The Knewton Blog

Our monthly newsletter features edtech and product updates, with a healthy dose of fun Knerd news.

Reverse Engineering and Knewton

Posted in Knerds on September 30, 2009 by

When evolutionary biologists encounter a trait in nature, they perform a process known as reverse engineering to understand why that trait existed in the past and continues to exist in the present.

Take, for example, the peacock’s tail.

Evolutionary theory is based on the idea that every adaptation must increase the organism’s reproductive fitness or it would long ago have been bred out of existence. On the face of it, the peacock’s tail poses a problem to the theory. It’s big, heavy and impractical to the point of being downright counterfunctional. The recent theory is that the tail’s very cumbersomeness advertises the peacock’s high level of overall reproductive fitness. The tail announces to peahens, “look at me, I can schlep around all this excess plumage, I must be a pretty impressive peacock.”

Reverse engineering doesn’t just have descriptive power, it has prescriptive power as well. Understanding that the peacock’s tail’s counterfunctionality gives a sexual selection advantage gives biologists insight into other seemingly counterfunctional anatomical and behavior traits. Whenever an animal has flamboyantly, uselessly large horns or throat sacs or frills, we can expect that sexual selection is the underlying adaptive mechanism.

Our approach to curriculum development at Knewton uses the same reverse engineering logic that biologists use on animals. We’re analyzing every GMAT and LSAT question ever published. For each question, we’re asking, what concept does this question test? What strategies does a student need to understand and answer the question? Because standardized tests are intrinsically formulaic and predictable, they lend themselves well to reverse-engineering. We can figure out what concepts the test-makers are addressing on each question and tag them at a high level of granularity.

All of the metadata and tags we’re creating go into our Adaptive Learning Engine (ALE). Using this and other data, the ALE can guide your progress automatically, identifying which concepts trouble you. It can then supply you with more questions or exercises addressing your specific learning needs. By turning test prep into a science, we think we can deliver a better and more effective study experience.