Removing code doesn't really solve the problem

Removing code doesn't really solve the problem

corbet's right that people have talked about replacing code a lot, and not just regarding McHardy. Some have even proposed there should be a "litmus test" about enforcement views before accepting any contribution to Linux and refusing to permit code in canonical versions unless each contributor agrees not to enforce. It's a scary prospect to think about some sort of purity test on enforcement before merging contributions. That's just a form of CLA, so I do hope such a proposal will be DOA, but in the current political climate, that proposal is gaining fans in the for-profit companies that use Linux.

But, removing code is actually not as useful as it looks, particularly given the business model that McHardy is chasing. The main strategy that McHardy appears to be using is to focus on companies who are small system integrators who don't do their own engineering of the actual operating system in their products. I've done plenty of GPL enforcement in that kind of situation, and these companies are usually just confused and need education to get into compliance (usually by picking a reputable, instead of a fly-by-night, upstream vendor). Conservancy does that education work in those scenarios, and helps the violator comply.

McHardy, by contrast, just asks for a cash settlement plus an agreement to pay more money if they can't figure out how to come into compliance in a few months on their own. Then, he cashes in on that clause when the confused company, having received no education, can't get into compliance in the time allotted. (I've been leaked a few of McHardy's settlement agreements, and they've been structured as I describe.)

So, how does this relate to writing his code out of Linux? Well, most of these types of companies use very old versions of Linux. I still see Linux 2.6 active, in the wild, in products, on the shelves today. Yes, they're low-end products from companies we've all never heard of (or that we all avoid because they are known to have sub-par products). But there are hundreds, perhaps thousands, of these kinds of embedded Linux manufacturers. It'll be at least a decade before they're all using versions of Linux being released today, so writing people's copyrights out with the goal of preventing GPL enforcement is just not very effective. (And with a ten year lag, it's not even a long-term solution. After all, no one had any thought that Patrick would decide to do this when he started contributing.)

The obvious solution to me is that we need lots of no-cost, freely licensed educational materials out there to help people understand how to comply with copyleft. That's why I work on The Copyleft Guide, and I have an open door policy that anyone who asks me about compliance questions, I try my best to answer them and/or improve the Guide to answer their questions. I wish more people would help me with this, but a lot of people with the expertise to help in this way have built a business around it, so they charge huge fees for training courses and/or consulting. That's why we see so many compliance tutorial products that are non-Free (in both senses of the word).