Three years ago I wrote a post exploring the new integrated programming language, environment, online editor and pipeline called Dark. I thought it looked pretty cool but lacked a lot of necessary features that would make it useful at business scale, and it also came with major vendor lock-in risk. I concluded by planning to look back at it in a year or two. I did so and, woo, do I have some concerns.

Something happened to Dark, (or DarkLang as I feel they should be marketing themselves for easier searching) because they were everywhere and now silence. If you search for Dark online you find a flurry of promotional posts on various communities and subreddits along with a few articles that seem like guest posts on various tech and programming sites. These are clustered around a few years ago. Since then there isn’t anything much out there and certainly not much from regular users. Hell, my own original blog post shows up. It all looks like there was a big marketing push to create posts, podcasts and social media buzz around that time that then dried up.

But lets go to the source. What are the Dark folks themselves saying about how it’s going? Well, things get weirder. The last post on the official DarkLang blog is almost a year ago and it’s a bizarre post about how DarkLang is going “All In” on AI and Chat GPT. This post landed days before April Fools Day and I remember hoping it was some kind of joke, but apparently not. Tracing backward, their last release was Release 9 in Jan. 2023… but notably the last release tagged in the DarkLang GitHub as of this writing is release 6. Where is the work that the blog is talking about? Did they stop open-sourcing later development? Are they working on part of the infrastructure that isn’t open source?

The fact that the last blog post is an AI/ChatGPT announcement that also drips with nihilism – referring to themselves as an “ageing rockstar” and author Paul Biggar stating “If you’re wondering what we’re doing, I’ll tell you honestly: I don’t know,” alarms and confuses me. They were onto something that suited a niche but they seem to have wandered off into the weeds. That entire blog post is a red flag to me.

While there are still commits happening to the Dark GitHub repo, a search of their roadmap site doesn’t reveal the term “AI” anywhere. So their published roadmap hasn’t been updated to reflect their last paradigm-shaking blog post in almost a year.

My analysis of this is grim.

At worst the Dark project was an investor grift. They pumped a lot of effort into making an MVP (Minimum Viable Product) that looks polished and exciting, created buzz, attracted investor money (they still have a few GitHub Sponsors too), then went nowhere. Jumped on the AI buzzword train to maybe attract some of the manic investment money around AI and LLM technology, and then seem to have lost interest in community outreach.

At best perhaps they are finding their footing in this new AI/LLM landscape, where automatic code generation is making a lot of developers sweat. Doing some re-evaluating and rethinking the plan for Dark. Maybe they will come back into the light in a year or two with something exciting.

I will tell you though, when I see corporate sites for projects like this that are in the early-access, rapid development stage, that don’t keep a public record of their progress on their own blog, it rubs me the wrong way. You’re only doing as much as you are publicly seen to be doing, if your business model relies on public uptake of your product or service.

Conclusion: I hope that Dark is just getting their plans sorted out. But for business and enterprise: give DarkLang a pass for now.

Update Feb 8, 2024: Paul Biggar from Darklang reached out to me in a Twitter DM. He asked if I’d like to chat verbally about this article. I declined a phone call but offered to add any written comments from himself or Darklang to this article in the interest of fairness. I have not yet received any written comments.

Update Mar 15, 2024: Darklang has published a blog post which clarifies their current status and acknowledges the limitations of their ‘classic’ version. Sounds like it’s mostly back to the drawing board and their interest in AI seems related to being able to take advantage as developers of AI code generation tools like Copilot, so they are keeping this in mind.

It’s a shame that the classic version has been mostly written off as flawed. A lot of it’s flaws for professional developers, I see as benefits in an educational setting. Darklang classic would be perfect for educators taking the leap from visual gameified code education tools like Scratch, where vendor lock-in isn’t a concern and the fact that it’s all contained in one web-based editor without infrastructure to worry about, would be a boon to educational uses. Oh well. For future coverage of Darklang, follow them online and keep an eye on their blog.