Constructive Competition Simulation vs. Regret

If rather than plurality's {+1, +0, +0, ...} weighted scores without duplication of non-zero values, agents would instead compete for {+1, +0.5, -0.5, 0, 0, ...} allocations, how would the signaling of the agents and scores from the audience shift in response to the change?

Don't make this partisan, treat it as pure math (game theory).

Consider the fact that plurality restricts choices to the situational inputs of blind cooperation or competition, whereas the partial weights can be applied directly to the outcomes of situations which create nash equlibria, thereby selecting win-win rather than risking asymmetry.

Note that every distribution of scores could be enumerated and is thus a "single vote" regardless of format. The "less flexible" allocation (finite non-zero values) compared to Approval/Range maintains constructive partisanship and the nursery effect rather than creating a monolithic purity test over the course of many iterations.

Analysis of any code, simulation output, or visuals encouraged.