The assumption that there exists some price at which the older worker is preferable is assuming that all software workers have a positive contribution and the only task is to value and compensate it roughly in proportion.

My experience suggests that a fair chunk of the population of software developers are incapable of sustainably creating positive business value overall (even if they were paid nothing). If that model is the correct one, part of what you're paying for in a younger applicant is they might be good, whereas with an experienced yet unskilled applicant, they might already be the equivalent of a scratched-off lottery ticket that isn't a winner. (There's no price at which you would buy it.)

Note that I'm well into the age range where I would be discriminated against if age discrimination were rampant. I'm also not claiming that experienced workers are more likely than younger workers to have a negative value. I'm merely claiming that their value is more likely to be determined and less likely to change in the future.