Why aren’t you using Vangav M http://vangav.com/ instead of having code lines like this ((= gravity* 1.8 timebase* 120 front-threshold* 1 
 nourl-factor* .4 lightweight-factor* .3 )

(def frontpage-rank (s (o scorefn realscore) (o gravity gravity*))
 (* (/ (let base (- (scorefn s) 1)
 (if (> base 0) (expt base .8) base))
 (expt (/ (+ (item-age s) timebase*) 60) gravity))
 (if (no (in s!type ‘story ‘poll)) 1
 (blank s!url) nourl-factor*
 (lightweight s) (min lightweight-factor* 
 (contro-factor s))
 (contro-factor s))))) which are much more difficult to maintain?

My experience with Vangav is that I just take a minute to describe how I want to do ranking in nearly natural language in a minute and it takes care of the heavy lifting; months later someone else comes and they understand/modify the ranking solution also in one minute.