Create

Admire first

Life is made of small pleasures. The biggest ones here are probably music, food, games and learning.

Before every creator, there is first an admirer: someone who was moved by something, and who keeps a pleasant feeling about it that they want to make last. That is where the urge to create comes from: finding that feeling again in a new form, even closer to one’s own sensibility, and then, if possible, letting others live it too.

Few things are as satisfying as making, with your own hands, the very thing you were longing for. Only one is more so: seeing someone else find happiness in it as well.

Games: a stranger, one evening

In early October 2026, a player from Poland started playing BlitzVolley. A stranger, who arrived no one knows how. He invited friends, then carried on alone in career mode, exactly the way the game was meant to be enjoyed. He spent several hours in a row on it, something even its author has never done yet.

The game is still very far from perfect, and yet people are already having fun with it. If a creation can offer a moment of happiness, even just a break from the harshness of the world, then it has fulfilled its mission. That holds for every field of creation.

Music: an emotion that never wears out

Composing started from the wish to hear sounds, harmonies, emotions, sometimes a story written down in notes, that satisfy the ear and the mind and move you deeply.

Sometimes a single passage takes a considerable amount of time, polished until it sounds right. Once finished, a piece of a few minutes can loop for hours without growing tiresome, and with every listen the same emotion comes back, the one from the moment it was written. It is a fascinating property of music, and probably what makes it timeless.

Music carries emotions, stories, experiences, and everyone makes it their own. The listener recognises the composer’s emotion and ties it to their own life. No words, no language barrier, no barrier of era: only emotion, which connects all human beings, and maybe more. Perhaps the most beautiful invention of humankind.

Cooking: caring for body and mind

The emotion takes another form at the table (nobody cries with sadness over a plate), but it is there all the same. A good dish comforts, fills you up and lifts your mood. It is our fuel, and we come back to it several times a day. Hence the importance of taking care of it, for balance as much as for taste: one looks after the body, for today and tomorrow; the other looks after the mind, through the pleasure it brings. The whole art is balancing discipline and pleasure.

Being able to cook, even just for yourself at first, is a real privilege. It takes time, but that time is quickly paid back: the pleasure of preparing, then tasting; eating healthily, often for less; cooking exactly to your own taste, cuisines from everywhere, without depending on anyone; letting others taste.

And sharing: passing on what you have learned, putting a comforting dish on the table that brings everyone together. Nothing compares.

The same pleasure carries on into tasting: coffee, wine, chocolate, an aged cheese, with their depth, their aromas, their acidity, their body and their finish. And the balance of flavours echoes the balance of chords in music: those are the analogies.

Being surpassed

The greatest pleasure in programming: being beaten by what you built. That is probably why neural networks and heuristic engines keep coming back from one project to the next: fish learning to feed, a second chess engine with a learned evaluation, a Barricade engine, the BlitzVolley bots. They are creations that improve before your eyes, game after game, generation after generation, until they do better than their author.

With BlitzVolley, it was especially striking. The neural network bots are given no hand-written strategy: they learn on their own, by playing. After very short training, from the “normal” level up, they started winning. Close to a hundred matches later, not a single win against them. And they found techniques nobody would ever have suspected, in a game designed from start to finish by the very person losing.

That moment was a milestone. It could feel like a defeat; it is quite the opposite: pride, admiration, almost jubilation. The goal was never for the creation to stay at our level.

An engine in its maker’s image

With OptiChess, the chess engine, the turning point came later. For a long time it stayed weak tactically, yet dangerous in actual games, thanks in particular to good time management: the losses kept coming anyway. Then its tactics firmed up, its evaluation grew sharper, and another threshold was crossed: in analysis mode, you start trusting the engine more than your own judgement. The rare times doubt remained, a check with a far stronger engine, Stockfish or Leela Chess Zero, proved the engine right, not its author.

Its search takes the opposite approach to brute force. A grandmaster calculates few moves compared to a machine, and still reaches enormous depth: a glance at the position is enough to discard most moves, then they follow the lines until a stable moment, where nothing is critical any more and the evaluation becomes reliable. Classical engines examine almost everything up to a given depth, without asking whether the position is critical or not. OptiChess relies on a home-made tree search, close to MCTS: it deepens the promising lines first, discards the moves that are almost certainly bad, and focuses its effort on a small share of the available moves. Raw speed is then just a bonus, letting it go much further than a human.

Its evaluation is made in its author’s image: every function was written the way a position would be judged by eye. Part of intuition remains out of the machine’s reach, which is precisely what neural networks are for, and they are not integrated yet. But the exercise is to formalise what gets evaluated by instinct: writing down rules that are written nowhere, and that still give the values one would have given spontaneously. It is hard work, and fascinating.

King safety is the hardest of all. Without calculation, it is almost impossible to estimate: either it is mate or it is not, and the evaluation diverges completely depending on which. Anticipating that, generalising it, boiling it down to a number that holds in most positions, is very complex. It is the function that took the most time, and it is deliberately aggressive, even if that means slightly overvaluing the attack. So the engine plays with tremendous aggression, never hesitating to sacrifice material to go after the enemy king. Against Stockfish, that is not optimal: everything gets refuted with near-perfect precision. Against a human, it is terrifying. It is exactly the chess one wants to see when building an engine: a way of seeing the game and loving it, close to romantic chess, where the value of a piece weighs nothing against an attack on the king.

In the end, apart from its better tactical understanding, the engine plays in a strikingly similar way, and that is what makes it so endearing: a bit like watching yourself play, only better. Flaws and bugs remain to be fixed. It is a long project, started with far less experience than today, and it still bears the marks. But it remains a great source of pride, with the hope that one day, despite a style so unusual for an engine, it joins the best of them, in a world of brutes. The rest of the thread continues on the chess side.

Coming full circle

The idea does not stop at machines. A game truly succeeds the day its players find strategies its author never imagined. A piece of music takes on a new life when someone hears it through their own story. A recipe gets better in the hands of the person it was passed on to.

It all starts with admiring something that moves us. We try to understand it, then we make something of our own. And the most beautiful part is when the creation grows to the point of surpassing us: the one who admired ends up admiring what they themselves brought to life.