Hacker News Reader

Comment by GistNoesis | Hacker News Reader

It's all about computation.

The computer running universe-simulations needs to be able to run fast. Regularities can be integrated quickly. Life not so much.

Life can make computers which can ask computations which can't be reduced.

Therefore as soon as a big enough computer is built somewhere in the universe, the whole simulation-time starts crawling to an halt because the inner computer use all the resources, and future times in the universe where there are many of those computers never get to be materialized.

Alternatively, solutions using time-travel allows to run simulations of longer simulation-time, by introducing a time-coherence force which prevent such computer (aka biological-numerical instabilities) to ever emerge.

View on HN

You give life and computers some special place in the universe for being to able "ask computations which can't be reduced". I don't think that's warranted. In the end, it's all fundamental particles and forces and interactions. I don't see why life would be special in that regard. A small part of material in the core of a star "asks" many more questions each second than a piece of brain tissue or a silicon chip.

I give computation and consistency a special place once you assume you are in a simulation.

If a computer inside the universe runs some proof of work computation like in BlockChain, (aka finding a mathematical one-way function that when iterated is terminated by a big enough trail of zeros), this computation has to be done in the higher plane universe, or a witness from inside the universe could deem the universe not consistent.

The alternative are time-travel explanations where you substitute questions, by replacing questions for which you can have already an answer.

You can approximate a star or any collection of particle by while preserving its observable statistics aka effective-super-particle and get huge computational speed-up. Life are run-away processes for which you can't define statistics.

But their very point was that the simulation wouldn't have to compute every single individual particle, only the "interesting" ones. The rest would be approximated to whatever level required.

I basically agree with you but I can't see how your argument applies here.

It's a very interesting point that I hadn't noticed before. Of course, this relies on assuming (believing?) the universe is a giant simulation run on a computer with limited resources.

I did not get your last sentence though. Could you elaborate?

I dig the universe-is-a-simulation theory, given how we live in an era where our own computation power and simulation accuracy has only continued to increase for the past 70-odd years. Which is a tiny amount of time on a galactic scale.

I wish suspended animation and the like was a thing, I'm super curious about the future and development of humans. I can imagine that within a thousand years, assuming we don't hit a cataclysmic event that wipes everything out, I could no longer understand anything that's going on anymore. (I mean I'm starting to struggle with the younger generations' stuff already... is this it? :o ).

Rules of the universe appear to us relatively simple. But the question of numerical stability is not solved specially for later times.

In physics it is natural to start from an initial state and integrate through time, starting from initial time and evolving by time-stepping.

Physics helps but this doesn't prevent the emergence of numerical instabilities like a big-computer being built.

So if you were running a universe simulation, when a life form start to emerge and starts building a big computer you can "erase" them from existence by approximating them away.

When your simulation starts crawling to an halt, you modify locally the current phases at the time of the modification so that the alien go extinct, and you propagate backwards in time towards the origin of the universe, to find the initial parameters of the simulation that when inputted make the alien never be a problem in the first place while not modifying too much the rest of the universe at current time.

From the perspective of a being inside the universe, the only thing we could observe by such a parameter search on the spaces of possible universe, are being amazed at how some constants of the universe seem to be fine-tuned.

And why exactly would someone go to all this trouble instead of just throttling that region of space? Surely you don't mean to say it would all run on a single chip?

And if this was a simulation there'd presumably be a purpose to it and once heavy computation emerges the entity running the sim would throw more resources at this interesting phenomenon, not try to get rid of it.

It'd be like "each time the neural net achieved AGI, the humans tore it down to refocus on the real goal, bombproof traffic sign recognition".

> It'd be like "each time the neural net achieved AGI, the humans tore it down to refocus on the real goal, bombproof traffic sign recognition".

Except that if the entity who's running the simulation is itself AGI, it really does not have interest in discovering yet another AGI in its own simulation.

It's not so easy to contain life. Potentially it can send information at light-speed and your whole universe become infested with a boring paper-clip AGI. Throttling will create problems at the boundaries for the conserved quantities, and for reproducibility.

But the mechanism I suggest, is a rather elegant (chaos resistant) way to allow to give control over a deterministic universe simulation without having to restart from the beginning, it is akin to re-rolling the dices if you don't like the result.

It is a somewhat natural way to search for the initial configuration of the universe. It present some nice properties. For example you can archive a whole universe by its seed. Which allows many more interesting search strategies. Like in chess you can create variations to guide the search towards region which are of interests for you.

I'd like to entertain the idea, that Bottled Universe Inc. , has to ship some Tera-Year old universes by Xmas. They have created plenty of variations in the universes and selected one just for you. You plug the seed, and you are guaranteed to see the emergence of beautiful universe when you run it inside your less powerful computer-brain.

> From the perspective of a being inside the universe, the only thing we could observe by such a parameter search on the spaces of possible universe, are being amazed at how some constants of the universe seem to be fine-tuned.

That's a good point, and I've also always wondered how fine-tuned those constants are. One problem though, is that changing the initial parameters would most likely have chaotic effects as the system (the whole universe) seems unstable and fragile. For example, it would be extremely difficult to fine-tune the parameters such that billions of years later at this moment, I don't write this comment.

One advantage of QM is that from outside the universe you still have some degrees of freedom : you can modify the phases without impacting the observable of the world.

You can make a pretty strong local change like flipping a coin, at any point in space-time without impacting the rest of the phases elsewhere too much, because the farther away in time and space from your modification, the more decoherent the phases would be and therefore their impact would be minimum.

And because the QM equations are time-reversible, you can propagate them backwards through time up to the origin of time. And you now have a new almost identical initial configuration, which when simulated forward will give rise to a universe that behave the same way until the time of the modification but then bifurcate.

Why do people assume a universe simulating our own runs on comparable physics?

Reply on news.ycombinator.com