On Open Source and Diffuse Contribution

There is something I think is fundamental to understand before talking about open source: the companies that provide services to society do not only respond to ethical principles, but also to a completely legitimate economic need. Those who invest money do not do so out of charity, and those who work expect something in return for their effort. It does not necessarily have to be money, but very often it is. The world works, to a large extent, on the principle of "I contribute in exchange for someone else's contribution".

This is not a flaw of the system, but it does have an inevitable consequence: there is an infinity of human problems that, under this logic, will never be solved. Not because they do not matter, but because they do not generate enough return to justify the investment. This is where open source and non-profit organisations occupy a space that the market, structurally, cannot cover.

The internet stands on open source pillars. Behind every application, every web service or every artificial intelligence model there are tens, hundreds or thousands of open source tools that someone built and made available to the world. PyTorch, TensorFlow, Linux, the very protocol on which the web runs: all of that exists because there were people who decided to contribute without expecting a direct return.

But open source is not just technological infrastructure. Not long ago I started contributing to OpenStreetMap, adding local businesses and parking spaces for people with reduced mobility in my city. It is a small contribution, almost invisible. I will probably never get to see its effect. But that information will be there for whoever needs it, and if more people do the same, in a few years the people who depend on that kind of data will be able to move around with considerably less uncertainty.

That is, to me, the most interesting thing about open source as a form of contribution: the reward is radically diffuse. There is no immediate applause, no clear metric of impact, no guaranteed recognition. Contributing this way requires a certain strength, because the only possible motivation is the accumulated effect of many small actions that, over time, change something real. The butterfly effect, but applied deliberately.

I do not know whether contributing should be a universal moral obligation, but I do believe it is one of the most honest ways to take part in something larger than oneself. And at a time when technology concentrates more and more power in fewer hands, open source remains one of the few forces that pushes in the opposite direction.

Some of my favourite projects are the following:

OpenStreetMap goes far beyond roads and businesses. It collects information that no commercial map has any incentive to include: rural paths, local heritage, infrastructure for people with reduced mobility. Each piece of data added by someone is a piece of data that will be available to anyone, forever.

Tor has a reputation marked by its criminal uses, but that obscures something important: it is also the tool that allows journalists, activists and civilians in war zones or under authoritarian regimes to communicate with the outside world, ask for help and report what is happening. That a tool can be used for evil does not make it evil. Tor exists so that freedom of communication does not depend on the government allowing it.

Home Assistant. Digitising a house can be a fascinating project, but closed ecosystems turn it into the perfect scenario for handing over control of your own home to third parties. Home Assistant lets you do exactly the same without any data leaving your local network.

Open source LLMs are not the same as open weight. Open weight shares the weights but the training remains opaque. True open source makes everything auditable: data, methodology and decisions, which allows the work to be continued and to examine whether the model was trained with questionable data. Opacity does not imply bad faith, since legal restrictions on large corpora are real, but it does prevent ruling anything out. That said, sharing the weights is already a huge step. Hopefully we keep getting closer to true open source.