There are lots of people working on this problem. The NSF Office of Cyberinfrastructure has funded quite a few projects working on long-term data storage and discoverability. The largest of those are probably DataOne (http://www.dataone.org/) and Data Conservancy (http://dataconservancy.org/). The hard part is convincing scientists to use the tools that are available. The NSF already requires that all new proposals include a data management plan. I imagine it won't be long before they start requiring projects to deposit their data in a public or eventually-public repository.