Source code does not a software product make

Recently, there has been a lot of hoopla around software that has been cloned using coding agents (an impressive suite of Adobe products and Microsoft Office app clones and a Canva clone). Want a more faithful clone? One can also first reverse engineer using AI and then clone using AI as well.

I’m not here to debate the quality of these coding-agent clones implementations. Or whether everything now is open source because tokens are all you need to both reverse engineer and implement. I’m here to talk about why software products are in essence services. Let me try and explain this to you with 2 points.

Successful open source projects are almost always backed by a community or a company

Enterprises don’t want a lump of code. They want support and they want assurances. They want help with deployment, patching, integration and training. And for all this, they’ll happily pay the support tax. I do not know of a single enterprise that thinks they can save money by running a bunch of open source software all by themselves. Any disruptions caused by said software and the whole equation changes. Software forms the central nervous system of most enterprises and they want someone to assure them they can depend on it: versioning, migrations, and security are hard to do and require a lot of thought.

The same goes for individuals using software for professional use. A community around an open source software project brings similar benefits as a company. It’s even better if there’s a commercial entity built around an open source project. It allows maintainers to sustainably maintain it.

As agentic coding explodes, you’re likely to find impressive applications built and abandoned all over GitHub. Without a community or a business backing them, they will just be Git repos and nothing else. Real software products that can be put to use, on the other hand, need maintenance and support.

Source code leaks did not lead to rivals in the past

One has to reason, as to why after the source code leak of popular software, no rival software companies popped up or why businesses behind said software packages did not suffer commercially. In fact it seems, it hardly mattered at all. Here are some popular leaks:

• Windows XP and Windows Server 2003.
• Portions of Adobe Photoshop, Acrobat, etc. About 40GB of material leaked.

Real Software Factories will need to automate software business, not just write code

We’re at a stage where the ability to write complex software with very little time and effort is a novelty and that surprises us. This will wear away very soon. If Software Factories are to be interesting, they should not just automate the creation of software, they should also automate various aspects of running a software business as well. But, they’ll start with the automation of software engineering, of which writing code is only one part.

Software, whether open source or not needs a vision, roadmaps, customer feedback, but more importantly: maintenance and upkeep.