Interview: Ravi Iyer, Ranker on Dealing with Inherent Bias in Crowdsourcing Data

We discuss the challenges of analyzing crowdsourcing data, tools and technologies, competitive landscape, advice, trends, and more.



AR: Q9. Which of the current trends in crowdsourcing appear the most interesting to you? Why?

RI: As a society, we are increasingly solving problems in the physical world, leaving room for crowdsourced answers to questions that are more psychological in nature. We are seeing more interest in our data concerning "why" people have the opinions they have, as opposed to just what those opinions are, in service of our collective move up Maslow's Heirarchy of Needs.
maslows
I see this trend continuing to be our differentiator, as the only platform that provides this level of nuance, and also being of increasing interest to companies that offer similar products, where the only differentiators are going to be how a particular purchase helps a consumer self-actualize. Increasingly, that kind of strategy will be data-driven as well.

AR: Q10. For a data scientist role, what skills would you categorize as "must-have" and which ones as "good-to-have" ?

RI: A lot of the engineering work that data scientists do can be outsourced to engineers who do not have a background in statistics or research, data-scienceso I'd classify them as 'good-to-have'. If your process takes too long to complete and needs to be optimized, the need for a solution is often obvious.

In contrast, if you frame your question wrong, misinterpret your data, or create a faulty research design, there is usually no fall-back within an organization to correct those kinds of errors, so I believe a core understanding of research methodology is the most important skill.

AR: Q11. On a personal note, are there any good books that you’re reading lately, and would like to recommend? great-beanie-baby-bubble

RI: I just finished The Great Beanie Baby Bubble and Bad Paper, two non-fiction books that have a psychological dimension. Since most data is generated by people, understanding people is as important as understanding machine learning algorithms, as the biases in any analysis could involve either.

Related: