The X principle: how we approach problems that seem impossible

Written by Yasin TANIŞ

Every letter in OMNIA-XSI carries a meaning. The X is the one we come back to most often. It stands for the rejection of limits: the point where human intelligence and technology intersect, and the place where the words “can’t be done” are refused.

A sentence like that is easy to print on a wall and hard to practise. So we want to spell out exactly what it means for us, and what it does not.

Power is not knowing everything

The first part of our name, OMNI, refers to absolute power: the power to do whatever we can imagine. Taken literally, that can sound like arrogance. We read it differently. Absolute power is not already knowing everything. It is the will to turn a thought into action, and to keep going when the first attempt fails.

That distinction matters a great deal in deep technology. Neuroscience, electronics and machine learning are full of problems where nobody yet knows the answer, including us. A company that pretends otherwise will eventually ship something that does not work. A company that treats every unknown as a wall will never ship anything new. The X principle sits between those two failures.

The principle in four steps

We wrote it down in four short sentences:

If something is not possible today, investigate why. Discover what is missing. Develop a new method. Try again.

Each step is deliberately ordinary. The discipline is in not skipping any of them.

1. Investigate why

“Impossible” is rarely a single fact. Usually it is a conclusion drawn from a set of assumptions, and the assumptions are invisible until someone writes them out. The first step is to turn “this can’t be done” into “this can’t be done because of A, B and C”.

Brain-computer interfaces offer a good illustration from the field. Early spelling systems based on the P300 response were slow: a few characters per minute. It was easy to conclude that the brain was simply too slow for typing. The real reason was different: a single P300 response is lost in the background EEG, so the system had to repeat each flash many times and average the results before it could decide. The limit was not the brain’s speed. It was the signal-to-noise ratio of one measurement. Once the reason is stated that precisely, it turns into something you can work on.

2. Discover what is missing

When the reason is clear, the next question is what is missing to change it. Sometimes it is a measurement nobody took, sometimes a component that does not exist yet, sometimes data, sometimes a person with a different background in the room.

In technical work, the missing piece often sits on the boundary between disciplines. A software team blames the classifier when the problem is electrode contact. A hardware team adds filtering that also throws away part of the signal the software needs. Nobody is wrong within their own area; the gap is between areas. This is one reason we keep AI and software, neurotechnology, and electronics and embedded hardware in one company rather than handing problems from one team to the next.

3. Develop a new method

If the same method failed, adjusting its parameters is seldom enough. A new method usually means changing the question. In the P300 example, researchers did not only try to filter harder. They redesigned the stimulus layout, the flashing patterns and the statistical models, and they added language models that predict likely letters. As a result, each measurement carried more useful information. The answer came from looking at the whole loop, not one stage of it.

For us this step has a practical rule: before building, write down what result would show the new method is better, and how it will be measured. A method that cannot be tested is a hope, not a method.

4. Try again

Then do it. Most attempts will not work the first time, and that is not a failure of the principle; it is the principle. Each attempt either moves the result or reveals a new reason why it did not, which sends you back to step one with better information. We keep a written record of what failed and why, because the most expensive mistake in research is repeating one you have already made.

What the X principle is not

Refusing the word “impossible” can be misread, so we set some boundaries for ourselves.

  • It is not denying physics. The skull spreads electrical signals, and no algorithm will change that. A real physical limit is a constraint to design around, not an obstacle to argue with.
  • It is not overpromising. “Not possible yet” is an honest answer, and we will give it when it is true. The principle governs how we work on a problem, not what we claim before the work is done.
  • It is not working alone. The X sits where human intelligence meets technology. Many of the missing pieces in step two are other people’s knowledge.
  • It is not speed for its own sake. Trying again quickly is good. Trying again without learning why the last attempt failed is just noise.

Why we built a company around it

OMNIA-XSI is not tied to a single product or technology. That is a deliberate choice, and the X principle is the reason. If the right answer to a problem turns out to be a circuit rather than a model, or a better sensor rather than a better algorithm, we want to be free to build it. OMNIA-XSI Labs carries the research and neurotechnology side of that work, and OMNIA-XSI Systems the electronics and embedded hardware side, so that a question can move between them without changing hands.

Our slogan, “Beyond Intelligence. Into Creation.”, says the same thing in a few words. Understanding why something is hard is intelligence. Building the thing anyway, carefully and honestly, is creation.