Over at "In the pipeline" he is talking about "hybrid" folks. By this he means those who do both chemistry and biology, and there is a bit of worry about whether or not they will be able to get jobs afterwards. He thinks academia will be more open to these folks than industry. I don't really know enough about this to comment on his example, but I would suggest a seperate example.
Recently, I was asked and answered about a woman who got a degree in bioinformatics and what should she do. I don't think I was overwhelmingly helpful to her, but I do think it points to the exact oppisate problem that Derek is talking about. I think the Bioinformatics degree is TOO specialized and not broadly useful. I think bioinformatics in the context of doing biology is good, but the bioinformatician has to (in my opinion) understand biology. They need to understand it at the bench level. In my view of the world, and I have worked with and managed several bioinformaticians, I have to say that I think the best ones were at one point bench biologists. They just seem to have a better grasp of the real biology behind the scenes. One can do a lot of math on things, but if you don't know what you are really modelling and whether or not that is real then I think you have problems.
Pure Bioinformaticians lack this.
Pure Bench biologists don't, as a general rule, have the programming skills necessary to code this. Most of us also don't have the math to get it done and make do by building a team of biologist/project manager/developer (with the project manager getting the boot a lot of the time). The project manager's job is to "translate" biology to computer and back again.
SO - This is sort of the flip of what Derek is talking about. Here, the bioinformaticians started as the hybrid folks, evolved to a "pure" version, and I think they have more problems now than they did before.