STUFF THAT NEEDS DOING

[thl19git]

• [ ] Our loot and fight proposals are very basic
• [ ] As leader if we have the fight or loot imposition, we just go with the most voted for proposal and leave it unchanged
• [ ] We don't trade, therefore generosity SC value is not updated
• [ ] Leadership SC value isn't updated
• [ ] Very little non-determinism so all the agents do the same thing
• [ ] We don't message other agents
• [x] Can someone (@Razvan00Rusu ) check that the current implementations of HandleLootInformation and HandleFightInformation are okay? - i.e. am I right in assuming that all agents are sent a StartLoot or StartFight message at the beginning of the discussion phase?
• [ ] Collective Intelligence and Group Think, what needs implementing @Issa-Bqain?
• [ ] REPORT @catastretcu @1rre @Issa-Bqain @Razvan00Rusu obvs needs to be done by about 14:00 tomorrow
• [ ] Experiments - our agents die every time. What experiments can we run, how can we make our agents better?

[Razvan00Rusu]

Sorry for getting to this so late, HandleLootInfo and HandleFightInfo look good to me - will start adressing some of the other points here. Will look at the non-determinism point now

[Razvan00Rusu]

Additionally, do we have an overleaf link for a report or something 👀?

[Issa-Bqain]

Overleaf link on

[Issa-Bqain]

I will finish writing up the CI stuff for report by 2-5am.

for the CI implementation, it should be very simple. We can just calculate one metric, which is just a combination of the other social capital values.

If this value is higher for any other agent than us by say 30%, then we should elect them as leader.

we dont have to use the co-efficients, since we don't have learning etc.

[Issa-Bqain]

Interms of messaging other agents, the CI idea actually tries to minimise communication and so if not implemented, it fits within this way of thought, shouldn't be an issue.