UCL recently hosted their first ever TEDx event, and to my amazement (probably after seeing some seriously dodgy maths themed stand-up I did) they invited me to give a talk.
So if you have a spare ten minutes and fancy watching me ramble on about snooker balls, the solar system, leopards and how they all relate to complexity science then you’ve come to the right place.
Alternatively, use this link.. Or find out more about the event here.
EDIT: I’m totally mind bogglingly amazed and thrilled-thrilled -thrilled that the lovely chaps at TED decided to promote my video as their talk of the day on the 4th August 2012.
Here is the link to the TED page: http://www.ted.com/talks/hannah_fry_is_life_really_that_complex.html
and to my bio on their site: http://www.ted.com/speakers/hannah_fry.html
Hi, do you have a published paper on this? I’m quite interested to understand more about it. Probably the Robbery thing might prove useful in my country [Philippines] (or I may apply it to other implementations similar to that). I’m a Computer Scientist by the way. Thanks
Hi Raf,
Yes – there is a paper. It’s being submitted as we speak so should be online fairly soon. I’ll keep you posted!
¡Gracias! pretty much excited to use this as reference.
Hi, you really have inspired me. I’m studying physics and I was thinking about studying something related with complex systems… you practically convienced me
Do you have any literature you can recommend?
BTW Do you think that after implementing these models the observer will affect the observed lowering the effectivity of the implemented measures, or do the models take into consideration that?
Hi Rodrigo – thank you for your message! There are lots of really good books you can look at, but for starters try these three:
Johnson – Two’s Company, Three Is Complexity. A Simple Guide To The Science Of All Sciences
Mitchell – Complexity: A Guided Tour.
Boccara – Modelling complex systems (although this one is a bit more technical).
This feedback between the models and the people’s behaviour is a big question in complex social systems – one which often gets debated but hasn’t quite made it into the models yet as far as I know. Say, for example, we could predict in real time where the next traffic jam was going to be. If everyone had access to that information, everyone would change their routes, the traffic jam wouldn’t happen, and there would have been no need to avoid that route in the first place. A real time traffic predictor would be a fantastic thing to work towards, but in this example case, the model hasn’t really done anything useful. We are a little far away from this being a problem of course, as the models at the moment are more designed to give us insight and understanding into our social systems than to be used as accurate predictors. There are some groups though who are moving closer though – particularly these guys: http://www.futurict.eu/ who you may want to look into.
Hope that helps! And best of luck with everything.
Thank you Hannah, you’ve been very helpfull and illustrative
Great ted talk great topic, you raised my already great interest in complexity, if this pattern-matching, that is discovering useful similariatis would be automated based on how the brain works you would have some neat AI i think! I can imagine a hierarchy of those kinds of patterns just like the human brain to do what the research team would currently do, anyways very interesting, thanks
Hi! I’m a quality analyst for a game development company but I’m an independent game designer on the side. I’m interested in complexity theory, although I don’t have that much background in math(but I’m willing to read up). Can you suggest any references that I can look up? Read about complexity theory in a game design book(its on the topic of emergence and cellular automata). I’m interested in it because I might be able to use it in some of my designs. Great TED talk btw!
Has this paper been published yet? Can’t seem to find it.
Also what mathematical techniques were used in the models?
Great subject, great examples in your talk. I am working witk the complexty of large organisations an their combination of strategy, business architecture and it architecture and the dynamics of theese areas. You notion of transferring proven models from one knowledge domain to another is a very rewarding thing to do. Keep up the good work and me updated on new results please.
Maneger and Enterprise Architect
Yours Sincerely
Nils Bundgaard
Hi Hannah,
Not sure if this is the best way of contacting you but Luigi Bonomi mentioned your name to me and I was hoping I could tell you a bit more about my company Transparent TV. Do drop me an email and we’ll speak soon I hope. Best, Richard
Hannah’s Research is flawed. Totally flawed. The idea that she can model the riots is a ridiculous. She mentioned the same nonsense at the DeMorgan dinner.
Anyway, I am against her research as well. This is because if she was able to model riots and if it were to show low errors amongst other riots world wide as training data, then this would be worrying? Why? Because this allows governments to suppress protests, especially peaceful ones.
I was not a fan of the 2011 riots, but shooting a man who was unarmed (irrespective of being a criminal and having a gun in the boot) was not on. Also Hannah neglects to realise that the riots were not spontaneous. Media coverage of it revealed it was the former, but my contention is that it is not so.
Why? Because I was there when the riots took place and for all intents and purposes, how does Hanna not know that just one individual started an entire riot in one region alone.
Hannah treats the material (2011) riots as if it was spontaneous and therefore predicting where it would spread to. As I said her data was flawed, because rioting in Hannah’s home town for instance is subjective on wether I wanted to riot or not. If I riot, then many others will follow me. Why? Because I am organised and I know how to cause havok. Thus why some towns did not experience riots and others did not.
If Hannah wanted to model it truly, she should have taken other factors that were not part of the model. Such as ethnic minority bias and inner london community bias.
Anyway, her research is flawed. When I bump into her myself, I’ll tell her as much.
Hi Pedant,
Thank you for your post – I’m very grateful that you took the time to look up my website and watch my TEDx talk after seeing my speech at the De Morgan dinner, and I always welcome a debate about the work.
I wonder though if you have had a chance to actually read our peer-reviewed paper on this work? http://www.nature.com/srep/2013/130221/srep01303/metrics It goes into a lot more detail than the two minute – and deliberately simplified – model overview in the video above.
The paper includes details of how we incorporated the spatial structure of the city into the framework of the model, along with the effects of deprivation (including income, education, unemployment etc) and population density. Our colleague Peter Baudains has done a more in-depth statistical analysis of the dataset (which served to inform a great deal of our work) and included some work on ethnicity. You’re welcome to read his paper too: http://onlinelibrary.wiley.com/doi/10.1111/1745-9125.12004/abstract
The Nature Sci Rep. paper also discusses at length many of the points you raise: how one person could initiate a riot in one region; and the process by which others will join in after a riot has begun. I quite agree the riots were not spontaneous – indeed I believe they were merely a manifestation of a much more serious underlying problem in our cities. This is an important point, and one which I touched on in a spin off project (the result of which you can see here: http://hannahfry.co.uk/2013/02/21/deprivation-youth-services-and-the-london-riots/) but was not the focus of our academic work.
I would also add that this research is not intended to be predictive. Instead, the model is capable of replicating the general patterns seen during the riots – offering insights into the mechanisms at play and an explanation of why some areas of the city were more susceptible than others. It has the potential to help the police understand the implications of their strategies in a theoretical setting, but could not (and would never be able to) predict where the next riot would ‘spread to’.
The London riots were relatively unique in two ways. First, the rioters were not united under a common cause (and hence were relatively easy to deter, as evidenced by the situation being resolved with increased police presence). Second, there was a strong emphasis on looting (and hence rioters tended to congregate at retail sites across the city). These two points are heavily built into the model at every level, and the assumptions break down entirely without them. For that reason, this model could not be used to suppress riots in other countries or situations (Arab spring for instance): it is a very different scenario when deterring otherwise-law-abiding-citizens from stealing a pair of trainers than would be in suppressing a nation of people fighting for their freedom.
Thank you again for your comments and please do stop me if you see me around the university – once you’ve read the paper. I would be more than happy to discuss this work, and the other ideas we’ve been working on.
Best,
Hannah
PS. It is generally considered a common courtesy to use your real name when engaging in debate on a public website.