Thought

Years of work in electronics, robotics, network security and artificial intelligence have taught me that technologies change, but one error repeats with remarkable consistency: we choose the answer before we have understood the problem.

Technology is usually encountered where the answer is already visible—a product, system, algorithm or device that works. Yet the decisive choice was made much earlier, when the problem was framed. That framing determines what will be seen, what will remain invisible and which answers will never get a chance to appear.

I believe that when the question is wrong, the best technology merely produces a more precise answer to the wrong question. Choosing the tool is therefore the last stage of thinking about technology, not the first.

We cannot understand the world without models. They help us contain complexity, make decisions and build. But every model reveals one part of reality while pushing other parts outside the frame.

I enjoy models, but I owe permanent loyalty to none of them. Trouble begins when we forget that a model is one way of seeing and allow it to take the place of reality itself. Past success makes this danger worse: a model that has answered correctly several times gradually changes from one possibility into the only possibility.

This is not abstract for me. At Targoman, there came a point when continuing with statistical machine translation was no longer defensible, despite the contract, the plan and the money already invested. Technical reality had changed, so the plan had to change with it. Persisting with yesterday’s answer merely because we paid for it is not a virtue.

The art lies in changing the model, not worshipping it.

Expertise is valuable. It shortens long paths, prevents repeated mistakes and makes complex work possible. Yet expertise also accumulates mental constraints: this cannot be done; this was solved before; this is not how the field works.

Many of those constraints were once correct, which is precisely why they are difficult to see. Ignorance is not a virtue, but knowledge does not guarantee vision either. Expertise retains its value only while the expert can treat even deeply held knowledge as a provisional model.

My objection is not to experts; it is to the worship of expertise. An idea should earn authority through reasoning, evidence and its power to explain reality—not through the name or position of its speaker. In a real conversation, ownership of an idea matters less than where the truth stands after criticism and correction.

Abandoning a belief is not failure; becoming unable to abandon it is.

Freedom in technology does not mean the absence of structure. Without structure, collaboration, scale and continuity become impossible. The problem begins when structure stops being an instrument and becomes the authority that decides what is true, eliminating alternatives before they can be tested.

For me, the value of open source is not merely zero price or access to code. Its deeper value is that it preserves competition among models: no answer can make itself the only possible answer, and the option to choose remains alive even when we do not exercise it today.

For the same reason, I have never equated support for local technology with removing its foreign competitor. At Targoman we deliberately refused to ask for foreign services to be restricted. If a product has value, it should be able to stand in a field of real choice and rely on its own quality.

The problem is not structure; it is the moment structure becomes sacred.

My path through work and life has never been a straight, clean line. I moved from software to electronics, from robotics to network security and from there to artificial intelligence. I do not believe that delay or failure is virtuous by itself, but much of what I know took shape inside those changes of direction.

Philosophy that never meets a real constraint can become a game of words, just as technology without philosophy can become a game of tools. Building is where mental models are tested against matter, time, cost, people and environment.

That is why I distinguish sharply between an elegant idea and a system that must work tomorrow morning. The same is true of ethics. Transparency, fair competition and opposition to corruption are only positions until they become visible in working methods, output quality and a willingness to accept responsibility.

In teaching, the names of tools and commands rarely endure. Tools change. What is worth transferring is the ability to break a problem apart, expose hidden assumptions and reframe a question for which everyone already has a prepared answer.