Milestone 3 RATH: Analytics & Algorithms: HIT Feeds and Pricing Engines
How might we utilize data and algorithms to mitigate human challenges?
Analytics & Algorithms
We opted to sort through the needs and discern if any of the needs could be mitigated first through historical data (analytics) and then though dynamic data (algorithms). Can anecdotal distrust lead to validated trust through transparency of data?
- HIT Feeds
HIT Feeds will generate a listing of HITS that the worker is prequalified for based on past worker performance and/or demographics. Machine learning will be utilized to amass an ever refining worker profile of the workers experience and expertise. Much like "Facebook knows you better than your friends and family" the system will know your work better than your last manager (or requester). By continuing to increase compatibility between workers and HITS, requesters will gain trust that their HITS are being completed in a quality manner and workers will be able to more quickly identify HITS for which they are best qualified. It will also allow requesters HITS to be completed in a timely manner due to the feed distribution. Perhaps requesters could pay a premium for HITS to show up in applicable feeds more often. The frequency could also increase at a reasonable rate as HIT deadline approaches. This way even novices (though with a larger standard deviation) will almost immediately have an applicable analytic.
- Pricing Engine
We have the opportunity to introduce a dynamic pricing structure that is more flexible and responsive to market forces.
- Dynamic Pricing: market based compensation structure could be utilized to manage supply and demand of workers and requests. If workers are busy and the market is flooding with HITS requesters would have an opportunity to increase their price in order to get attention, workers would in turn make a premium during high demand times.
- The requester could set a compensation of the HIT that could increase as it approaches deadline.
- We could also introduce an e-bay inspired HIT compensation where workers could bid on HITS or "complete now"at a given rate.
One last possible idea in this realm is to include a followers mechanism. It could be a way to socially validate a requester by following the requester. A requester with more workers following them could be similar to being on the good requester/good hit thread of the current message boards. A worker could have a twitter like feed that shows all the HITS from the requesters the worker follows. There are also possibilities for the requesters to follow workers that have done good work for them. It veers away from the data science but does offer a familiar tool for participants to utilize.
- Workers need to be able to quickly find tasks they'd want to work on
- Workers need to feel they are being fairly compensated for their work.
- Requesters need to get their HITs completed (quickly / correctly)
- Requesters need to have workers who have the appropriate skills and demographics do their tasks
- Requesters need to price their tasks appropriately
- Requesters need workers to trust them