Even if it's a small percentage of calls that have problems, a few annecdotes I've seen may suggest that those drops are concentrated on certain phones (or in certain areas): so most people get AT&Ts usual connection reliability, while a few get much worse. Of course, forum postings aren't evidence of how widespread the problem is, but I'd rather see Apple pay needless attention to a NON-issue (and address the few problem cases individually) than overlook a real problem!
Note, according to Daring Fireball, "Nomura analyst Richard Windsor," who started this media storm (and seems to be the source of fears that only an internal hardware change might help), is the same guy who started a blatantly fictitious report of hardware problems with the FIRST iPhone: an "increasing" wave of dead spots on Apple touch screens.
Like Windsor's 3G/Infineon report, his iPhone scare last year was packed with convincing-sounding technical and business details--which were entirely false. He said that iPhones use a chemical film on top of the glass that senses heat to detect touch, using technology from a Finnish company,and that this film breaks down several months after purchase. Awfully detailed and convincing! Better not buy an iPhone!
Of course, iPhones have NOTHING on top of the glass to break down, and don't use heat at all, and it takes no effort at all to uncover those facts. Windsor's FUD re touchscreens had to have been intentional--it's hard to imagine mere incompetence leading to such detailed false info.
See:
http://digitaldaily.allthingsd.com/20070810/iphone-deadspot/
So, I take Windsor's latest claim with salt. A problem exists, at least for some small number. How many? Who knows--it's only natural to post when you have a problem and not when you don't. But fears that it needs a hardware change seem to come from Windsor, so that's the detail should not be taken too seriously without more evidence.
I hold out hope for a software fix, and will watch the issue with interest before I buy my 3G.