I think one of the issues with reward weight is how badly it fails at comparison of content from two different time periods. Times when the token was $2 vs the last several months of 12 cents.
It doesn't seem like the reward number can at all be used to sift through and compare content from separate years. Perhaps if we had voting share data instead of final reward amounts that depends on speculative market prices that would be better.
How are you approaching this issue?
Peakd.com is excited to work on totally revamping our search UI and happy that you have a head start on what you're doing while we do a few other things. Happy to see its already getting better.
We are thinking to remove reward weighted ranking or make it decaying with time so it won't affect or put higher weight on ranking of content for specific keywords. Also by default more fresh content should have higher value than older content due to outdated information/technology/feedback etc. It is tricky to find a balance, so that's why idea is to try different variations and analyse results, iterate.
Is it at all possible to get that vote shares data (or whatever they call it)... i'm pretty sure @asgarth told me it wasn't possible and that the API only gives us this reward amount. If so then the only solution is a price history for every day and then that gets WAY to complicated.
Rewards from any timeframe is taken in HBD form so there is not much of difference really, unless I am not understanding question properly. Hivesearcher doesn't store/index vote data, only post data (number of votes, of course part of it but that's it).