The challenge is finding a fair way to compensate talented engineers. Many might do it for free under a "code for science" campaign. But that would require a lot of scientist to realize and then admit how deeply incompetent they are at programing.
I know very few scientists who are terribly pleased with their coding skills, and many who actively admit they're not great coders.
The problem is you can't rely on a "code for science" campaign. A research study isn't something done in a weekend Hackathon, though they are occasionally helpful. What happens if, 2 years into your 5 year study, your helpful volunteers, leaving you with an incomplete code base beyond your skills to maintain or extend?
I know of two ways currently being tried to address this.
The first is to have scientists break their problems down into chunks that can be performed by volunteers but which aren't completely beyond the ability of the scientist to manage the resultant code. This is doable for much of physics and computational biology, less so where a scientist isn't a programmer themselves. We're taking this approach with http://solvers.io.
The second is to have the scientists mentored by programmers to help them become better at it. This is the approach being taken by http://interdisciplinaryprogramming.com.
In both cases, any particular volunteer dropping out is probably not a massive blow. If a project is going to rely on a particular programmer long-term, they probably need to find the funding to pay them.
I'm really interested in solvers.io, and I've been trying to come up with things like that. Small, bite-sized chunks that take just a bit of time, and then everyone can go their separate way.
Thanks! This is our number one priority at the moment. We'll have a new small-task focused design up in a few weeks and are working with project owners to break things down.
