Of course, you need to count the received shares, but in practice the difference is not big.
The problem can also be in the case of excessive overclocking and the mining farm is unstable.
All important points but as you say the difference is not big, and it's not big enough to be significant.
But it's being used to claim dishonesty in hash reporting when there are many other factors that could cause
bigger differences.
The history is a feud between 2 closed source miners and claims of which was the fastest. One of the miner's
devs decided to prove it with a challenge and a test tool. The feud seemed to have died down but this may be
an attempt to revive it.
For example, I use T-Rex, I learned how to fine tune this miner for my AMD video cards.
No article will convince me to change the miner, because even if I lose 1-3% of the shares, this will not greatly affect my profit.
Miners do not pay attention to such articles, because they understand that identical video cards can have different hash rates and different overclocking options.
"For example, I use T-Rex, I learned how to fine tune this miner for my AMD video cards." Are you payed from T-Rex?? 1st time I read T-Rex in AMD ...