The problem I see with constant-reward inflation is that if we expect the network and transaction volume to grow over time, then we can expect the value to perpetually trend upward and never stabalize (more work, same reward). A fixed percentage block reward, if sufficient, leaves little incentive for transaction witholding attacks and spiraling fees. I actually think the best solution is a variable, "smart" inflation based on volume/hashrate that you and I talked about in the economy thread. However, that brings its own challenges and I feel a 1% debasement to be a reasonable first step.
A finite money supply is the exact cause of the four issues stated above, and constant-reward inflation is just a fancy-pants way of making a finite money supply look like it isn't finite. The significance of that static block reward will perpetually drop until we're right back in the same situation. The block reward must always be proportionally significant, or we will have a problem.
I've spent a night thinking about it and I've come up with a few more thoughts about this.
Both constant reward and constant % inflation will serve the exact same purpose for years to decades.
Consider both of these most likely simulations (provided by drawingthesun):
0.33333333 MRO released every 60 seconds.
http://pastebin.com/GwQJQzjeInflation at 1%
http://pastebin.com/3K0dMnZe1 - For 1% inflation, at year 18: Year: 18.000000 Coin: 21530654.562360 Coin Increase: 213174.797647
2 - For .3 subsidy, at year 20: Year: 20.000000 Coin: 21503999.964960 Inflation: 0.821425%
For 1 and 2, the total coins in circulation have a difference of about 26,655, which is about .123% of the inflated simulation and .124% of the fixed reward. After years 18-20, they begin to diverge and the inflated supply yields many more coins than the constant reward supply. My point here is that this is a long time to make a more informed decision about whether or not a fixed subsidy, fixed inflation, variable of either, or some mix of all of these would be a correct move for the currency. Up until almost 2 and a half decades from right now is the time I am talking about, where both of these routes serve the exact same function (within <<.1% thresholds)
With our most recent update, Taco has set a tx size limit that won't drop the subsidy of more than 9% of the intended value. What this means to me is that a targeted inflation of 1% or 2% may very well have an actual effect of being a .91% or 1.82% inflation. As such, any targeted fixed inflation will yield a window of uncertainty where it will be .91% - 1% or 1.82% - 2% fixed inflation. Before I can be convinced that a constant inflation model is appropriate, that uncertainty would need to be removed. At least with a fixed reward subsidy, we have a solid model for 25 years that people can plan their futures accordingly. At least with the constant reward, we continue to have an uncertain emission only to the day that that is put into effect.
What I'm getting at here is that ... with the rather large amount of time between now and 25 years ... I think we're really just splitting hairs. Unless the constant reward model was subject to the same extra uncertainty about the emission of the coin that will definitely be present in the inflation model, I don't see this as a necessary change that needs to be tackled right this second. I just simply don't think there will be blood in the streets over this in two and a half decades. It's a tough issue definitely, but this is something I have full faith that anyone capable of making decisions in the future will be able to solve. Of course, I'd like only the best to happen after 2.5 decades, but I'm just not convinced that the fixed inflation is more correct than the fixed subsidy right now. The best we can provide them is an open door to either fix the issue or change it into something better,
when if the time comes.
