Above: Nick Poole with Mike Lowndes, Mike Ellis and Ross Parry
The talking began with Jeremy introducing some ideas and perspectives from outside the museum world.
Jeremy Keith: (Web 2.0 guru) does anyone from a museum here let their visitors tag their objects? Let visitors add meta data?
Mike Ellis: (Dana Centre and Science Museum) We will be, as of the end of the year.
Jeremy K: Is anyone familiar with the term folksonomy? (many hands go up) OK, most of you. It has scared the hell out of a lot people because it basically means giving up a lot of editorial control – which has traditionally been the way that things like museums and other collections have done it.
It basically means someone else besides yourself settles on a taxonomy, the semantics, the words they use to describe things and folksonomy comes from a completely different direction – whereby it’s the people themselves that decide what to call it.
And the only way to cross the threshold, to get to the wisdom of crowds stage, like the interesting thing on Flickr where 70% is the emergent kind of semantics; all this is words that people are using to tag stuff that we never would’ve thought of, coming from an academic background or our linguistic background.
We never would’ve thought of using this word or this tag to describe this range of objects. We were talking earlier about how do you quantify cultural value or the emotional value of things. It’s very hard to come up with a taxonomy but if you let users tag stuff they may take something as funny or sad, or these terms that generally don’t get used in metadata or museum vocabulary.
In the emergent sense it becomes very, very useful because they know they found something funny in a museum and they can go back to the museum website and search for something funny and find the thing that they found or other things that people found were funny.
But it is scary, it’s a scary thing to do, because essentially you’re giving control out to the people – they are the ones that are trying to find this stuff, that’s what it’s all about. So, I think there is a lot of value there. It’s kind of moving backwards, but the semantic way it is a big vision of what the Web could be.
Going back years now this whole vision of the semantic web existed right from the very start. When Tim Berners-Lee invented the World Wide Web his original vision was for a read-write web, where as well as reading web pages you’d be able to write web pages in situ.
And things like tagging folksonomies and wiki’s, for example, are a little bit of a step towards going back to that original vision of the web, rather than what it has become, almost publication-driven like publishing a book or a magazine, where there is a pyramid structure with a person that publishes and an audience that consumes, where this is much more back and forth where the audience are publishing and adding to what’s out there.
Adding this ability is pretty straightforward in terms of the backend database, it’s not that tricky – you just say we will not even control what’s to add, or how you add it and then see what sticks. It’s a tiny thing – but in the aggregate it works really well.
Nick Poole: (MDA) Can I just briefly comment on folksonomy? There’s been a lot of discussion on this, and about this kind of either/or discussion – do we plump for one approach or the other? The fact that traditional classification is in some sense scientifically objective and that folksonomies are necessarily a mass of subjectivity – I just don’t think they are an either/or.
I think that when a person comes to a collection and applies a tag to it as a curator that’s still a subjective experience. So what you’re really doing is democratizing that process – and you’re not removing the central authority of the curator. You are saying, that if you move up to large scale patterns of use, that either validates the tag that the curator gave. That’s particularly interesting – the extent to which informal, folksonomic approaches validate highly structured taxonomic approaches.
Or, possibly, it invalidates them. It says there is a totally different way, for example with art history questions, there is a non art-historical totally emotional and subjective reaction to this. What you get is large scale, clustered. So I think if we can avoid the either/or juxtaposition which just doesn’t help
Mike Ellis, Science Museum: I think there’s something important at the front end of that as well. Because I can’t see a situation in which we are all going to present just folksonomy stuff. And actually, I have a new take on this which is that actually it isn’t that challenging, that when we started off talking about this it was “oh, we are losing the authority” and actually it’s not. It just simply isn’t the situation.
The front end will always be the curatorial information, object name and the other stuff: and then tags are there as well. People are used to that stuff, they see it on Amazon. They see handwritten reviews, they know how to separate those two different bits of interpretation. They know that this bit is authoritative and this bit possibly isn’t. I don’t think anyone has a problem with that.
Mike Lowndes: there are two places where I think it falls over. So I’d like a robust rejection of the following if possible but let’s see how it goes. One is the long tail. Is everyone familiar with the concept of the long tail? In general terms in web 2.0, it’s about the use of keywords / tags. You have rare keywords and you have small groups of people using those keywords and they are the long tail. What you would normally call the ‘Bell curve’ (of distribution) and it has a really long tail (gestures). These are more specialized keywords (often for more specialised content).
Mike E: the more important thing about a long tail is the stuff down the tail is way bigger than the upstream stuff.
Mike L: the problem is that the number of people tagging is small. Which means in a generic folksonomy context that small numbers of people can generate really bad keywords, and sometimes deliberately. He’s (Mike E) shaking his head already.
Mike E: I’m shaking my head because I don’t think it matters.
Mike L: the less ‘popular’ and more specialized the knowledge that you’re trying to get out on the web, the cruder (the results of) the folksonomy approach is going to be, until you have some kind of unknown threshold of numbers of people aggregating these tags. Now that is why I think the Steve project is interesting, because you have a small number of people tagging so I’m really interested in looking at the Steve project progress to see if it actually generates sensible results. (Note from Mike L: Since the Brighton meeting we found out that the Steve users are selected, i.e. not free public tagging – so its not a good model for testing this issue).
Jeremy Ottevanger: (Museum of London) there are some results out now
Mike L: Well that’s going to be really interesting. That’s maybe one of the solutions to (arbitrary tagging) or maybe one of the challenges for the semantic web: to then use that folksonomy; give it a context by applying, or at least matching up in some way, thesauri and ontologies with the ‘key word long tail’. The ontologies themselves are more formally ‘top down’, if you like, and are from ‘authorities’, whoever they may be. But, in this context it applies a context (to the folksonomy) and then you know that associated keywords like ‘polish’ and ‘shoes’ is about shoes from a certain country or about something else, do you know what I mean?
So, those (the crudity of the long tail and lack of authority) are the two things that I think a slightly worrying for me about folksonomies.