrift. took years to sort out.

    Shibboleth is good. i like it. I really think the ux is currently a bug. We have to tell people “we don’t use athens…. don’t click on ‘log in by athens’…. unless it says ‘log in via athens or institution login’ “. When users hit an article page, too often the publisher promotes the ‘pay money to buy’ and too little on ‘click here to signin to your institution to see if they provide access’. The need to accomodate other authentication systems for different types of customers doesn’t help. I’d also like to see publishers avoiding needing special setup for shib, or requiring unique attributes. I’d also like more to tell service providers that the person accessing this is a Chemistry academic, so this information can be presented in the COUNTER stats.

    One final thing. Users wouldn’t search for ScienceDirect to then do a search there (why would they? though oddly they do for jstor), they would normally search with some keywords (when they are writing an essay and need to find stuff), or the title of an article or journal (when they know what they are looking for).

    So, in summary, I think a combination of auth methods is right, and ezproxy is a good ingredient to have in this combination. 🙂

  3. nicole Post author

    Thanks Clare, as I said it was an unfair statement, but trying to make a point. However I disagree that ‘lack of queries’ = success. What about your students that hit the resource from the open web without visiting your portal? They have no way at all to access the resource. So at least with Shibboleth and Athens you have some chance of logging in if you don’t visit the library first.

  4. nicole Post author

    Chris, I’m interested in your comments that users find EZProxy ‘seamless’. I’ve heard this often, but is there any evidence to support it? I certainly didn’t find the Huddersfield experience seamless. Perhaps we need some empirical evidence from user testing to back this theory up?

  5. Stuart

    Dear Nicole

    Your post seems to imply that you don’t think that librarians look at the way students use resources. Or that they haven’t tried to lobby publishers’ to change their websites.

    But it doesn’t matter at what point the user bumps up against having to login. With EZproxy they do that at the start of the process, rather than at the end with no confusion over which login they’re meant to use. That’s why librarians and students like EZproxy.

    There are issues with EZproxy but those have to be weighed up against its benefits. I enjoy a good straw man argument as much as anyone, but to pretend that people use EZproxy from a position of ignorance is somewhat patronising.

    Also, students are there to learn, developing them into sophisticated users is part of what a good library does. I didn’t realise that learning was a bad thing.

    Your post also doesn’t explain what you’re doing or who you are, so how am I to judge, unsophisticated user that I am, what you’re doing and what your credibility is in this situation?

  6. Chris Keene

    nicole, good question.

    Can’t comment on huddersfield (I hear their IT staff are a little slap dash) and do not know of any user testing.

    I can see that the links from our systems: the Catalogue (to e-journals and e-books), the Link Resolver, Journal title A-Z all take the user to a login prompt *if* off campus (some argue we should also do this if on campus for consistency, though my feeling that users probably don’t even notice that sometimes they don’t have to login). The login box is the standard University login box which is used for other systems on campus. Once passed the login they are taken to the article, journal title, book or resource in question. It looks just how they would normally expect and they can access the full text if we have subscribed to it.

    So to me that seems quite seamless, though can’t comment on how users find it 🙂

  7. Chris Keene

    @Stuart “Also, students are there to learn, developing them into sophisticated users is part of what a good library does. I didn’t realise that learning was a bad thing.”

    I agree Information Literacy is an important part of what a Library does, especially effective searching and analysing of information. However we often also have to teach users the limitations of our systems.

    I’ve been in touch with academics and postgrads recently who like to work with CiteULike and Papers (mac app), they like to discover new articles by following people they respect and know, and by following certain tags. But when they discovery stuff they want to be able to access the full text through our subscription. We were able to help them by suggesting they use our LibX toolbar (which was able to add a Link Resolver link to each article page on CiteULike) and make use to shibboleth where available.
    So I think we do live in a world where researchers find their own ways to work and discovery research, and this will vary by subject and by individual researcher, and our job is to help accommodate this.

    So I guess my point is that while training is important, and teaching info skills are part of any good undergrad degree, teaching “you need to start by using X otherwise it wont work” will hopefully be no longer required in the future.


  8. Pingback: EZproxy and Summon – "Self-plagiarism is style"

  9. Clare

    Sorry, I should have said that we have moved to EZProxy as the main way within the library system – we still have Athens for those who may want to start their search elsewhere. However, I have firm evidence from folks who still can’t get in from, say, Google Scholar, because the Athens login presented to them is so confusing (you will tell I have a big gripe with the ‘alternative login’ system – if the push is to get users to use one login via OpenAthens etc, then it should be the automatic login box presented). Given that a logical way to access anything your library subscribes to is to start with its own search tool, I don’t see a problem with advising that. If we want to get to a stage where folks can use any search tool then I have to say, the Athens route needs to be improved.

  10. Pingback: Interesting Things of the Week #107 « Lore Librarian

  11. Vincenzo

    Being an EZproxy adept, I still deeply agree with the argument.
    At least in Italy it is mostly a cultural issue, kind of of Humanist vs Nerd war.
    Librarians simply DONT’T WANT to be involved in the overall technological process; they still think it’s just “computers”, and keep asking IT to run their library the same way, only… faster.
    The idea that IT could actively suggest not only new WAYS to handle the library or interact with the user, but also new PROBLEMS, is heresy.
    However there are compromises: OK, we still “teach” users to use the library main site, but we also inform them about the federated authentication available in the publisher sites.
    Since the library tools, the proxy and the publisher sites are all shibbolized, we hope to somehow evangelize them, at least about the authentication interface.
    But sooner or later this model have to change, otherwise the library will lose any role in the knowledge creation and transfer chain.
    And there are many around fully willing to take their place… at a price !


  12. Pingback: Authentication from E to Z « Paul Stainthorp

Comments are closed.