A recent Wall Street Journal story described the tools and techniques
used by the Internet Archive to scan out-of-copyright books:

http://online.wsj.com/public/article/SB113111987803688478-2GrTWaqO20C8bl6vQd0qMe6_7qk_20061110.html?mod=blogs

The question is: What tools and techniques are being used by even
larger scale book digitizing projects, such as those of Google, Yahoo,
and Amazon?

I am particularly interested in: 
(1) The hardware used to do the scanning.
(2) The software used to perform OCR on the scanned text.
(3) The resulting sccuracy of the processed, searchable text. E.g.,
One incorrect character for every "N" characters scanned.
(4) The cost per page processed.

(Note: The WSJ article did not address items #2 or #3.)

---

Clarification of Question by ssturgis-ga on 14 Nov 2005 12:29 PST

This company makes a robotic book scanning machine that can scan and
perform OCR of books in a hands-off fashion:

http://www.kirtas-tech.com/

The system uses OCR software from this company:

http://www.abbyy.com/finereader_ocr/

Looks like a wonderful tool, but apparently costs $150,000.