In light of the several talks that we've had recently about scoring essays and other free text response sections on tests, I came across this example of a poorly written essay that could get awarded a top grade from the E.T.S. e-Rater system.
The accompanying article notes the following flaws in the technology:
1. The e-Rater cannot decipher truth, so a student will not be penalized for writing that the War of 1812 started in 1945.
2. Longer essays (even if they are filled with nonsensical sentences) are awarded higher scores than shorter essays. The e-Rater also prefers long sentences to short.
3. When in doubt, using a big word in place of a smaller word increases one's score.
It's fascinating to see how far technology has evolved so that we can now use an algorithm to measure the quality and development of an argument -- and it's also interesting to see how much further there is to go before a computer scorer can fully replace a human grader.





Recent comments
1 day 1 hour ago
1 day 17 hours ago
2 days 1 hour ago
2 days 5 hours ago
2 days 5 hours ago
2 days 5 hours ago
2 days 6 hours ago
4 days 22 hours ago
4 days 23 hours ago
5 days 1 hour ago