Sebastian Schultheiss talked at BOSC 2010 about the lack of maintenance of bioinformatics web services after the paper that describes them has been published. This also applies to bioinformatics software in general. What happens when the PhD that wrote them moves on? Or the university recycles the server? Or the funding grant doesn't allow for maintenance? How does this affect reproduceability of results?
Sebastian observed that papers with live web services were cited twice as often as papers with services that had gone away so that in itself should serve as encouragement to consider maintenance seriously, but it doesn't answer the question about funding that maintenance.
Maybe a common archive or repository is called for where published bioinformatics software and data can be deposited and looked after by an independent third party. Much like what is currently required for sequencing data supporting a published paper.