Holding onto code gives researchers a competitive advantage over other researchers when performing follow up papers. I'm not saying this is a good thing, rather

Holding onto code gives researchers a competitive advantage over other researchers when performing follow up papers. I'm not saying this is a good thing, rather this is something to be aware of. Also, the burden of proof is usually on someone to disprove a paper, rather than proving it, so having someone reimplement it (possibly in another, but similar, manner) allows a fresh pair of eyes at the research topic.