Abstract
On traditional information markets, rewards are tied to the occurrence of events external to the market, such as some particular candidate winning an election. For that reason, they can only be used when it is possible to wait for some external event to resolve the market. In cases involving long time-horizons or counterfactual events, this is not an option. Hence, the need for a self-resolving information market, resolved with reference to factors internal to the market itself. In the present paper, we first offer some theoretical reasons for thinking that, since the only thing that can be expected to be salient to all participants on a SRIM is the content of the question bet on, a convention will arise of taking that question at face value, and betting accordingly, in which case trading behaviour on SRIMs can be expected to be identical to that on TIMs. This is the ‘face value’ hypothesis. If this hypothesis holds, SRIMs have the potential of incorporating the accuracy of TIMs while shedding their limitations in relation to long-term predictions and the evaluation of counterfactuals. We then report on a laboratory experiment that demonstrates that trading behaviour can indeed come out highly similar across SRIMs and TIMs. As such, the study can be thought of as an experimental case study on SRIMs. Finally, we discuss some limitations of the study, and also points towards fruitful areas of future research in light of our results.