I have quite a bit of BTC left at Nicehash that they are slowing releasing after the hack (that was a nice surprise as I just assumed that .5 BTC was long gone but so far they seem to be living up to their promise (well, 10pct so far)). Once the next round drops I will go give it another try. Will probably just test the SHA then cancel and use another Algo as have enough SHA for now. Particularly on SHA on NH, I hate having to constantly go back and see if prices have moved and my contract isn't live anymore. And the fixed contracts are pretty expensive. Just feels like I am getting gamed. MRR just is more predicable - and typically cheaper. A lot of people love NH, an it is a little easier to use.
On the L3 point - I reverted back to vardiff and it seems fine. Not sure you changed anything but it now seems to be moving it between 120 and 240K nicely. Previously it just seemed like everything good pegged at 240K on vardiff. Have 4 pointed at you and they are all between 120 and 160K at moment. 3 are overclocked to 550Mhz, one is not. I may move another 3 or 4 over next weekend as returns look good.
A few days ago we tried using Nicehash for sha rentals again and have confirmed that one of our previous plans is a good option. Basically we want to be able to support a broad array of devices. People who still have miners operating in the GH/s range need a lower difficulty option and those using S9s or as high as Nicehash demands need a separate option so each category can run smoothly. I know we have been testing what would work best for everyone and we want to go through some further testing before making anything public. Making sure earnings are properly recorded is a big thing to ensure before making this live. Occasionally some of the sha restarts have been to apply some of these changes to see how machines adjust or improve with changes based on pool hashrate and machine type.
We are going through each algo and looking at what can be improved, many cryptos have increased in difficulty or decreased in profitability so we will continue monitoring each option as efficiently as possible. After many years in mining, I know how important it is to maximize your miner's efforts for time spent. There are a lot of things in the works that we think will help with that.
On your note about scrypt, we haven't adjusted anything difficulty-wise on our end, but the available selection of what cryptos the profit-switching is trying to find has a bit to do with what var diff is selecting. Certain coins just have better performance in job relay and we try to adjust depending on availability.