REGARDING THE NFL API DATAMAGEDDON (thread):
I& #39;m obviously not "in the know" here, but from my cursory digging around:
I don& #39;t think the "easy" NFL feeds are coming back. It& #39;s gonna be a lot more like whoscored, which is to say, a huge PITA (but not strictly impossible) (1/n)
I& #39;m obviously not "in the know" here, but from my cursory digging around:
I don& #39;t think the "easy" NFL feeds are coming back. It& #39;s gonna be a lot more like whoscored, which is to say, a huge PITA (but not strictly impossible) (1/n)
This is especially frustrating for me, as I& #39;d spent a [sigh] fair amount of time over the past weeks documenting, categorizing, and otherwise "linking" a number of the NFL endpoints, in service of basically building a historical @seanlahman database, but for football (2/n)
As an educator, given how instrumental Sean& #39;s DB has been as a first touch in sports data analysis for so many, the idea of having something similar for football -- even if it& #39;s not as novel as dots or even event data -- was really enticing. That may be, uh, "delayed". (3/n)
For me personally, I& #39;d struggled with what to do with what I& #39;d learned about the APIs. I& #39;m 100% a proponent for OSS, and was eager to share what I& #39;d learned. But I feared incentivizing dozens/hundreds (more?) of people to hammer those endpoints without remorse. Whelp. (4/n)
I still hold out hope that perhaps this will go back to "normal", and regardless, there& #39;s *always* a way. But the reality is, who& #39;s signing up for implement named-entity resolution on (e.g.) "J.Clowney" to try to link it to an ESPN user ID? It& #39;s not "hard" per se, but.... (5/n)
Some people are of the "Dirty data builds character! This is what the real world is like!" camp, and I absolutely see/understand that viewpoint. But it inarguably hampers "actual" progress, and there are diminishing returns on the level of effort required to *do* something (6/n)