Satoshi had no limit on block sizes at all ... There was a 33.5MB limit on message length and since blocks are transmitted as a single message it would have limited blocks to only 33.5MB but even this wasn't a hard limit as new message type could have been added which transmitted blocks in other ways (i.e. header & txn hashes vs full transactions).
The 1MB "limit" was added as a temporary anti-spam measure 18 months later ... the commit wasn't made by Satoshi.
Yep. Good enough for me. I'm of a mindset that you "don't change the original design unless there is hard evidence of a need". I'm against Gavin's plan simply because the numbers are arbitrary, and I don't like arbitrary. I'm in favor of rolling back to the original message length limitation and letting the miners determine block length based on their internal cost/benefit analysis using fees and orphan probability.
I don't want a "new" blocksize; I want the old blocksize back. When that gets full, we can discuss it further, and it'll be for the same reason that one day, we'll need to extend the decimal precision to allow division of satoshis.