Logic Recognition

For the past two years, I have been teaching C# to blind students. One of the major pains of the internet is image recognition systems. If a site uses image recognition without any alternative, then a blind person must contact tech support, etc. This turns their registration into a week worth of figthing with tech support before they get an account.

Sites that do offer an alternative should be commended, but there is still a problem. Sound recognition is often used as an alternative. However, these are sometimes very difficult to understand. Imagine a person with both vision and hearing loss, it remains difficult and sometimes impossible.

Also, bots are increasing their ability to deceifer the images and sounds and will eventually solve them just as well as a human.

I have thought of a better alternative: Logic recognition.
Basically, create simple logic puzzles that must be completed by the user.
Example:

Enter the third word of the following sentance:
I like to eat eggs in the morning.
Answer: [to]


Obviously this could easily be solved if a bot could be programmed to solve this type of problem. The solution to this is variety: problem types, randomness in the problem, using abbreviations, incorrect grammer, typos, etc.

For example (If you wanted to make it really hard for a bot):

Enter TheWordAfterThe 5th Word of the Following Sentance:
If you are a bot you will enter 'bot', not 'you'.

If a bot were designed for the first type of problem, it would think it should use the 5th word. We actually want the 6th word in this problem.


Another type of problem:

Sort the following words in alphabetical order. Make sure to leave a space between each word:

dog
cat
apple

Answer: [apple cat dog]


Again this is easy for a bot to solve, let's make it harder.

Sort the following words in alphabetical order. Make sure to leave a space between each word. Also, leave out the animal that meows:

dog
cat
apple

Answer: [apple dog]


Image and Vocal recognition are simple compared to the variety of logic puzzles you can create.


Another great benefit of this, is that the problem generation is processor fiendly. It is simple string manipulation. On the other hand, image and sound generation would need a dedicated server for large sites.

Also, the sentances and words used in the puzzles can be placed in an xml document. The xml document can be customized for the site to include slogans and other advertizing propaganda.


I hope this will become widely used. We are going to implement this on a site for which I program. I will add a link when it is done.