Laser barcode scanners are in a different league from the barcode scanners on smartphones that decode a photo. the bottleneck is how fast you can position the book.
(of course, in his case, he's got to wait for his thing to go look up the price, which may take considerably more time.)
I don't know what this guy did, but what I did (and this would become more useful if I had more clients than just me) was that I cached the data for a certain period of time... so the first time I asked for it within a certain time period it'd actually ask, the second or third time it'd hit a cache.
Now, if I was making software for other people to do this, and I had a bunch of guys like this one using my software to search for books, assuming it went through my db before amazon's, you'd build up a pretty large cache.
Of course, if you could somehow arrange for the whole damn thing to be available offline, that'd be even better, but I don't know how amazon would feel about it. When I was doing it, there were rate-limiting things preventing me; but if you are bigger, there's really no reason for amazon to not give you the data if you are willing to pay for it or what have you; this product is ultimately helpful to amazon.com as a merchant.
I wonder what the economics are here. Using his setup, you can get an Axim X51 on eBay for around $100 and the scanner he uses goes for around $250 on eBay ($384 new). That's at least $350 in initial costs.
Assuming a profit of $5 per book (just a wild guess) you'd need to sell 70 books to recoup that cost. Using his ratio of 1 "buy" to every 30 "reject" books, you'd have to go through 2,100 books to recoup your initial costs!
Taking it a step further, if you could scan three books per minute (another wild guess), it would take 700 minutes, nearly 12 hours, of "hunting" to earn back you startup costs.
Based on all those numbers and assumptions, your hourly earnings are roughly $30 per hour. ($350 / 12)
Based on a standard 2,000 hours of work per year (basically, 5 days a week for 50 weeks), this comes to $30*2000 = $60,000.... awfully close to his estimate of $1,000 per week for a person working alone.