Quote from: CryptoMF on February 17, 2014, 02:54:55 AM

Quote from: wawaderwaran on February 17, 2014, 01:10:02 AM

Quote from: CryptoMF on February 17, 2014, 01:01:27 AM

Quote from: nrg_wolf on February 17, 2014, 12:28:47 AM

wonder if the price will be pushed down lower after this rape fest.....

No rape fest as they they were not able to get many coins due to the difficulty rising quickly. You can see the hashrate getting back to normal after they gave up. No need to worry.

Are you sure? It didn't go up quite as quickly as it should have to prevent the reaping.

Looking at the block chain: http://explorer.karmacoin.info/chain/Karmacoin it appears you are somewhat correct. The diff didn't quite rise as fast as it could have. This is public knowledge so I feel there is nothing wrong with posting this here.

My block counts per hour are (all times PST for 2/16/2014):

11 to 12 am = 59
12 to 1 pm = 211
1 to 2 pm = 190
2 to 3 pm = 102
3 to 4 pm = 6
4 to 5 pm = 9
5 to 6 pm = 11

So the average over those 7 hours is 83 blocks per hour which is above the 60 per hour ideal. Remember though that whomever put the amount of hashrate that we were seeing at this coin could have made a nice profit on at least a dozen other coins if they didn't choose Karma. The best protection from these attacks is for the price of Karma to rise so more miners are on the coin on a consistent basis to minimize the % hashrate increase. Thankfully this will average out over time and the length of the attack was very short in the grand scheme of things. If you are a consistent miner of Karma then you should see a decent average return as the diff and total network hashrate comes down.

It is good to see we are making progress and I hope we have more to come. I think there is room for improvement here though. Dev's what do you have to say about optimizing the KGW to react faster to massive hashrate changes? Maybe reduce the number of blocks before re-target or change the algorithm for calculating the next black diff to better reflect total network hashrate? Vertcoin as an example seems to have a good system as the difficulty varies very little and the total network hash stays fairly high and constant. Of course Vertcoin uses N-factor Scrypt but that shouldn't make a difference in the KGW implementation.

The trick with reacting faster to massive hashrate changes is if you react too fast you have the problem where a multiminer with decent hash rate can dominate miningThere are a number of coins that have just implemented KGW that still have this issue (Reddcoin was one). See https://forum.megacoin.in/index.php?topic=2791.0

The exploit was they would turn on their miners for about 20 blocks, then turn them off. This caused difficulty to spike (since the reaction was quick), then drop way below where it should have been when they turned the miners off (because the algorithm tried to normalize over 14 blocks- since there was one block with a huge difficulty it drops the subsequent blocks to very low difficulties until it rises again). They turned them back on again and would take advantage of a low diff rate for about 10 blocks, at which point the difficulty came back. Meanwhile normal miners are basically blocked out by competing with a high difficulty or a huge hashrate.

To immunize against this, we changed the settings to almost what Vertcoin (and several others) have, this was the update that kicked in on block 15700 Feb 15th. The difficulty does not adjust as quickly, but we are also not susceptible to the exploit.

We thought about other strategies but given the popularity of the coin, did not want to risk a new algorithm which could cause destabilization. Instead we went with settings that have a good track record.

As always, open to suggestions.