Repeat the dead-column census on the current feature matrix

Determine whether any of the 25 raw features in the current feature matrix are identically zero on the Base, Arbitrum, and Gnosis chains by repeating the dead-column census on the matrix used for the reported transfer results.

Background

The reported transfer results use a current 25-feature input matrix, but the available dead-column census was conducted on an earlier 20-feature input epoch. That earlier analysis found seven features that were identically zero on Base, Arbitrum, and Gnosis; aggregate features were subsequently backfilled on those chains.

Because the census has not been repeated on the current matrix, the paper does not make an effective-dimensionality claim or assert whether dead columns remain. Repeating the census would resolve this specific measurement gap and clarify the effective information content of the features used in the reported experiments.