Toast in the Machine

I recently saw yet another exchange about developers being expected to provide support for open source software. The debate on this topic isn't anything new. It's gone on long before I got into open source and will probably be going on long after I'm dead. I've heard it all before, but this time one of the takes made me stop and ask a question.

The basic idea was that it's irresponsible to your users to release software if you can't support it. It sounds pretty straightforward at first, but there's some very important assumptions you have to make in order to resolve that sentence. Specifically, what do you mean by “release” and “support”. I don't think any two people reading that sentence will insert the same definition for either of those terms and that's a recipe for misunderstanding. In reality, both release and support are sliding scales.

First let's consider what we mean by support. Is support responding to an email from a user with a question? Is support acting as a maintainer for the project? Is support providing ongoing bug fixes with an SLA? Is support adding new features that get requested? Is support having an on call rotation? These are all valid definitions of support, but it's pretty clear that some of these would be an absurd standard to hold every open source developer to. But where do we draw the line? I think that varies from project to project and largely depends on what release means for that project.

Just for example, let's take my text editor. I've released this code by putting it on GitHub, but the intention isn't for people to actually use it as their text editor. The code is there for other developers to use as a reference for how to build terminal applications from scratch. The only documentation is a single README, and that begins with a note about how it exists for educational purposes. Now consider Helix. Helix has a logo, marketing page, nicely packaged binaries for multiple operating systems, and extensive end user documentation. Helix has a very different definition of release, and I think as a result the expectation of support is much higher than for my text editor.

Beyond that there are also external factors that can influence what support should be offered. What if the user would rather the software exist without support than not exist at all? What if your software results in a critical CVE? The context of a specific situation can't be ignored.

Ultimately, the point I'm trying to make is that the level of support a developer owes their users is a sliding scale that's influenced by a variety of factors. Reasonable people can disagree about where exactly we draw the specific lines, but treating it as a yes or no binary is always going to be wrong.