Could the Universe Shake an iPhone into Existence?

Talking to Myself in the Rain

Can the Universe Shake Out an iPhone?

God bestowed SwiGLU upon us.

Today I saw something Feng said and found it rather interesting, so I am sharing it.

Quotation | Fengge Wangming Tianya

Fengge, you say the cup-shaking theory is wrong: if you put infinitely many iPhone parts in a cup, with infinite time and parts, could you shake out a complete iPhone? You say a monkey with a keyboard cannot type Shakespeare even with infinite time. But with infinite time, the monkey could type Shakespeare. That is how Poincaré recurrence is defined.

Let me tell you a simple thing: you say shaking iPhone parts together will not work. Do you imagine that “shaking” means an iPhone assembling itself? Are iPhones not assembled by people?

Earth has existed for five billion years; humanity evolved to its present state in one hundred thousand years. In that time it shook out an iPhone. Is that not the universe naturally shaking out an iPhone?

The process is not an iPhone clattering together by itself. First people emerged; people invented computers, the internet, chips, memory, and cameras. Eventually, after all that shaking, an iPhone emerged.

From becoming upright animals to today, humans took only a hundred thousand years to shake out the iPhone. The airplane, too, was shaken out by the universe.

A hundred thousand years after humans left the East African Rift, the Wright brothers and the airplane emerged. Did the universe not also shake out Shakespeare? Where did he come from? He wrote his works, a natural product of the universe, nothing special.

Imagine the universe as a vat—or rather the Milky Way. The galaxy shakes and its systems spin. Life appears on Earth, life gives rise to humans, Shakespeare appears, and then his works.

Five billion years of Earth, hundreds of millions of years of life, one hundred thousand years of humanity: Shakespeare, the iPhone, and airplanes all emerged. Did they not? That is how they were shaken out.

The supposed path to “shaking out an iPhone” Earth Life Humanity Industry iPhone Not parts assembling themselves, but layers of emergence in a complex system.

This does not mean the universe shakes chaotically without any rules.

Rather, in a vast complex system, local processes can follow rules while the whole still involves contingency.

1

This reminded me of Does God Play Dice?

When I saw this passage, my first thought was the book Does God Play Dice?

But I have not read it. I looked it up briefly and found it does not really address the same thing. The book discusses randomness and determinism in the development of quantum theory; Fengge is talking more about “random generation” in cosmic evolution.

Two different questions about “randomness” Quantum debate Randomness vs. determinism Does the world fundamentally roll dice? Cosmic evolution Contingency + necessity How do results emerge from complex systems?

The claim is not that shaking a cup of iPhone parts makes them clatter together into a phone. The universe first gave rise to Earth, Earth to life, life to humanity, and humanity to science, industry, chips, and the internet; only then came the iPhone.

I find that fascinating.

2

Effort does not determine outcomes; it changes the distribution.

Many things are like this. Looking backward, a result seems to have a clear causal chain; viewed beforehand, there is a great deal of randomness.

Gaokao and final-exam scores, the people you meet, whether you get concert tickets: all involve chance. Effort matters, of course, but many specific outcomes are still random—or, in other words, luck.

The crucial one or two points on a gaokao score can sometimes come down to a grader’s momentary judgment. Effort is more like shifting the probability distribution, moving the center of a normal distribution, than directly deciding the final result.

Effort changes not a single point but a distribution Original center Center after effort The actual landing point still fluctuates.

This is not to say everything is luck. On the contrary, awareness of chance makes the meaning of effort clearer.

We work so that, when chance comes into play, we are more likely to fall in a good range.

We cannot control where the dice land, but we can try to change their distribution.

3

Activation functions: great results often come from trying things out.

An amusing example that came to mind is the activation function.

Most of my readers may not know what an activation function is. You can look it up if interested; AI makes knowledge easy to access now. Put simply, today’s large models still fundamentally work with probabilities: based on pretraining, they predict what the next word, or token, should be, rather than truly understanding you as a person would before replying.

Activation functions are an important part of training a model.

Figure: formula for the SwiGLU activation function.

SwiGLU, used in many large models today, performed well in the Transformer architecture and was subsequently adopted by many models.

You might expect such an important research result to have a beautiful theoretical explanation—or at least an account of why it works so well. Interestingly, the Conclusions of the famous paper GLU Variants Improve Transformer (click Read Original at the end for the paper) include this line:

From the paper:

We offer no explanation as to why these architectures seem to work; we attribute their success, as all else, to divine benevolence.

That is: we cannot explain why these architectures work; we attribute their success, like everything else, to divine benevolence—to God’s blessing.

Often, humans do not first arrive at an absolutely correct theory and deduce an inevitably successful result from it.

More often, they have an intuition, run experiments, discover that something seems better, and keep using it. Explanations supplied later are not necessarily the whole of the real reason it works better.

That is rather like Fengge’s “shaking.”

4

Local rules, overall contingency

It is not that the universe shakes without rules; rather, in an enormous complex system, local processes can follow rules while the whole remains full of contingency.

You can explain where every iPhone component comes from and how a chip is designed. But ask further: why did life arise on Earth in particular? Why did humanity reach this point? Why did these particular people, countries, companies, and technological paths come together to produce the iPhone we use today?

That question is hard to answer definitively.

Two perspectives on a complex system Local Parts, engineering, formulas Explicable Whole Paths, eras, encounters Full of contingency Determinism has not vanished; it is layered together with randomness.

So sometimes I feel humanity is too attached to “certainty.”

Everything we see now did not appear out of nowhere. Over a vast stretch of time, it was built up from countless contingencies and necessities.

A final note

Sometimes when I face trouble or worries, I let my mind go blank and think beyond myself, beyond the city, beyond Earth.

In this boundless universe, I am only a tiny speck of dust. What seems like a long time to me does not amount to a moment on the cosmic scale.

And then I think: what do the problems right in front of me really amount to?