I have been asked about the potential for internal review mechanisms to improve the quality of papers for REF.  Perhaps, it was suggested, have some sort of review just before the paper is submitted, just as some institutions do for grant submissions.

I’m sure that, in principle, this would improve the quality of submissions, and I could imagine doing this for, say, open journal submissions.  However,  in practice for venues with deadlines such as conferences or journal special issues, we will all been remember times when we were still writing the final parts of a paper in the hours running up to 5pm PST.

My suggestion is to
submit as normal (frenetic as ever!).

However,
when (if!) the paper is accepted there is usually a period of a few weeks when final changes can be made.  Use this window for an internal review and especially one that includes non-experts, people within your department, but not n the same core area as you are.

The purpose of the review is two-fold:

• craft abstract
• avoid gotchas

By ‘gotchas’ I mean typos, numbers in tables or graphs that do not agree with this in the text text, equations that in some way look odd.  We have all made these slips or even things that are correct but look wrong, but in fact just need a word of explanation.  Fixing these makes your paper more readable and credible anyway, but in particular during that 15 minute REF reading if anything serves to knock confidence the paper will not get that elusive 4*!