Harddisk encryption

Bruce Schneier is very sceptical about the new hard drive encryption standards released by a group composed of virtually all major storage vendors. He points out that the established software solutions have worked just fine, while this new standards adds complexity (and probably flaws, too) at the hardware level. I'm not so sure myself in which direction the balance goes:

• Encryption (without the key management, which I do not suppose the standard covers) is not that complex to get right and it can easily take advantage of hardware acceleration. (Yes, I don't quote numbers here. Schneier claims he hasn't had problems with encryption speed, but doesn't have numbers either ...)
• It should be a lot easier to subvert a software component than it is to subvert a hardware component.
• It is a lot harder to correct errors in hardware / firmware than in software and get users to apply the updates.
• In many cases the key management is the weak point, not the encryption itself. And key management is hard to get right regardless of where the encryption sits.
• I'm sure the abysmal track record of encrypted drives affects only the cheap USB drives and not the type of hardware we're talking about here. But, can I really be sure? (Serious) software solutions come from companies where security is the main business, whereas the hardware encryption will be implemented by storage companies which so far haven't had to care much about these issues.
• Finally: This standard should replace a number of proprietary solutions since some vendors already do offer encrypted drives anyway. Assuming the standard is sane, this should certainly be a Good Thing™ in any case.