I think the discussion is centered around preserving the component of Open source which includes funding for the software but retains availability of the source. In such a case, it may not be possible to retain the full pure classic OSS definition.
It might be best to just use Unreal Engine like license where the source is available for inspection and personal use but still being charged for commercial use.
One model that might work is the Unreal Engine model, so an OSS license that is free for personal use, but for commercial use, it might require 3% (say) of your profit from sales of the software product.
One of the most intriguing models to fund OSS that I've seen lately is ICOs. The only downside to it being that it fits a narrow band of software which can have a marketplace. So if you're developing the next coreutils "ls", I doubt you could fund it with an ICO.
Another model is that of sqlite, where the core code is open but unit tests aren't available. In effect, this is somewhat similar to selling consulting but with fewer chances of you having to complicate your software.
At the end of the day, what's important is that the base computing be available in an OSS fashion. Base Computing = Decent OS with GUI + An App store. I think that's the direction Canonical should be going in and attempt to make money by hosting an App store for Ubuntu, just like Google makes money off PlayStore.
I've been thinking about this for some time and I agree with the gist of the article.
Most OSS that comes out:
1. Either funded by giant megacorps because they're trying to commoditize their competitor's edge: See kubernetes, LLVM etc.
2. Common frameworks that people sell consulting around. This is tricky because if your software is easy enough for consumers to use, they won't have use for your consulting. This leads to this bad incentive of complicating software where not necessary. See: Pivotal selling consulting around Spring and Redhat etc.
3. Anything that doesn't fit in #1 or #2 above is mostly not possible with OSS. To think about it, we can just examine the most consumer facing software that we use. Where is a OSS developed messenger app that is as popular as Facebook or Hangouts? 20 years back, we had OSS for most consumer facing software, Unix coreutils etc. Basically, today OSS is reduced to professional frameworks and middleware libraries because that is beneficial to megacorps and they fund this kind of software, but OSS by indie developers is pretty much dead.
Haha you stole the words from my mouth. The fact that CentOS exists undermines all of Redhat's business model.
Honestly, the future of Open source software is probably going to be like Sqlite, in the best case scenario:
1. The core of the software is open source
2. Useful extras like unit tests and documentation from developer might only be available to paying customers.
There is no other way to make money from OSS otherwise and please don't bring up consulting. It doesn't scaleup and to make money you're incentivized to keep making your software more complex, see a nearby comment on Elastic admitting that.
It might be best to just use Unreal Engine like license where the source is available for inspection and personal use but still being charged for commercial use.