I've just come across this other article, which pushes the whole concept of simulation to its extremes: Do we live in a Matrix simulation?

I know it's not 100% relevant, but it's on theme.

Existing software - what's around??

I thought it might be instructive to check out for any existing software.

Jamel (http://p.seppecher.free.fr/jamel/)
This is a Java-based framework for simulating complex monetary economies. Here's a screenshot of the demo running...

Jamel Demo running

All those graphs are updating in real time, and some of the numbers and ranges look pretty interesting! I assume the demo is using a specific set of startup parameters... Have to check out the code. I just downloaded it and clicked on Jamel.jar! It's doing quite a bit - the Mac's fan has cranked up....

The beginnings...

The big problem with politics and economics is that there is no acceptable means of testing out alternative means of behaviour. Further, the existing maths-based economic models are neither tested against reality, or apparently capable of predicting economic events that actually occur.

Here's an email exchange between Giuseppe and me on the subject:

On Mon, Sep 19, 2016 at 10:35 PM, Giuseppe Mura <giuseppe@muraonline.com> wrote:

Now, how about this as a project to entertain us: using graph databases to create an agent based model, which would describe Homo sapiens  rather than bullshit homo economicus...


It’s time to junk the flawed economic models that make the world a dangerous place
http://www.theguardian.com/commentisfree/2016/sep/19/its-time-to-junk-the-flawed-economic-models-that-make-the-world-a-dangerous-place?CMP=Share_iOSApp_Other

On 19 Sep 2016, at 22:46, John Gouk <johngouk@googlemail.com> wrote:
Hmm. Agree with the proposal (change the modelling approach), but not sure how graph databases help... Unless the are a means of representing the connections between the simulated entities i.e. people??

On 19 Sep 2016, at 23:35, Giuseppe Mura <giuseppe@muraonline.com> wrote:

Yep, graph database to hold the network of connections, with the nodes being people. Nodes can have a umber of characteristics, personal traits, wealth, etc. And they have relationships with other individuals. And those relationships are of different kinds, just like in real life: partners, children, friends, colleagues.

But then of course that's just the data, we would then need to make things happen, and that would probably require some algorithms to change the state of the nodes or their relationships. 

We could have it quite simple, few variables, but not so few as one (maximising profit, whatever the fuck that means) and then play different scenarios, see what happens.

For example, we could define a set of basic resources (food, water, energy) and see what happens when we tweak some of the characteristics of our population: can we make resources disappear faster/last longer based on some specific characters being more present than others?

What would impact more the distribution of resources? Specific human characteristics or the strength/number of connections?

I haven't even begun to think of this, it's just that the idea intrigues me, on so many levels: a) I think there must be ways of modelling reality better than those asshole neolib wackos, b) I left off my studies as an undergrad at neural networks,  c) I'd like to learn more about the tech d) you never know we might come up with something interesting from an economic point of view.

On 20 Sep 2016, at 08:09, John Gouk Gmail <johngouk@googlemail.com> wrote:

I did something like this 30+ years ago... My boss (Mike Sidnell - CCd) and I read about people testing the effectiveness of various interpersonal interaction protocols eg tit for tat,tit for two tats,kill them all. Good fun. Not hard to do either. We built a simple programme that started with N algorithms, M copies of each, did an iteration of all against all, then awarded more or fewer copies on the subsequent rounds depending on the outcome of each cycle. It was fascinating, some algos dominated early but ran out of victims and died off, others were sleepers and eventually did well. I recall tit for tat and other more altruistic ones were pretty good, a lesson there I think!

Obvs, the hard part is to determine what parameters are important and what the values should be. One might have to look at some sociology stuff, eek! As well as psychology.

That's cool, yes that's the kind of thing I was thinking of. What did you do with all this? I'd you end up writing some papers on it? Or did you just play and learn?

What did you write the algorithms in?

Well, probably easier if we chat about it sometime. Not sure today's going to work out, but I'll try

And my answers are:

• We did very little apart from enjoy some programming and an interesting outcome
• We wrote no papers, nor saved anything
• Algorithms in maybe Fortran or PL-6 (Honeywell CP-6 programming language)

None of which is very useful. So, let's start from scratch!