I just answered my first 'real' reference question today (or I will have, when the patron receives it). In fact, that last bit is somewhat of a problem in SL, as I would like to be able to send the response in the form of a document, for several reasons. First I would like to be able to 'brand' output from the SLML with a graphic, and the current notecards don't permit such formatting.

Second, with the patron offline, and unless I have asked for their email (which may be perceived as a breach of privacy), I must either send the response as an IM, or hop online on and off through the day to see if the patron is online. I'd love to see Linden (or someone!) create a communication system more sophisticated... perhaps there could be a way to email in-life, or to have a private cubbyhole at the library, so that documents and requests could be asynchronously exchange in a confidential way that also permits branding - or images, if that is appropriate to the question. This should be an opportunity to push tech forward!

One other thing I'm pondering is the nature of online ref in this particular environment, with the constraints we have. I spoke with Lori and told her - I used several proprietary databases to answer the reference question, and to be able to provide subject headings the patron could use in searching those databases in a RL environment, which may be her/his preference. I could probably not have answered the question without using those databases... in fact, I'd searched PubMed and Google Scholar, then used what I found (PubMed had nothing that I could find) to chase down subject headings. We make an assumption (don't we?) that a) we can answer the question adequately using free resources; b) the patron is willing/able to follow up with a rl librarian; c) that questions will not include the need for more in-depth information or even training. Would you have done just-in-time training for PubMed, Google Scholar, and then referred the patron to a rl librarian when nothing was found? Is Saving the patron's time possible in the SLML environment, or do we risk engendering confusion? Even with a shared understanding that we're beta, it's a problematic thing.