> Finally, the main technical argument given for capped supply (that issuance undermines the security of consensus) is simply not correct.
Here I disagree.
> Enough issuance will eventually drive down the price of every individual unit of cryptocurrency, it does not necessarily end up driving down the total “market cap” of the cryptocurrency.
This assumes that the gain in market cap from increased coin numbers will exceed the loss from per-unit price. I personally believe that, in the long run, changing annual issuance from 2% to 4% will more than halve the market cap, so both the market cap and the total USD value of rewards (and therefore the total USD value of deposits) will go down.
> In fact, as long as the currency has a positive market capitalization, it does not matter how low the price of an individual currency unit goes, it will be possible for there to be positive dollar-valued payments of fees (and issuance), and therefore possible to have economic security for the consensus protocol.
Yes, but we’re not interested in “positive dollar values” of economic security, we’re interested in maximizing economic security. Incidentally I do agree that there exist scenarios where long-run economic security is maximized at nonzero issuance (eg. if txfees are at 0.01% of market cap per year, then 0.09% annual issuance will dectuple economic security without correspondingly large long-run losses from market cap reduction). That said, there is definitely a Laffer curve here, and I actually would guess that even 1-2% annual issuance for security is in the long run on the right side of that curve.
I don’t see fixed-supply guarantees as price-pumping; I see them as presenting to users a clear economic model, where you can lower-bound the value of the coin of a blockchain as the sum of future discounted expected transaction fees minus the cost of running validator nodes (this is assuming PoS is used for consensus; lower bound because there could be a “store of value premium”). Adding in issuance is risky because it makes the economic model much more complex to analyze and reason about.
That said, I will reiterate that I do think it could be wise to wait until full Casper has been running for a while before making any final community decisions on this. Right now there is definitely a trust assumption that “the governance process” will figure out sensible parameters for everything else in Casper (and sharding), so we gain little by armoring against governance failures in setting one parameter but not others; but once the other parameters are roughly set, and we see how the validation economy roughly works, it seems reasonable to start constricting ourselves.