Monday 12 November 2012

MOOCs are not the only disruptors

As wellas writing about MOOCs and their potential to change the whole university experience, there's another disruptor out there - data publication.

No, really!

Up to now scientific publication has followed a nineteenth century model. Write a paper, send it to a learned journal, who send it to some other people established in the field to make sure it's not complete bollocks and who then publish it and retain copyright, meaning that people then have to buy the journal to read the content.

Academic journals are not cheap - several thousand dollars for an annual subscription. Basically universities and research institutions have to have them,  and yet the content has been produced by their own researchers.

The model has been very successful. The only problem has been that the multiplicity of journals has meant that research libraries can increasingly no longer afford them. The answer to this was bibliometrics - identify the journals with the most widely cited papers and buy them - that way you were getting best bang for your buck.

So less impressive journals withered and died and the more prestigious journals found they could ramp up their prices and still people wold buy them as they were the 'must haves' of the scientific publication scene. You also got strange aberrations where researchers were ranked by the number of papers published in high ranked journals - 5 points for a paper in the Journal of Very Important Things but only one point for a publication in Proceedings of the Bumpoo classical society irrespective of the worth of the actual paper

Things like open access publication represent only a partial fix - basically the researcher pays for the publication process and the validity checking. That way the content stays free or very cheap and libraries can afford them.

But they are still recognisably journals as we have grown to love them.

Possibly this is an important thing, but we are also beginning to see the emergence of a journal free publication model:

Sites like arxiv.org have shown that all digital publication is very cheap. Sure all the researchers do the work, but importantly the validity checking is done by the market place. If your research stands up people will cite it. If it's defective they won't. Basically a market driven model whereby what's good is cited and what's not isn't.

Arxiv.org is just one site. Reputedly it started on a server under someone's desk. The important thing about data publication is that it's building a web of trust and the infrastructure to let you identify and follow up with individual researchers.

Apply it to the arxiv.org model and you end up with a working model for journalless publication in the societies.

Interestingly we sort of have this in the humanities where researchers are quite often  ranked on how well their books are received and how often they are invited to give seminars - effectively post publication peer review ...

2 comments:

Anonymous said...

Over at In the Medieval Middle, where open access is warmly greeted, there has been occasional talk of a kind of digital commons model for publishing, whereby something one writes goes to the cloud/server and searches pull it out, and what ranking there is is based on popularity. I like some aspects of this--and with sites like academia.edu where works that probably shouldn't be are still shared anyway, without any distinction between unpublished, pre-published or published necessary, we start to see this I think--but there are problems. One is what happens to the guy with poor grasp of search terms; but maybe, to be fair, a badly-written piece is less actually useful. A secondary implication of the relation to searchability: things that no-one knows about yet will suffer a timelag before being picked up and understood, because relevant phrases are unlikely to occur. I mean, if someone learns how to synthesize a new material called Wellineverdidium, obviously no-one will search for that, though its unique and useful properties might still bring it hits I suppose. Tagging and such gets round this to an extent on academia.edu, so maybe could here too. Thirdly, though, is expertise: can what's popular necessarily provide a clue to what's valid, in fields where significant background knowledge would be needed to make a judgement? Academia already has this problem, of course, and it largely gets round it by reputation; that's something web fora have been doing for a long time too, so again the models already exist. (Academia.edu doesn't do this, except by number-of-followers metrics, and just as well, as I hate using it as an a successful paradigm example when it really robbed most of its functionality from Facebook.) I guess that perhaps it's all surmountable, but still, I can't get rid of the memory of this XKCD cartoon: http://xkcd.com/937/. Sometimes it really does matter whether anyone has the right basis to evaluate something...

dgm said...

The world's imperfect, and we still need a reputation based model - which is why I go for citation counts by analogy with software development - good and/or useful code is reused, code which is not re used is either obscure or poorly structured.

But yes - reputation is a peer based thing - you need to be enough of an expert in order to assess worth properly - otherwise the peseudo academic von Danikens of the world would overwhelm us with plausible sounding rubbish