There's a common mistake about $kas that's so common that even experts and academics seem to encounter it quite often (raised here in $nexa's Tailstorm paper). They keep assuming that GHOSTDAG is a variant of GHOST, and then when "comparing" their work to GHOSTDAG, they are actually comparing it to GHOST.

So let me clarify again: GHOST is not GHOSTDAG. In particular, the concept of uncle blocks does not exist in GHOSTDAG, and blocks outside the chosen chain receive fewer rewards. More generally: the "unfair dynamics" of Bitcoin and GHOST (a fancy name for the high variance in block rewards obtained during a week of mining, assuming no selfish mining) do not exist in GHOSTDAG. In fact, I speculate that GHOSTDAG solves these problems better than Tailstorm, because Tailstorm reduces the number of orphan blocks, while GHOSTDAG eliminates orphan blocks completely. I expect simulations will show that GHOSTDAG has lower block reward changes than any other PoW protocol, especially at 10BPS.