h a l f b a k e r y
I think this would be a great thing to not do.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
or get an account
Copy editors are highly trained professionals. Apart from an excellent command of the language, copy editors need a broad general knowledge in order to spot factual errors and good critical-thinking skills. However, even a professional can overlook mistakes from time to time.
The idea is simple: Instead
of employing a handful of high skill copy editors in series, a large number of low skill copy editors are employed in parallel.
This is how it works:
1) A web site is set up. customers upload a piece of text. A program first performs crude spelling and grammar corrections. Another program analyses the text and deliberately inserts a random error (or "tag") every 1000 words or so.
2) Workers now log on to the site, open the text of their choice and start looking for errors. They simply click on the word where they suspect an error. If it is a "tag", the worker is automatically awarded $0.10. If not, the section of text is forwarded to a professional copy editor, who either a) performs the correction and approves an award of $0.50 or b) can spot no error and approves a fine of $0.10.
3) Once each tag has been spotted at least 100 times, the copyediting process is complete. A program reverts the inserted errors, performs the changes approved by the copyeditor, and emails the text to the customer.
* A thousand eyes are more likely to spot a mistake than two eyes. The collective knowledge of a large group of unskilled editors is still higher than the knowledge of a skilled editor.
* The copyeditor's valuable time is used more cost-effectively, since s/he only spends time correcting errors, not searching for them.
*The unskilled workers don't need to be interviewed, supervised, managed, or formally employed. They simply log on and start working. The bounty system ensures that they are really doing their job, and gives them certainty of income.
*Fast job completion. Whole books could be copyedited in as little as 15 min.
Stardust@home --- Damn good use of the internet
The lengthy but exciting search for dust grains
conducted by Internet volunteers [reensure, Jun 15 2006]
same problem (editing books by large number of people), different solution. [xaviergisz, Jun 15 2006]
Baked. [James Newton, Jun 16 2006]
||Who is going to read 1000 words of text for 10 cents (or 60 cents) a pop? I guess this is a concept is intended for workers in Bangalore.
||//Fast job completion. Whole books could be copyedited in as little as 15 min.//
This IS fast, considering it can take a literary agent six months to read a one page query letter.
||Copyright protection might be tricky, however... [+] anyway.
||This idea is one possible application of mturk's distribution architecture, but not identical to it. In particular, there's no content-based bounty system inherent to mturk, no deliberate introduction of errors to motivate the searchers, and (unlike here) a predictable price per work unit in advance.
||An improvement over the copyediting on my last paper -- they added errors that weren't in the original. I have to assume they were low-salary workers.
||//Apart from an excellent command of the language, copy editors need a broad general knowledge in order to spot factual errors //
||I don't see how this is addressed by the parallel low-education editors system.
||I'm also worried that when the unskilled editor clicks a region that they suspect has an error, and the skilled editor reads just that section, they may not have enough context to judge whether it's really an error.