I was talking about halving the difficulty after 20 minutes without finding a block, so from (say) 500M to 250M instead of the current 1.
I'm sure you'd have people rushing to submit 20 mins + 1 second blocks they had already solved. *shrugs*
I think removing the 20 minute structure period would prevent all of this.
It's a rush to race against that timer, make the timer go away and the network should flow(?)
I think LoyceV's proposal does not depend on the block timestamp but rather the node's clock, which obviously is going to be even easier to abuse since tempering with clock time is fairly easy and takes no work.
No matter how you look at it, you really can't solve testnet, even my proposal which is the most extreme of them all might not work, some jerks are willing to lose money just to upset someone else.
It feels solvable to me. Just remove the 20 minute stuff and the difficulty that adjusts every 2016 blocks.
Testnet people need a pool that pays for shares found. A CPU will still be able to submit valid shares easily and quickly receive a small amount of Bitcoin Testnet.
I'll never understand the argument "People can just faucet 0.0001 TBTC because that is all that is needed for testing!" but then "We need people to be able to mine a 50 TBTC block with a Windows 98!!!!"
Seem like the best route is to ditch all that stuff, go with a more traditional Bitcoin reset, and let people give that hell in a testing network.
We gain:
1. A Testnet network that is harder to apply gamesmanship.
2. A quick easy way for CPU miners to get a small amount of Testnet.
3. People can't estimate these blocks and then rush their presolved blocks.
*shrugs*