How the Hacker Manifesto change my life

I had access to the internet at university in 1994. Today, people confuse the web and the internet. But at the time it was very different. Even though the web technically existed, no one used it. It sucked. The web was really born in 1995 with Netscape and Yahoo.

Before that, anonymous ftp was used on the command line to exchange text and images. I wondered why to list the files we used “ls” and not “dir”. So I discovered the Unix world and Linux. And in the E-Zines of the subculture of this early internet, I quickly discovered the Hacker Manifesto.

I didn’t understand what was happening to me. I didn’t even think about it at the time. I became interested in cybersecurity, probably for the wrong reasons.

I was fascinated by the internet and wanted to understand how it worked in its deepest foundations. Some guys on IRC told me that I had to read the RFCs. So I read the RFCs. But it was dark and arid.

Without much difficulty, by drawing inspiration from existing code, I created a few IRC bots. But I wanted to go deeper in my understanding of the net.

And the more I went into the depths of the net, the darker it became. So I started reading the Linux kernel sources. Combined with the RFCs, this is how I was able to touch the depths of the net.

As you always learn better by doing, I created a sniffer. At that time, Ethereal (now Wireshark) did not yet exist. But with the Linux kernel model and the help of RFCs, I managed to implement a basic network stack, with reassembly of IP fragments and TCP streams.

It was a different time. Everyone was using telnet and ftp. The local networks were made of hubs and not switches, so no need for a port monitoring config to see all the traffic. Madness.

I could have been kicked out of my university for that. It happened the year after I left to a friend of mine who repeated his master’s degree. He created a sniffer too. Except that he was caught.

I’ve always been a grey hat. I have never broken a system. On the contrary, I fixed many security flaws that the incompetent administrators of the university had created. Basically, I always protected people even when I wasn’t allowed to.

The day I entered the workforce, I decided to stop this kind of nonsense and become a white hat. Never again will I enter a system without an order from my employer.

But I don’t regret anything. The skills I developed at that time are still rare today. And the vast majority of my corporate experiences were based on these skills.

Twenty years later, I reread the Hacker Manifesto. And I realized that this thing is not about hacking at all. It speaks of the toxicity of our education systems. That’s why I say that I became interested in cybersecurity for the wrong reasons. Because this text had a strong emotional impact on me and I didn’t even realize what it was about.

In Western countries, education systems developed with the Industrial Revolution. The aim was to take country people accustomed to freedom and to teach them submission and boredom. And it hasn’t changed much since. Of course, the problem is more complex. It’s not all doom and gloom. They were also taught to read, write and count, so that they could read a leaflet, count their production and write a report.

The worst problem of education is the relationship to failure. In the real world, as long as we don’t get discouraged, there is no such thing as failure. But when you want to train factory workers, you don’t want failures. You have to repeat well-known exercises and perform them perfectly. Otherwise it’s considered a failure and at school, failure is the end of the journey.

The problem is that in real life failure is the path to true learning. The example of Edison, who failed 6000 attempts before finding the right filament, is often cited. I was an entrepreneur for 13 years and I learned this the hard way. But it freed me a little more from our toxic education system.

For the past two years, I’ve been reliving what I experienced at the time of my discovery of the internet, but with AI.

I already had the basics. Officially, my university course was not about hacking. My research project focused on multi-agents. I had also learned about neural networks with a course by Kevin Gurney available online. Then I continued to train all my life, for example in NLP, inspired by the blog of the French researcher, Jean Véronis, by learning to use Gensim, spaCy or NLTK…

I started interacting with LLMs through libraries like Hugging Face’s Transformers or APIs like OpenAI’s. But already knowing neural networks, I wanted to understand how LLMs work in depth. So I read the research papers on the subject. But like RFCs, they are dark and arid.

Today, I read the code from the Transformers library and everything becomes clear. It’s not that complicated to use Pytorch to create the different layers of an LLM. And when you start to understand how that works, the research papers also become much clearer.

And I want to share my explorations with you. Because we’re all alike.

If you’ve never read the Hacker Manifesto, it’s available here.