I’ve got it let’s invade Iraq!
Download 2.43 Mb.
|
- Navigate this page:
- Everett Kaser
- When I last went to post a response, I was commanded to key in
- There are always two groups of characters. If you make a mistake, are you less likely to be bounced if it’s in one group or the other
- Where do the character groups come from
- How is reCaptcha capitalized Or, to put it in language I actually understand, where do they get their money
- We’ve all at one time or another written a brilliant, eloquent, perfectly articulated and very
- Thank you, Everett. Just really fascinating information.
|
The Stupid 365 Project, Day 14: GOTCHA, reCaptcha! October 14th, 2010 
Many of you, like me, have wondered about those bizarro wordoids you have to enter before WordPress will accept a comment. So today, we’re going to let someone else write most of the blog in order to explain reCaptcha. I personally think it’s fascinating. Everett Kaser has been kind enough to look at all the Simeon Grist books as they’re formatted as e-books, and he’s pointed out literally thousands of mistakes that Kimberly Hitchens, my long-suffering e-book producer/factotum, has then fixed. A graduate in Arts from Oregon State University, Everett spent 20 years working at Hewlett Packard, where he rose from an entry-level production job to being a software engineer in the R&D lab. After writing many trivial games for fun through the 1980s, he started releasing shareware games for MS-DOS and later MS-Windows (and now Mac OS X). This hobby grew sufficiently until, in 1997, he quit his day job and has been in game-programming heaven ever since. He reads, he plays disc golf, and sometimes (he says) he even sleeps. He’s also one of the few people in my world who understands reCaptcha. So he said I could ask him some questions. When I last went to post a response, I was commanded to key in ores; prepolty. I mean, what the hell? Ores; is perfectly reasonable, even in English. Think of an article about mining containing a compound sentence: ”They were digging for rare earths and metal ores; little was found.” Modern newspapers don’t use the semicolon nearly as much as papers did 100 years ago. prepolty IS a strange one. But strange reCaptcha words can be one of several things. – A word from a non-English language. – The name of a person or place. – The beginning of a hyphenated word (from the end of a line) – The end of a hyphenated word (from the beginning of a line) – A technical term. – Slang. (Think how some of today’s slang will look to people in 100 years.) – An acronym. – An actual mistake in the original document. – The middle of a word. (Nothing says the reCaptcha system has to use entire words. A word may get broken down into several pieces because it’s best that reCaptcha terms be kept to a reasonable length.) – Etc.
This I don’t know for sure, but ONE of them has to be spelled right, and the other one doesn’t matter. But you don’t know which is which, and if I were the one writing the reCaptcha code, I would randomize which one you have to spell right. Otherwise it would eventually be known which was the important one, and people would start screwing with the system, which would defeat the second purpose of reCaptcha. The first purpose is to make sure it’s a human who is making the comment (or whatever it is that reCaptcha is evaluating), not some spamming computer program. It is still VERY difficult to program a computer to figure out what these mutated graphical words say, while a human can do it quite easily. The original system was called a “Captcha” and had only one word or series of characters you had to enter to prove you were human. ”reCaptcha” was invented to do that AND something else, and that brings us to the next question. Where do the character groups come from? The second purpose of the reCaptcha system is to rapidly, cheaply, and easily digitize and proof-read scanned-in documents. So the words in reCaptcha come from documents that have been scanned in. They’re digitizing what? And how does this relate to the zillion other digitizing projects that are underway? Documents, like the entire run of The New York Times from the 1800s up to the present and the hundreds of thousands of books that Google has scanned in from libraries. Scanners and OCR (Optical Character Recognition) software can do a pretty good job of converting printed text to ASCII text computer files, but they’re not perfect and many little errors creep into the text. So every time you correctly enter a Captcha you’ve helped this process by proofreading ONE word, or part of a word, from one of those documents. It sounds like a slow process until you realize that there are tens of millions of those reCaptchas filled out every day on the Web. That’s a lot of documents every day, day after day after day. The way it works is that one of the words in the Captcha is already known to the server that provided the graphic, and the other is being proofread. If you enter the known word correctly, then the system assumes you entered the other one correctly, too. But errors could still creep in, because sometmes you’d luck out and get the known word right while making a mistake on the unknown word. So the system requires the unknown word to be typed in the same way by multiple people before it’s accepted as the correctly proofread version of that word. There are many digitization projects underway, but most of them have no good proofreading associated with them. For example, Google has scanned in hundreds of thousands of books and converted them to ASCII text files, but they also contained many OCR errors. Hiring people to proofread all those volumes manually would be expensive and time-consuming. How is reCaptcha capitalized? Or, to put it in language I actually understand, where do they get their money? Initially, they were probably being paid by The New York Times for digitizing and proofreading the papers. But in September 2009 Google bought the reCaptcha company, and Google is now applying the technology to improve the quality of the digitized books — making them more useful and more easily searched — as well as continuing outside projects, such as The New York Times. Some folks may think that Google is becoming the new Evil Empire, but having available high-quality scanned versions of all those old books is really a good thing and is a huge step forward into our information future. We’ve all at one time or another written a brilliant, eloquent, perfectly articulated and very long comment, and then seen it flushed when we screwed up the reCaptcha. And ideas on how to avoid that or recapture our lost brilliance? The best thing to do is select all the text you entered and copy it to the clipboard before you enter and submit your reCaptcha words. That way if something goes wrong and your submission fails, you can easily paste your brilliant thoughts back into the edit field and try, try again. If you don’t know how to select and choose “copy,” you should learn. If you failed to have the forethought to select and copy your text, you may be able to recover it by selecting the BACK button on your browser. Sometimes this will work and sometimes it won’t, depending on your browser and the implementation of the specific web page you’re viewing.
Directory: ebooks ebooks -> Hcm cộng hòa xã HỘi chủ nghĩa việt nam ebooks -> "That was our view and I believe it right." ebooks -> The Teachings of Don Juan ebooks -> Appendix 9c sa army unit histories1 Infantry ebooks -> 8. 0 Cellular Phone Systems 01 Cellular Voice Reference Model ebooks -> War against the machine accountability meter has expired. Time ebooks -> Reader Description ebooks -> Vedic Astrology, An Introduction by Das Goravani Download 2.43 Mb. Share with your friends:
|
.