Typos. They're killing me. I really think that the typos detection algorithm (or whatever fancy name you gave it) should be more flexible.

For example: if you type 'if' instead of 'of' you'll automatically lose a heart, regardless of how close the letters 'i' and 'o' are on the keyboard, you'll lose a heart because 'if' is an existing word.

I can give a more specific direction about where exactly can we improve it if needed.

October 17, 2014


The big problem with algorythms like this is that where the typo produces a valid, but incorrect, word how can you decide that it is a typo or just an incorrect answer?

There's already this problem as well actually. That's a point to think about while designing the algorithm, And that's why I gave this specific example - you can define letters that are close to one another on the keyboard, Google does it perfectly well.

The issue that you run into is different keyboards. Someone using a US standard keyboard is going to have different typos than someone using a French standard keyboard.

Duo would have to 1) ask users what keyboard they are using, and 2) set different algorithms for acceptable typos across all the different keyboards in all the courses.

This is a great idea!!! Could skip the survey part and just code for all typos of this sort ( adjacency ).

I have the same ANNOYING PROBLEM

Well yea it's generally a bug.

i have notest they are some typos

Learn a language in just 5 minutes a day. For free.