Crowd labor is outsourced information work that can be provisioned automatically. It is used to perform intelligent tasks that machines can’t do. Innovative applications can be built by interleaving crowd labor and machine labor. Ideally, crowd labor is inexpensive, on demand, and elastic. This report examines how crowd labor platforms — which have not been commoditized — are shaping up and competing with one another.
Key findings in this report include:
- The key disruptive opportunity that crowd labor platforms can exploit and improve is work quality. Few quality-control methods are implemented in Mechanical Turk, for example. For this reason, Turk has been plagued by low-quality results, leaving the customer responsible for ensuring correct results. However, the need to manually verify results is not scalable.
- Another critical disruption vector is business process outsourcing (BPO). Platform suppliers will leverage BPO and work quality for the most competitive advantage. Other opportunities include building out crowd labor marketplaces and their accompanying business models; taking advantage of application specialization, especially for the long tail of underserved common tasks; and offering specialized labor pools.
- Similar to the cloud, crowd labor has a stack: Applications provide a distinct service like transcription or sentiment analysis; platforms provide quality results from the crowd; and Amazon and other crowds provide infrastructure. Currently the primary limitation in the crowd labor marketplace is the immaturity of the platform layer. There are no robust crowd labor platforms with high-quality labor and native workflow support to third-party crowd labor application developers on a self-serve basis. For this reason, crowd labor application development requires vertical integration with the platform layer.
- Number indicates company’s relative strength across all vectors
- Size of ball indicates company’s relative strength along individual vector