I think that assumes we're in a state where people are buying coins for their supposed use. But at the moment, demand is mostly speculative.
There's also a very crowded market for coins right now. Most will not take off. Even if coin X has perfect fundamentals, there's a self-fulfilling equilibrium where nobody buys it but instead buys coin Y which is slightly worse.
My argument is that in the long run, a very inflationary coin would not be attractive. Using it for transactions would rely on greater fool theory. You buy it so you can sell it immediately. But that means you must sell it to either (a) someone who holds it and loses half his money, or (b) someone who buys it and sells it immediately, again to either (a) or (b)....
The intent is to be outcompeted as an investment vehicle. The goal is approaching zero investment.
I'm thinking that inflation would be priced in as platform fees for exchanges. Thinking of transactions in terms of throughput, the question is "how long do I have to store EUR in INF to be able to match an EUR sell with a buy." The transaction fee would be (at least) the inflation over that range.
The notion that blockchains should produce value by competing for investment is exactly the wrong idea that I'm trying to avoid.
An inflation half-life of a week is probably excessive. It really only has to be bigger than the IRL currencies it interfaces with. (Which in turn only have to be bigger than the value growth in their market.) But proposing a high inflation like that puts the focus of efficiency and monetary gain on improving transaction performance, which is imo where it should be.
Also having a high inflation lets you limit the size of the ledger - eventually, old transactions simply become irrelevant by being too comparatively small to matter. You could even reward people for erasing their historic transactions - ie. allow them to roll several historic transactions into one, as a way to reduce the inflation tax. (Or phrased differently, tax ledger entries.)
I understand that you don't want people to buy it for investment. My point is that even buying a coin for transactions relies on having someone to sell it to later. Take an equilibrium in which everybody uses InflationCoin for transactions, paying the cost implicit in the inflation rate. Why would somebody not offer FixedCoin as an alternative, identical except for having lower inflation and hence a smaller cost? Why would FixedCoin usage not then spread to replace VanishingCoin?
Well, the idea is people would use LessInflatyCoin as long as the inflation was still high enough to make it unviable as an investment vehicle. But once inflation was high enough to make the coin usable as an investment vehicle, it would become a worse unit of transaction because you'd expose yourself to Bitcoin's attendant bubble risk. For instance, if you were to sell something for FixedCoin 1mil, you might have a hard time predicting how much your FC 1mil were worth tomorrow, or in a week, and your incentive would be to keep the money and gamble that it goes up, whereas IC's value development is highly predictable (ie. just slightly negative) because it's only driven by current use. I'd expect this to make IC more appealing to "boring" corporations.
There's also a very crowded market for coins right now. Most will not take off. Even if coin X has perfect fundamentals, there's a self-fulfilling equilibrium where nobody buys it but instead buys coin Y which is slightly worse.
My argument is that in the long run, a very inflationary coin would not be attractive. Using it for transactions would rely on greater fool theory. You buy it so you can sell it immediately. But that means you must sell it to either (a) someone who holds it and loses half his money, or (b) someone who buys it and sells it immediately, again to either (a) or (b)....