May I ask what method did you use to test it and establish that the claimed 2.8% fee is accurate?
We ran Vnish on about 3PH worth of hashrate the last summer, and we tested the fees against the monthly reports, and we took the mining into account (which you probably failed to do) it's safe to assume that while running custom firmware especially if it's on auto-tune it won't be 100% online, so while you do lose a certain % of your total hashrate it does not go to the dev's pool, it is just lost.
You can use AwesomeMiner or manually collect the information, it's highly unlikely that you will find a miner with 100% mining/uptime on your primary pool, below is a screenshot of the 30 days report on one of my miners
This miner is underclocked, runs cool all the time, and looks perfectly fine, but for some reasons, it failed to stay connected to the pool 100%, which is pretty normal.
We have also done some packet sniffing and checked the bandwidth to dev vs main pool, you can perform that yourself, you will notice that while your main pool stays connected the transmitting of data does change when dev fees take place, this does not show on the mining graph (for most pools) probably because they use XNSUB.
I also believe you can use the DiffA to get a somehow accurate guess about the fees and here is an example
The last two rows belong to the dev's pool if you add them together you get 56,198,434, compare that against the total which is 1,930,069,282 and you get exactly 2.9117%, keep in mind that this is AwesomeMiner's version with license included so it's 3%.
Of course, the second method could be off by an order of magnitude if the dev is displaying false figures out of the box, but Vnish and Vnish distributors like Marc and Patrick from AwoeosmeMiner are long-standing community members, Patrick probably makes millions from his monitoring software, it would be a completely stupid move to try and lie about the fees he collects especially that at least half of it goes to Vnish in the first place.
Of course, this is just my point of view, I am not saying my tests were 100% accurate and that my way is the only right way of doing this, I could be wrong after all.
I don't name anyone, because I don't want to go to war with anyone, and I don't want to be called a detractor to put my firmware ahead of others and say that I'm the best because it's not the case...
This is not a valid excuse, forgive me for my honesty, you could always use a throwaway account to expose scammers, if you are afraid just send me the findings and I'll post them, if anyone can provide solid proof which the community agrees with and accepts as an accurate test, I won't hesitate to tag every firmware dev involved.
