Test To tell Computers and Humans Apart

After finishing work at 4D we drove to my parents house to help my father trim the shrubs in their backyard. They have grown to about ten feet high and I did the job with my father’s three legged wooden ladder. On the side of it it says, “This ladder is designed for use in orchards”. I worked my way around the row of bushes and was almost done when I cut into a wasp’s nest. I got stung on my head and on my wrist and almost cut the electric cord jumping off the ladder.

We left there and headed out to Peggi’s mom’s apartment where we had dinner in our favorite restaurant, Le Petite Bistro”. As we sat down to dinner an instrumental, easy listening rendition of “And then I go and spoil it all by saying something stupid like, I love you” was playing. But forget about making reservations here, it is open only to residents and their guests.

Peggi’s mom still gets the Wall Street Journal but barely reads it. I glance at the rabid right wing editorials  and usually find a few interesting articles. Today there was one about the guy who invented the Captcha system (Completely Automated Turing test to tell Computers and Humans Apart). I am really glad to know there is such a test. I get them confused all the time. Clusters of letters are distorted and presented for you to tell the computer what you see in order to prove that you are a human. Sometimes it takes me three tries to get it right so I am contributing my share of the estimated 500,000 hours a day (I had no idea there were that many hours in a day) that people spend solving these inane security clearance issues.

This guy has developed a new version, called ReCaptcha, that puts those hours to good use. Most people have used OCR software. I use this package that came with the free Canon printer/scanner/fax I got  with the last Mac we bought. Today I scanned an old article on Scorgies that Bob Martin’s father left behind when he passed away. Bob photographed the twenty five year old article and I  OCRed it rather than typing it. There were many words that looked like cartoon swearing so I had to go back to the photo to make a human call on what the word was supposed to be.

Google and other companies have been scanning printed books from the pre-computer age and they plan to put them online someday but their OCR software has the same problem as mine especially with books that are over a century old. And paying humans to make all these judgement calls is very expensive. So ReCaptcha funnels scans of the words that the software is stumbling over to the online companies that need the captcha service and it has people like us make the human call on what the word is. Others have already guessed at the same word and if a certain number of people all agree what the word is, they settle the issue. The system doesn’t sound exactly foolproof to me but I love the concept. Our security hassles will be worthwhile for future generations.

1 Comment