I doubt any of us will convince the others of their views. Agreeing to disagree is boring, so I think the best solution is to turn it into a competition. Lets create two alternatives and let the three battle it out. 
But what would we call the alternatives? I think that my alternative would be pretty easy to implement by making some small adjustments to the Bitcoin code. But I'm not sure how Suggester's idea would work from a technical perspective because generating a larger number of blocks per hour makes it harder to propagate those blocks across the swarm.
Alright that sounds interesting. I don't know how fast can propagating blocks go, but let's assume for starters that it takes 10 seconds to propagate a block across the network. There are 8,640 deci-seconds in a day, which means that on average, the system will be able to support something like 2,880 daily blocks (third of 8,640) at a maximum to reasonably avoid collisions. In other words, if we froze the difficulty of generating 1 block so a modern computer would need to work for 1 day on average to create it, it wouldn't be good enough because we hope for much more than 2,880 users to join soon. But what would we call the alternatives? I think that my alternative would be pretty easy to implement by making some small adjustments to the Bitcoin code. But I'm not sure how Suggester's idea would work from a technical perspective because generating a larger number of blocks per hour makes it harder to propagate those blocks across the swarm.The solution would be to make generating a block extremely difficult right from the start. Say, a modern dual-core computer would need to work continuously for a month to generate 1 block regardless of how much CPU power is in the network. With the 10-seconds assumption, this will give us room now to accommodate about 86,400 (which is 2,880 x 30) connected machines without considerable problems. You might complain today that a month of waiting is too much, but assuming the current model succeeds, it will soon need much more than a month to see a block coming under it. Additionally, coin production will be done via specialized businesses and dedicated computers left to work only for that purpose who'll sell to you and me, so a month wouldn't bother them and it would create a killer proof-of-work. And as NLS always fears, a botnet can and will stop all of us from producing under the current model. A botnet will not be able to do so under the suggested model. It will only make its owner rich, which happens anyway.
"What if we listened to your stupid idea, and soon enough we got more than 86,400 connected users?" you may ask. First, please don't resort to name-calling. Second, remember that as time progresses, internet speeds improve faster than computer speeds. So by the time we get 86k PCs connected 24/7 (which may take a couple of years for example), we'll have a propagation time of only 5 seconds. Two more years and it will be 3 seconds, naturally allowing the system to accommodate more newcomers (including a couple of botnets
) without destroying the economic system via perpetual deflation. The network however wouldn't grow too large because Westerns would prefer to buy coins from people who live in cheap-electricity countries to save on their bills, keep reading...Can you explain:
I think that request was for NLS but I'll answer them for my model too 
1 How yours solves the issue of inflation?
Under my model (and indeed, the current one) the generating costs are generally tied to the price of energy. Those may go up and down all the time, but aren't affected by how much people are in the network and don't double every four years either.2 How yours can calculate a meaningful price system?
Prices will be determined in my model (and also in the current model) by how much cost (i.e. electricity mainly) was spent to generate a block of say, 10 BTCs (or whatever).3 Outside of a commodities market, how does yours have a meaningful value for the BTC?
My favorite question . If we assume that running a high-end computer in a Western country continuously for a month costs $10, then under my model, a ฿10 block will more or less have an agreed-upon value of $10 ($1=฿1). As more people from countries with cheap electricity (eg. Venezuela) join to exploit that difference and compete with each other in selling to Westerns, price will more or less stabilize just a tad higher than the average cost of electricity in these countries. This will also serve as a way to limit the network so it wouldn't grow too large for propagation (because why run your PC for a month in the US and pay a $10 bill when you can purchase that block for $4 from an Indian dude while avoiding the melting of your machine -and nervous system- via all that heat and noise?). The network will be filled with generating Mexicans and Zimbabweans (and botnets) because they're most efficient at electricity cost. It's a shame because botnets may eventually become our main providers, but that's better than having them ruin the whole system by stopping participants from producing under the current model, right?Now that Bitcoins would rarely increase in value (because people will continuously find a place where electricity is 3% cheaper), nobody will be hoarding his coins. People will gladly use them as a means of exchange and even spend them ASAP before they lose another 3%, causing the ฿ economy to flourish. Additionally, even 20 years from now, anyone will be able to generate a coin by running his machine for a few weeks, granting newcomers who don't want to purchase from botnets for ethical reasons or from Mexicans for security reasons a chance to join at any point in the future.
I wholly appreciate NLS's model, which is exactly like mine with the difference of freezing the generation at, say, 2,880 daily blocks. Under his model, if one computer is connected, it creates 2880 blocks/day. If 10,000 computers are connected, they create something like 1 block/4 days each. The problem with this is that the cost of generating a block would continue to deflate for the foreseeable future until the number of participating nodes stabilizes, which could never happen as long as the price keeps increasing because of the increasing cost due to more people joining the network (due to more people using the internet and hearing about Bitcoin, for example), inevitably having us stuck in the perpetual-deflation scenario. Additionally, if a couple of large botnets join the network (which will happen sooner or later, to be honest), the average user would then have noway to generate his own blocks and would be forced to buy from them instead. Some people afraid of government witch-hunting those who purchase BTCs might be afraid to use conventional methods like paypal and would thus have almost noway of acquiring coins (both because botnets/crowds make generation difficult and because 16 years have passed effectively making it impossible). Finally, if those botnets/supercomputers belong to a government, they would not sell their 2,879 daily blocks in order to hinder Bitcoins's usability. Ego aside, that's why I prefer my model of user-base-dependent-block-generation over NLS's of perpetually-fixed-block-generation.
Satoshi, please do consider changing the current model to one of those two (or something new). We're not trying to win an argument here; we're trying to find the best way to make this project successful. This becomes more difficult with time. If we're going to change anything before the word catches on, now is the time to do it.
(maybe) 
