Microtask wants to make distributed work a “highly-liquid, tradable commodity that can be easily bought and sold.” They take large volumes of simple tasks that require human labor and distribute them in small chunks to workers around the web. For example, they could take high volumes of hand-written forms and convert them into a database for easy analysis & storage.
If you’re interested in crowdsourcing, you’re probably aware of Amazon’s Mechanical Turk service. Mechanical Turk enables you to create “Human Intelligence Tasks,” or small tasks that require a human’s insights. However, anyone who has used Mechanical Turk knows that there’s a lot of setup and customization required to get the results you want – you basically have to code the tasks from scratch.
This is where Microtask steps in: they make the process of crowdsourcing straightforward, alleviating the burden of setting up the projects. They’re focused specifically on high volumes of repetitive, menial and boring tasks – a category that you don’t want your employees focused on.
Microtask’s design is fantastic, and they’re on a truly “new tech” mission. We’re excited to have them presenting at our March 2011 meetup.
As I watched Microtask’s demo last night, I realized that in my rush to write this post before the meetup, I had neglected to mention the way they get people to complete the tasks. This is part of the genius of Microtask, and how they are able to take on massive projects.
They take chunks of writing or scanned text that normal OCR algorithms are unable to recognize correctly, and insert them into word-based games. As people play the games, they identify the words, allowing Microtask to validate the text.
Here’s an example (showing Finnish text):
We think this approach is brilliant, and hope to see it applied to other areas. We’ll post the video of Microtask’s demo when we have it prepared.
Here’s the video: