2. The ASPI's methodology is flawed. It's based only on citations in the "top 10% of journals", which includes most IEEE or ACM subgroup-esque journals. This is not to say Chinese research is ahead or behind, but the methodology used is just whack as it doesn't take production capacities into account or R&D funding capacity - only publishing. This is why Iran was found to lead in a couple subfields by the ASPI, because the Iranian govt started citation gaming a couple years ago [0], HPE ranking on top for HPC despite IBM and Nvidia being the primary private sector orgs in HPC R&D, and a number of subfields (eg. "distributed ledgers" and "hardware accelerated AI") showing a supposed Chinese lead even though in action some of those subfields are bullshit (distributed ledgers) or R&D is not published due to national security reasons (hardware accelerated AI research @ National Labs)
The one by Nature felt like a more accurate gauge than this one
[0] - https://www.timeshighereducation.com/blog/citation-counting-...