OpenTF Announces Fork of Terraform

SQLite is also a particularly interesting comparison here in that they do not accept outside contributions at all. In fact, their integration tests are not public. So there aren't any experienced outside contributors who could maintain a fork of SQLite without significant personal and technical ramp-up time, if SQLite were to change their license for future versions, and there wouldn't be any room to complain about old versions of the CLA talking about the importance of open source. And on the other hand, Terraform requires an ecosystem of outside contributions (providers/integrations) in order to be a worthwhile product, whereas SQLite is self-contained.

I'm not sure what the conclusion is from this. I think we'd all find SQLite relicensing to be much more shocking than we found Terraform relicensing, despite what feels like a much bigger "moat" for SQLite.

Anyway, it's interesting to me that Vagrant, the original HashiCorp product, started as a "personal side-project' (as Wikipedia puts it) and was funded for a while by Engine Yard, which itself never really grew to be huge but still exists today. There's definitely a world in which Mitchell Hashimoto kept working on it without trying to make it into an IPO. On the other hand, that's likely a world in which Terraform doesn't exist the way it does today, if it even exists at all.

In reading about how exactly HashiCorp came to be, I stumbled on https://mitchellh.com/writing/the-new-normal which I think at least partially explains it - he happened to be in a culture (which I also lived through, being a college graduate who moved to SF at the same time) where the thing to do, if you wanted to work on a particular technical problem ore, was to make a startup. "Startup" is a very specific word: it's not just a newly-formed company, but one with ambitions of essentially unbounded growth, at least until it "exits" in the form of an IPO or acquisition. The SQLite folks, meanwhile, seem to use 6th-century monastic life as their "normal" (see https://sqlite.org/codeofethics.html), which has vastly different metrics of success.

Perhaps we need to get better at establishing cultural norms and stories and exemplars of growing just enough to get the stability to focus on the work you're passionate about, but not beyond that. It definitely happens - apart from SQLite, it's what Linus Torvalds and Guido van Rossum have done, for instance - but there isn't a "script" for it in the same way there's a script for making a company with superlinear growth, getting an outside CEO, and getting to an IPO.

(Another separate but interesting point is that Terraform is basically _only_ useful for large for-profit companies. It's open source, yes, but it's kind of hard to defend Terraform as the sort of product that's valuable to the crowd that runs Linux on our desktops and cares about our technological autonomy - the whole product is basically about being able to run stuff on cloud providers! We very much need some sort of cultural script for saying, I want to work on this interesting problem that's only valuable to big companies running at big scales, but is really better for those very companies if it's run as a commons instead of a for-profit product, please fund me modestly but reliably. See also RHEL/CentOS.)

> But if there's too much adoption, then what does anybody need you for? Software like Terraform and Vault are used by SRE and DevOps teams, which means companies already have teams doing and creating what a company like Hashicorp ostensibly would want to sell to you as value-add professional services or products.

I think there is quite the big market for professional services, but that would be a rather labor intensive operation which most investors don't want - to really get the valuation up you need huge market adoption and then get into rent-seeking. I think this really only works with older software products with a lot of inertia and enterprise adoption.

> This looks to be a successful fork

The fun thing is that even though there had been that announcement over a week ago, and the deadline expiring one week ago, there actually isn't a fork yet that's publicly available. I'll call it successful once it's public & there's some evidence of a diverse set of contributors from various companies.