Banon: License changes to Elasticsearch and Kibana

> Back to the topic, see the video starting at 14:00. Basically FSF/RMS left glibc at LGPLv2.1 for fear of losing users. Have to agree the optics don't look good with the FSF trying to convince others to upgrade to *GPLv3, but not having the courage to do it for one of their own flagship projects.
> Adding insult to injury, jra's baby (Samba) switching to the GPLv3 cost them a lot of popularity. Guess I would be pretty miffed if I was in his shoes.

I think that's a big part of it. But the GPL2->GPL3 transition seems almost quaint at this point. Legal enforcement of linking is far less important nowadays than the degree to which services and Big Tech are eating the world. The GNU Affero License helps bring services back on a level playing field, but it doesn't change the economics of tech behemoths throwing a tsunami of resources at a project and stiffling the viability of anything else.

> Obviously I can't speak on behalf of jra, but he did go into some details in his copyleft2020 talk at https://archive.org/details/copyleftconf2020-allison

Thank you. That helps understand more of the details.

+1 for the recorded talk. I missed it earlier. His key analogy is that cloud infrastructure is the next iteration of the foundation-role that the PC revolution played in the 1980s, where the standardised hardware of the IBM compatible microcomputer (the "PC") enabled the scale necessary for the community of open source software (and at the same time, the scale opportunity that justified investment in a lot of proprietary software). He treats this standardised environment as a black box with a well-defined protocol, and he says that's all we need. To make cloud infrastructure the equivalent for "freedom", we don't need AWS to publish its source code, we need all the service APIs or protocols etc documented. Success is when we can write an open source client to talk to a closed server, or vice versa. So a focus on licence enforcement is not only futile (jra's experience and observation) but it is solving the wrong problem. In the cloud context (or IOT device context) having the source code is actually mostly useless for taking advantage of the freedoms that open source licences are supposed to provide (although I suppose it is a way of documenting the API or protocol). He says, I think, that the best hope of getting protocol standardisation and documentation is enlightened-self interest, and as proof he discusses the rejuvenation of SMB after Microsoft chose to go down this path (it's not the only example he gives). [Although did Microsoft choose, or was it forced by the EU?]

The funny thing is that for the purpose of selling a service, Amazon only needs to implement Elastic's APIs, just as jra says. AWS probably doesn't use a real elastisearch server, they probably would just middleware it to some massively amazing backend. But as early as 2019 they prepared an independent project and now they have activated it. It's interesting to speculate what Amazon's motivation is. Enlightenment?

He really believes community is at the heart of open source (not licence enforcement), and he concedes there is one big value of a licence: it helps builds a community around a common statement of values, which Elastic should think about, perhaps.