Cost of REF doubled to £471 million for 2021 exercise

I wouldn't be surprised if much of the additional expenditure was caused by the open access policy, and specifically the requirement to deposit on acceptance rather than on publication. Very few publishers routinely provided date of acceptance in their metadata when the policy came into force (and many still don't) leaving institutions with no option but to put in place hugely labour-intensive manual processes to gather and verify this data. It’s worth noting that this policy remains in force…

This works out at over £2000 per academic in the UK. Why not scrap the REF and instead give that directly to academics to spend on their research? I'm in a low-cost research subject, so just the £300 per year spent submitting me to the REF would be plenty for all the research funding I would ever want!