I had always assumed hash rate predictions were already very good, but I also hadn't depended on them directly so I had no reason to question them.
It's possible that large scale miners could be interested in an improved hashrate estimate for their own planning purposes
They'd be able to predict their bitcoin profit more accurately, but have we gotten to this stage yet? I'd guess the price of bitcoin is more volatile than hashrate and that price ends up playing a bigger role in ROI estimates.
Also: TIL lopp writes scripts in PHP like a dgaf chad