The need for accurate and precise polarised parton distribution functions (PDFs) is becoming increasingly
crucial in view of the Electron-Ion Collider experimental program foreseen in the coming
years. Two global PDF determinations at next-to-next-to-leading order accuracy have been
recently presented, MAPPDFpol1.0 and BDSSV24. I review the former and I provide a comparative
discussion of other PDF sets accurate to next-to-leading order. I show that differences between
these PDF sets, due to the choice of experimental and methodological input, exceed differences
due to perturbative accuracy.