@crowdconf here in San Francisco. A lot of phdâs around here. Brainpower abounding. Interesting mix of academia and corporate representation. Many folks trying to harness all of the power of distributed labor. Not sure if anyone will ever fully own the space, but very interesting to see the adoption across larger segments of the market.
Panel:
- Matt Swanson, Silicon Valley Software Group
- Jason Rollins, Reuters
- Patrick Booher, Autodesk
Question to the panel was âWhat is the one thing you wish you knew when you started dabbling in crowdsourcing or outsourcing?â
Matt : Itâs not just unskilled laborers in a bucket. Employing skilled labor with special expertise (software development, technically skilled, engineering talent) alongside the broader mid-level labor pool in order to make a system work well is a necessary part of the equation, and it has scaled for many companies, small and large alike. Known answers, redundancy, and machine learning. These are the 3 pillars that make a system work in this space.
Jason: be patient. It takes time to put things together. We see sample data and we do experiments with customers.
Patrick: (paraphrased) itâs just a managed service, so we found that we can squeeze crowdsourcing / outsourcing services into our business myopically as it serves our needs specifically. There is no one silver bullet. We had to break a complex task down into many simpler tasks to increase the success rates. There are unique and specific formulated pieces that can be specified in a workflow of a task.
Moderator (crowdflowerâs Zack Kass): Itâs just âworkforce managementâ. We realized that we just needed a technology for workforce management.
Patrick (Autodesk): itâs important to understand the workers. Understanding their performance, mindset, goals, workload, etc. This is critical.
Summation: the process of creating a corporate approach to outsourcing is to first delineate the flow, elaborate on data provision, train the worker specifically for the work, test workers, be ultra specific, ensure proper sourcing methods, refine the process, and the iterate again and again until it is done right.