The key issue in my mind is that the coin must not be hard coded to todays tecnhnology. This is the problem with Bitcoin and its 1 MB blocksize limit and even Gavin's proposal is ultimately flawed since he is proposing a hard 8GB blocksize limit. My take is that age has a lot to do with this issue. Young people have a phonomenal grasp of the current technology; however when it comes to the rate of change, or first derivative, of technology the prespective of a 50 year old or a 60 year old actually wins hands down.
How many people on this forum have actually progammed with punch cards, used a teletype to communicate with a computer that had 2 KB of RAM, or got an error message from the mainframe of a major Canadian University because the program they wrote required over 2 MB (the total memory capacity of the mainframe)? I have. To put things into prespective a punchcard holds 80 bytes, and an 8in floppy holds 80 Kilobytes. Now today does it make any significant diffrence in the cost of sending an email that has an 80 byte or 80 kilobyte message? 50 years ago sending an 80 byte message over the telegraph network would cost around 10 USD (in todays dollars), while the other hand sending an 80 kilobyte telegram would have cost around 10,000 USD in todays dollars.
The critical advantage that Monero has is that it does not have a hard coded (requiring a hard fork) limit that throttles the coin to today's technology. This is the critical factor in my opinion.
Now while I am writing this post my sell Bitcoin buy Monero order is being filled. In the meantime let us see if Gavin gets his way and Bitcoin is allowed to scale from the punchcard to the CD 5.25in floppy disk.
Edit: Corrected CD (640 MB) to 5.25in floppy disk (640 KB)
We've got a founder on the project that was from the same background, keeps us very grounded in terms of exactly what you are saying. Frequently he tells stories of punch cards, tape drives and hard discs the size of a refrigerator
Even if he doesn't intend it, it makes me sit back after and think about things I could improve, change, remove to work on lower hardware. Its not a bad practice either, as low end hardware these days is crazy efficient, and so is our project as a product of that thinking
Im working on some JavaCard stuff atm, and thats the same as what you are describing in a way. A few kb of EEPROM, 2k of RAM, slow processor. You have to think about every byte and clock cycle you use. Currently writing an ECDSA signer in that environment (JavaCard 2.2 doesnt support ECDSA with SHA2 hashes natively, only SHA1) and its taxing!