Also: “Proof if Weak Hands” was successful in making a game out of spamming the network to the point of DoS.

When ETH was hundreds of dollars, the cost of running code on the blockchain (gas price) could get up to $50 during these attacks.

Imagine adding a column in a database table cost $50. Forget about it.

I created and run a cryptocurrency trading game called Cache (https://www.cachethegame.com) that uses ERC20 tokens in the “coin-op” sense. The solution when transaction fees go wild is to simply enable free play. Nobody wants to play $1+ in fees for something that costs a few cents.

It did become more expensive to use (on the order of 10x), but it didn't go down.

No thats not what happend. Nothing ever went down but if you wanted a tx to go through fast you had to pay more.

I was thinking of this story: https://www.bbc.com/news/technology-42237162 (Which, I understand, is for a general audience and might not be very accurate.)

I recall the founder of CryptoKitties saying that the transaction cost was significantly higher than the kitty cost. Isn't that a failure state for the network?

> Isn't that a failure state for the network?

it depends on what you define as a failure state. Technically, the network is still working - but business-case wise, it failed.