4 ms·

> Sure, but part of the issue is that we can make accurate predictions of pretty much everything we can readily observe today. Actually there are a lot of pred

> Sure, but part of the issue is that we can make accurate predictions of pretty much everything we can readily observe today. Actually there are a lot of predictions we can’t make today. Stuff that are very important 1) Why do most young people do fine with COVID-19 but some apparently healthy people do not? 2) How exactly do proteins fold 3) Will it rain tomorrow at location x at time x 4) when will my pet tiger decide to maul me? 5) will a drug with this molecular structure for this disease actually treat the disease? We know a load about the basic building blocks. About quarks and protons and electrons. But once we get to more complicated systems composed of these building blocks, we struggle to make predictions. See for example the scrubbed SpaceX launch due to weather. I am sure NASA and spaceX have scientists who are experts in physics, but they could not predict in advance that there would be bad weather on that exact day. What we do now is make calculations on “spherical cows” and call everything else “chaotic” and “too many variables” and pat ourselves on the back for reducing everything to its basic blocks.

mattigames 6y ago
That's what we do because it's all we can do. Meaning it's likely predicting all those things you listed are as hard to predict as the lotto numbers of next week, for which even if we knew the exact position of every ball in the lottery ball machine at atomic level we would need to predict the thoughts of the person who will spin the machine, for that level of predictions nothing short of a full simulation of reality (ala Matrix) would suffice, and even if we already had a machine powerful enough to do that inputting the state of every atom into that machine would still be outside our capabilities, and I'm afraid it will always be.
trixie_ 6y ago
None of the things you listed depend on explanations outside of the realm of physics that we already understand. Theoretically with a simulator powerful enough and using the principles we already know, all of the items you listed could be solved.
- Thorentis 6y ago
  I think there's a difference between understanding a system well enough to make predictions, and recreating a system (a simulation) and then seeing what it does. Many times in history before the physics was properly understood, the best scientists and engineers could do was create some kind of "dry run" and simulate/recreate whatever process they were testing, and use that to predict outcomes of the real one. But actually being able to look at the state of a live system and predict what it will do next accurately is quite different.
  - antepodius 6y ago
    It seems to me to be more of a spectrum, where a 'well-understood' system is one we can simulate accurately with extremely pared-down models. (I.e., we can simulate it much more cheaply than running the whole atomic simulation) because we found symmetries/simplifications/shortcuts. Approximating ballistics arcs with parabolae, for example, is sort of a compression algorithm on the full simulation.
- TheOtherHobbes 6y ago
  Some systems are simply random. No amount of simulation will tell you exactly when an unstable atom will decay. We've spent the last century arguing about what this means for physics, and we still have no idea - even though quantum events can have macroscopic consequences. (To give a contrived example - a decaying atom may damage some human DNA and cause cancer in a historically important individual who dies decades earlier than they would have otherwise, changing the course of history.) It's more accurate to say that we can model a small selection of mathematically 'pure" system types that are neither random nor emergent, on timescales that are comprehensible to us.
Thorentis 6y ago
I think the problems you described are in two different groups though. The first group, is where we understand everything well, but we don't have the ability to track every variable accurately and record the amount of data required, in order to make good predictions. Weather falls into this category. Then there are those like protein folding and animal behaviour where we understand the building blocks, but even of we could track every single building block there would be something missing for us to make predictions.
- adrianN 6y ago
  In case of protein folding it's mostly that the optimization problem is NP-hard.
mhh__ 6y ago
Fluid dynamics is one of those problems I was referring to at the end. I'm not a CFD guy but it's partly an issue of not having enough computing power - fluid dynamics doesn't scale the same way as regular mechanics so you have to compute the whole picture.

mattigames 6y ago

That's what we do because it's all we can do. Meaning it's likely predicting all those things you listed are as hard to predict as the lotto numbers of next week, for which even if we knew the exact position of every ball in the lottery ball machine at atomic level we would need to predict the thoughts of the person who will spin the machine, for that level of predictions nothing short of a full simulation of reality (ala Matrix) would suffice, and even if we already had a machine powerful enough to do that inputting the state of every atom into that machine would still be outside our capabilities, and I'm afraid it will always be.

trixie_ 6y ago

None of the things you listed depend on explanations outside of the realm of physics that we already understand. Theoretically with a simulator powerful enough and using the principles we already know, all of the items you listed could be solved.

Thorentis 6y ago
I think there's a difference between understanding a system well enough to make predictions, and recreating a system (a simulation) and then seeing what it does. Many times in history before the physics was properly understood, the best scientists and engineers could do was create some kind of "dry run" and simulate/recreate whatever process they were testing, and use that to predict outcomes of the real one. But actually being able to look at the state of a live system and predict what it will do next accurately is quite different.
- antepodius 6y ago
  It seems to me to be more of a spectrum, where a 'well-understood' system is one we can simulate accurately with extremely pared-down models. (I.e., we can simulate it much more cheaply than running the whole atomic simulation) because we found symmetries/simplifications/shortcuts. Approximating ballistics arcs with parabolae, for example, is sort of a compression algorithm on the full simulation.
TheOtherHobbes 6y ago
Some systems are simply random. No amount of simulation will tell you exactly when an unstable atom will decay. We've spent the last century arguing about what this means for physics, and we still have no idea - even though quantum events can have macroscopic consequences. (To give a contrived example - a decaying atom may damage some human DNA and cause cancer in a historically important individual who dies decades earlier than they would have otherwise, changing the course of history.) It's more accurate to say that we can model a small selection of mathematically 'pure" system types that are neither random nor emergent, on timescales that are comprehensible to us.

Thorentis 6y ago

I think the problems you described are in two different groups though. The first group, is where we understand everything well, but we don't have the ability to track every variable accurately and record the amount of data required, in order to make good predictions. Weather falls into this category. Then there are those like protein folding and animal behaviour where we understand the building blocks, but even of we could track every single building block there would be something missing for us to make predictions.

adrianN 6y ago
In case of protein folding it's mostly that the optimization problem is NP-hard.

mhh__ 6y ago

Fluid dynamics is one of those problems I was referring to at the end. I'm not a CFD guy but it's partly an issue of not having enough computing power - fluid dynamics doesn't scale the same way as regular mechanics so you have to compute the whole picture.