Friday 25 November 2016

The value of open access

Back in 2012 I published a couple of book chapters on a similar topic: 'large scale agent-based simulations'.  Both were published by Springer. The first was for the Encyclopedia of Complex Systems, and as such set out some definitions and what I thought to be some insightful guidelines on approaching the challenges of large agent-based models. And by 'large' I generally mean models with large numbers of agents and or significant complexity that basically grind standard PCs to a halt...
The other chapter was for a book 'Agent-based models of geographical systems' edited by a good friend and colleague (and now recently professor) Alison Heppenstall at Leeds, with Mike Bithell (Cambridge) as co-author. That went into more depth with examples, especially around the why and how you might parallelise agents vs spatial decomposition across processors when taking a parallel computing approach.

Recently I was invited to submit an updated version of the former (Encyclopedia) chapter to Springer.  So, will I accept?

Now a little further on in my career I don't immediately leap at such opportunities as I did in the past, and I am forced to weigh up the benefits against the ever growing backlog of half written papers I should have published and the list of topics I would like to start work on. So, instead, I ask myself some questions, which are something like this:

- is this going to move my career further towards where I want to be heading?
- Will this be a significant contribution to my field?
- Is this book likely to be well read and cited?

In this case, the answers were:
- well, quite a few methodological papers came out of my PhD, but today I think I need to publish a lot more on the scientific understanding gained from my models as that is where the true value lies in 'modelling' for my career now.
- I think there are some significant advances made in this area since that book chapter, most notably Repast HPC which is designed specifically as a library to facilitate agent based modelling in parallel. I think this still focuses on parallelising agents rather than space, though it does have spatial modelling capabilities so its well worth exploring... However, on the whole I feel my 'guidelines' and suggested approaches still hold up well today, and revising may not add a great deal.
- I actually had no idea how well cited that Encyclopedia chapter was. Turns out, not very: 6 citations according to Google Scholar! Very dissapointing, particularly compared to the contemporary chapter I wrote with Dr Bithell that has been cited a much more respectable 28 times. One factor, likely a big one, is the accessibility. Now I have looked at these again on Google Scholar I realise the latter can easily be downloaded from there, but not the former!

Overall, I think I will decline unless it really will be a quick job to revise. But thinking through my response has been useful, particularly to consider making sure all my publications are as easily accessible as possible so they get read and cited!

If you are interested, here are the two book chapters  to download: