Wednesday, 13 May 2009

Final thoughts on open hack day (and an imaginary curatr)

I think hack days are great - sure, 24 hours in one space is an artificial constraint, but the sheer brilliance of the ideas and the ingenuity of the implementations is inspiring. They're a reminder that good projects don't need to take years and involve twenty circles of sign-off, even if that's the reality you face when you get back to the office.

I went because it tied in really well with some work projects (like the museum metadata mashup competition we're running later in the year or the attempt to get a critical mass of vaguely compatible museum data available for re-use) and stuff I'm interested in personally (like modern bluestocking, my project for this summer - let me know if you want to help, or just add inspiring women to freebase).

I'm also interested in creating something like a Dopplr for museums - you tell it what you're interested in, and when you go on a trip it makes you a map and list of stuff you could see while you're in that city.

Like: I like Picasso, Islamic miniatures, city museums, free wine at contemporary art gallery openings, [etc]; am inspired by early feminist history; love hearing about lived moments in local history of the area I'll be staying in; I'm going to Barcelona.

The 'list of cultural heritage stuff I like' could be drawn from stuff you've bookmarked, exhibitions you've attended (or reviewed) or stuff favourited in a meta-museum site.

(I don't know what you'd call this - it's like a personal butlr or concierge who knows both your interests and your destinations - curatr?)

The talks on RDFa (and the earlier talk on YQL at the National Maritime Museum) have inspired me to pick a 'good enough' protocol, implement it, and see if I can bring in links to similar objects in other museum collections. I need to think about the best way to document any mapping I do between taxonomies, ontologies, vocabularies (all the museumy 'ies') and different API functions or schemas, but I figure the museum API wiki is a good place to draft that. It's not going to happen instantly, but it's a good goal for 2009.

These are the last of my notes from the weekend's Open Hack London event, my notes from various talks are tagged openhacklondon.


  1. I've been wanting to do this for ages - it even got a passing mention in my mw2008 paperMy feeling was that it should, initially at least, be focused solely on 'exhibitions' as the key social object. Which is why I started the exhibitions domain on Freebase, which now has a decent set of data that'd be a good place to start from.

    So if you wanted to actually think about how this might work, I'd be happy to help... :-D

  2. 'Exhibitions' feels like someone else is choosing my interests for me - ideally I'd be able to specify the particular periods of Picasso's ouvre that I'm interested in (or nominate key artworks and let the system suggest properly related pieces) - but we could take each of our interests as use cases and work from there. It'd be good to have test cases for related API and machine-readable outputs at work, too.

    In my own time, I can have a life from May 28 - September 28 (i.e. when I don't have taught classes for my MSc) - how's summer for you?

  3. The reason I picked 'exhibitions' is because it's a defined type of thing, and has a certain regularity with the type of facets you might pick to build the service around (dates, subject, venue, type of exhibition, and so on). Which makes it a bit less scary to think about how complicated a service you'd have to build.

    I agree though that'd it'd be cool if you could specify other areas or period of interest and build recommendations around that.

    The key stumbling block I thought of is that for all of these things, people would have to manually input all their interests/exhibitions they've been to. Whereas most of the successful social object services have had some degree of automation ( with scrobbling, iTunes Genius) or are based around an action which already happens online (buying a book from Amazon, sharing a photo on Flickr).

    Sites where you have to manually input your data/interests, in the hope that it'll be useful later, tend to be less powerful. For example, cataloguing your bookshelf on LibraryThing would take a while (unless you have a barcode scanner!), and even sites like Upcoming and Delicious can become a pain to keep up-to-date.

    This doesn't mean that this isn't doable, it's just that the act of putting in your interests/exhibitions you've been to needs to be as easy as possible.

    Dopplr allows you to forward your airline e-mail receipt to them, for instance. Perhaps the same could be done for exhibitions?

    Or design the service so that entering your data becomes a social act - inviting someone to visit an exhibition with you on a certain date, for instance.

    Or use Fire Eagle/Google Latitude/etc to track where people are, and automatically detect which museums/exhibitions they've been to.

    There are other alternatives too, but I think these are the key problems to solve. (And I'm up for it!)