I have reviewed many papers and there is never the time to recreate the work and test. That is why I love the "papers w code" site. I think every published CS paper should require a git repo with all their code and experimental data.