You're not going to solve a social problem (companies who use GPLed code are afraid of compliance suits) with a technical solution (rewrite a subset of that GPLed code under a liberal license). The problem isn't Busybox. The problem is that companies are either unwilling to comply, or are afraid that their attempts at compliance will be inadequate.

The risk of being sued for accidental infringement is just as real for the kernel as it is for Busybox. Unless you're proposing to replace Linux (which would be a strange thing for the head of CELF to do) then a year or so down the line you're going to be just as vulnerable as you were before. You could spend the time working on a replacement for Busybox, or you could spend the time working out how to convince companies that the only long-term solution is for them to ensure that compliance checking is present at every stage of the procurement and development process.

The first of those choices helps infringers more than it helps compliant companies. The latter helps compliant companies more than it helps infringers. You've chosen the first, and then seem surprised that people are assuming malicious intent. But, whatever. The good news is that it's spurred increased interest in enforcing the kernel's license, which was an entirely predictable outcome.