Michael Hart forwarded an email to me regarding the Toronto Internet Archive and the "addictive" quality of fast feedback loops. (Email text included below.) Further to our conversation a couple of week ago, maybe a process of fast-feedback error correction would look something like this: 1. The HTML version on the PG Canada site could have "live" links to the associated original page. (DP's jhutch's http://www.pgdp.net/phpBB2/viewtopic.php?t=14719 HTML output from TEI puts page numbers along the side of the text.) 2. When a reader spots an error, they click the page link. (They must be logged in, and perhaps have to validate themselves via one of those boxes that display obfuscated letters.) If they validate, it brings up an interface similar to the DP proofing interface (scan above, (HTML) text below). They make the correction. 3. On save, the correction is written the the HTML file but styled in a way to differentiate from the other text. A roll-over pop-up shows the user who made the correction and the original text. Therefore, corrections would be displayed to all readers, but it would be obvious that their post-publication corrections. 4. Editors, working from a dynamic list of pages with "pending" corrections, double-check the corrections, and, if they're okay, port the changes to the source document, update the version history, and re-generate the output versions and the archive file. I like this idea because it tightens the relationship between readers and proofers - it makes it easy for a reader to cross the line into proofing. (And, of course, I like the error-correction aspect very much.) ======================= Email Excerpt:
There are great differences between the Toronto branch of Internet Archive and the others. In the Dictionary of National Biography there are scans that are so faint they can not be read. Others have noted the complete jumble of the indexing system. Authors appear several times with erroneous spellings of their names; often under their given name rather than their family name.
I have on the other hand found the Toronto books to be of very high quality.
This is my impression too. There are some French dictionaries or encyclopedias scanned in Toronto that look so nice I wish I spoke French. I think that the early parts of the Internet Archive's "Million Books Project" have one mistake in common with the early Project Gutenberg, and that is too long feedback loops. One thing we can learn from Wikipedia (and PGDP) is that short (fast) feedback loops are both addictive (so it draws many contributors) and productive (so that it produces high quality contents). You spot an error, you fix it, the new version is published. That is a matter of seconds on Wikipedia. For PG's e-texts, you have to submit an error report by e-mail, and for IA's Million Books you fill out an error report on the web, but then you can only hope that some editor will attend to it. Any means to speed this process up would be a great help. ======================= jen.