The Split

Over the last three weeks there has been quite a bit of drama in the MeshCore community. Before I go any further, a disclaimer that I am just one man who has observed this dispute from the outside. Any claims made by either side are just claims, not necessarily facts. It is also possible and even likely that I get some of these details wrong. I’m not a reporter, I’m just a guy. Take everything I write here as my opinion and imperfect recollection of the situation. Do your own research.

What Happened?

So, yeah. A couple of weeks ago, there were suddenly a flurry of random posts across the “official” MeshCore Discord and Facebook groups, as well as on Reddit saying that there had been a split in the developer community and that the site we had all considered to be the official website for MeshCore - www.meshcore.co.uk - was NOT the official site, not to be used, and that the official site and firmware flasher were now at www.meshcore.io.

While at first this seemed like just better branding for an international community, it quickly became apparent to me that something else was going on behind the scenes. This was eventually supported by a blog post on the new site.

This post describes the view of the “core firmware devs,” and they essentially make two main claims:

  1. Andy Kirby has registered for a Trademark for the term MeshCore, personally, and without the blessing of the larger community.
  2. Andy Kirby’s companion app for the T-Deck and other standalone devices was “Vibe Coded” using AI.

Weirdly, it seems that the vibe-coded accusation was implied to be the more serious aspect here, both in the original blog post and in a lot of the discussion around the community. This is absolutely insane to me.

Filing for a trademark for a F.O.S.S. project like Andy allegedly did is wild and reprehensible behavior. Bad, bad stuff. Not a good look at all for Andy.

Vibe Coding

I don’t know why the AI tool usage, which is what I will call the “vibe coding” is the bigger issue. Andy made a product, offered it for sale, and people can buy it or not. He could have hand coded it himself in assembly, used an LLM, paid a guy in Bangladesh or Romania, had his 11 year old nephew review his code, etc, etc, etc. The fact is that as consumers of software, we don’t actually know who wrote the code, what tools, they used, etc. At least for closed-source software. It should be judged almost entirely by how well it works, if it has vulnerabilities, if there are other bugs or crashes, etc.

Vibe Coding is a term that has come to mean someone who doesn’t know how to write code using plain English and an LLM to generate a piece of software without the author ever looking at the output or reviewing / understanding any of the code. The acceptance criteria is “does it work and do what I want.” Of course many people have tried this but I will just say (as a software engineer), I haven’t seen any actual Vibe Coded apps reach the level of Andy’s project, MeshOS. LLMs just aren’t quite there yet unless at some point in the loop there is someone who has an actual dev background reviewing the code, guiding the architecture, etc. Initially I strongly suspected that the “vibe coding” claim actually meant “a person with a software dev background using an LLM as a tool” (which is not the same thing). In Andy’s response (see below) we find that he claims to actually have a background as developer. I tend to believe him there.

Regardless, I think the core devs’ focus on “our code is all hand-written, and his is written by AI” is a giant red herring.

➡️ The trademark filing is actually the very bad thing that Andy allegedly did. ⬅️

Quite frankly the “hand written” code in the main MeshCore firmware is quite sloppy, amateurish, and just not all that great. Every single time a new firmware has been released recently is has broken some part of the existing functionality in one or more of the hardware devices. There don’t appear to be any standard tests or code review process, etc. People who live in glass houses shouldn’t throw stones is all I am saying.

The other issue I have with the official “The Split” blog post heavily attacking Andy’s MeshOS software is that one of the “core devs” makes a competing product for standalone, Ripple, which is also closed-source, and which ALSO has some features behind a pay wall. It also is far from perfect, in my experience. But we have this open source firmware that is the “official” product, and then two closed-source options for standalone devices. The fact that one of the authors of the blog post has a competing product to Andy isn’t even disclosed in the post, which I think was a mistake not to do.

Andy’s Response

A couple of weeks after the MeshCore.io blog post, Andy published a video response telling his side of the story. I don’t think he did a great job with this, at all. Watch it and make your own decision there, I think I’ve addressed the key issues above already.

One thing I think people are glossing over is that UNTIL this split, Andy’s work was THE default “official” source. Andy’s Discord was THE MeshCore Discord. Andy’s website was THE MeshCore site, including the official way to flash firmware to a device. Everyone used these as such and none of them were ever contested as unofficial by any of the core devs. I saw Andy make a MeshCore youtube channel which would often explain new features in the firmware and how to get set up in a way that I never saw from the core devs. In fact I saw almost NOTHING from the core devs outside of the Github / public facing.

Without the work to reduce friction, build community, explain features, market the product, etc - the ‘core devs’ would have a neat piece of firmware with almost NOBODY using it. The fact is that while Andy wasn’t writing any of the firmware/protocol code, he was indisputably a major and important part of growing the product. He didn’t do a good job of advocating for himself about this in his video, and the core devs certainly have never acknowledged his contributions - claiming that basically they deserve all the credit. The fact is that their docs kinda suck and their marketing and community building was nil.

If I was the King of MeshCore

  1. Trademark is held by a non-profit organization.
  2. Organization has a board of people who control that, this would contain the core devs and, honestly, could also potentially have Andy as well.
  3. Ripple and MeshOS continue as competitors, which is what they are, NOT associated with this board or the organization.

But… I’m not King of MeshCore. I’m just a user and a builder of my local infrastructure. It will be very interesting to see what happens. There is talk of a re-brand new name / another split (keep in mind one could think of MeshCore as a splinter from Meshtastic) - I feel that would only be a bad thing for the brand and community.

At the same time I am, myself, taking a look at other alternative technologies in the space and seeing there is something better or a way to bridge these types of mesh networks together. Stay tuned for that.