It seems that choosing a good factor to use to expand Bitcoin's capacity is a good idea. The software would divide some of its calculated values by this factor, such as the target time between blocks during difficulty calculation, the number of blocks between difficulty adjustments, and the subsidy. To achieve the security of "one hour's worth of hashing power," the recommendation for confirmations would be multiplied by this factor. It seems that using a factor of two from time to time (whenever the community finds it valuable) would be appropriate.
Let's suppose we choose a factor N. This will allow the blockchain itself to grow N times as fast, providing N times the transaction fees to miners, multiplying the number of rewards by N (but dividing the reward itself by N), dividing the average wait for a confirmation by N, and multiplying the rate at which transactions are confirmed by N. This also reduces the variance of mining revenue.
This allows time (instead of network bandwidth and computer memory) to be used to expand the carrying capacity of the bitcoin ecosystem. Should such a proposal be written into a BIP?
The objections to this idea so far have been answered in the comments. If anyone feels that there is an objection adequate enough to drop the idea, please post an answer. If anyone wants to create a BIP, please let me know so I don't have to do it myself. I don't mind doing it myself, but I'll probably be slow.