I’m glad to hear the limits were increased, and it reduces my complaint by half, since they were doubled. There will still be a mis-match between difficulty and hash rate with oscillations between 50 and 150 blocks even if mainnet is huge, and substantial cheating with on-off pulses of hash rate (less than 17 blocks per hit) until the mainnet is big enough to stop it. Unless there is weirdness as a result of the median being required, removing the limits will not cause oscillations.
The difficulty is rising steadily as more people join, and there are no statistical anomalies in the reported timestamps so far. There has not been a 3x attack or time warp attack so far. The reason too many blocks are being issued is because the network is increasing too fast for the 17 block averaging period to adjust. The 16% up per block limit is not the cause. It’s a gentler rise than that, about 0.5% per block. After 4 hours of this, 100 blocks, you can see why it is releasing nearly 2x as many blocks: 1.005^100 = 1.64. [each block is 1-step ahead of the average] If the difficulty ever dropped at the same rate, these would be regained. As the % difficulty rise slows, the excess block release will slow. A cycle of fast ups and slow downs over the long haul would mean the coin will release ahead of schedule.
There was not anything in the time-warp attack that required D=1 of z9. He only needed to be at least 2x the network hashrate. 0.5x the network hashrate should be enough to do it on a smaller scale, but that’s better than him choosing to be a 51% attack. BTW, I wonder if a 51% attack could create new coins inside a falsified ledger without anyone ever noticing (since they would only really are if coins were stolen by being spent without permission).