OPTIMAES


Open Project to Investigate Money and Economic Systems

Comparing Grid Worlds to Fully Connected Populations

:: results

Previously, the OPTIMAES experiment was being run, starting with 100 agents, on a grid. The grid constrains the agents so that each has only four neighbours with which they can interact. This gives relatively few opportunities for bartering or gifting. So few that bartering appeared to produce an effect that was little different from selfish isolation.

Did this mean that barter was wholly useless? Or that there was simply a bug?

To look more closely, the results of all three strategies (selfish isolationism, gifting and barter) were run in a fully connected world (ie. a world where all agents have each other as neighbour). That gives every agent 99 possible partners for gifting or bartering rather than 4.

All other parameters were kept the same. There are still 100 agents in the population and 3 resources. Although this time there are 30 rather than 10 runs of each simulation. The results shown are still the average.

What we see is that, for both barter and gifting, being fully connected seems to have a far greater impact than the strategy itself. While in both worlds, gifting does far better than barter. Both barter and gifting in the fully connected world perform better than either barter or gifting in the narrowly localized cases.

<--
-->