RE: The External Game Theory of Cryptocurrencies
Monero found out the hard way how an adaptable-size blockchain wasn't a panacea. In fact the system was exploitable through a spamming attack - presumably an attack originating from rival Bytecoin
Just a small historical correction here. The reason for the spamming was to raise the block size to trigger a coding bug that occurred at 512 transactions. Only with a larger block size was it possible to have 512 transactions in a block. The first attack that was "fixed" with higher fees was a week or two before the bug-triggering attack and appeared to be a "trial run".
I'm not aware of a purely bloat-oriented spam attack that has occurred on any blockchain with significant transaction fees. The history is that that several coins with absurdly low transaction fees (such as 1 satoshi) have had to raise them or introduce other anti-spam measures, but outside of that, bloating doesn't seem very attractive as a method of attack. As I understand it the Ethereum spam/fork is also not bloat oriented, it aims to DoS nodes processing power because some virtual machine operations are severely underpriced.
I think there are some good external game theory reasons for this, for example if you spam then the coin developers will just introduce more anti-spam measures. You won't be able to successfully spam it all the way up the point where the blockchain is terabytes in size and it causes real harm, even if you pay the fees, so paying the fees doesn't accomplish much. This doesn't entirely apply in the case of Ethereum, because the fork to needed to change fees is going to break some existing contracts, causing real lasting harm.
Good post overall.
Most altcoins aren't even worth attacking so that's not surprising. But there are vectors wide open for doing so. For example, when I transact with Dash I don't even put a tx fee and it has conf within the next block. My "evil" self says "wtf? This is exploitable". As is running sybils to unmask mixing parties due to low darksend mixing costs. For a dash user it doesn't make sense to pay more fees than necessary. From Monero's perspective it makes sense to pay the costs to perform a sybil attack on the mixing parties of DASH to anonymize a few txs and then use it as a negative publicity against DASH. The mixing fees paid (perhaps a few bitcoins over a few days - until some mixing parties with only few-rounds of mixing are identified) would be worth it because then Monero can claim undisputed superiority in the field.
Coins which are "evolving" have both an advantage and disadvantage in that regard. The disadvantage is that evolution is directed by developers and hence a degree of centralization is necessary. The advantage is the quicker response during attacks. More "solidified" coins would be much slower to respond and they could also have a lot of users lagging to catch up with forks and stuff.