Remove ads

Transposition, or: Ship Happens [Worm/Aoki Hagane no Arpeggio | Arpeggio of Blue Steel]

Reader mode
Darth Thanatos said:
Yeah.

But the main goal of evolution is not producing "the best", but just "good enough" to meet the challenges posed by the environment. If by dumb chance you get better than "good enough", more power to you, but it's not what was intended.
Click to expand...
Click to shrink...
The thing about using evolution as your optimization tool is that you can continually reset the definition of "good enough." Often, it self-ratchets, because "good enough" is "better than the current best." (This is a gross simplification, but should get the point I'm makinbhere across.)

permeakra said:
Evolution is shit in designing from scratch. Actually, it can't design from scratch, period.
Click to expand...
Click to shrink...
Depends how you define "from scratch." Largely, intelligent design of any sort, including evolutionary algorithms, is not designing from scratch. It is always working from some element of the known.
 
Segev said:
Depends how you define "from scratch." Largely, intelligent design of any sort, including evolutionary algorithms, is not designing from scratch. It is always working from some element of the known.
Click to expand...
Click to shrink...
I guess that you could say that evolutionary algorithms are good for finding the starting point, and then it's better to manually go to hell with it with occasionally running another algorithm to see what you might've missed, and then manually trimming it down and improving it.

After all, evolution takes a ridiculous amount of time for simple things, while scientific method allows for much more efficient means to an end but also depends on a solid foundation.
 
Segev said:
And yet, evolutionary procedures are used in Computational Intelligence all the time to innovate and discover new, better ways to do things. I wrote my Ph.D., in part, on the subject.

I suspect you're operating from ignorance, here, and making assumptions about it that are not warranted. Primarily, you're forgetting that, while in theory, there's no difference between theory and practice...in practice? There is.
Click to expand...
Click to shrink...
It's a tool for your toolbox, one among many. It's normally used in combination with other techniques -- simulated/quantum annealing and their many varieties, backpropagation, artificial (perhaps manual) selection, etc. etc. etc. Use it on its own, and you're likely to build CPU clocks that secretly pick up radio waves from your instrumentation.

The Entities, as described, use only evolution. That's why they're dumb.

If you use your human intelligence to examine the results of your evolutionary algorithm and reject the unusable, crazy or far too situational stuff, then you're already not using pure evolution.
 
Last edited:
Baughn said:
It's a tool for your toolbox, one among many. It's normally used in combination with other techniques -- simulated/quantum annealing and their many varieties, backpropagation, artificial (perhaps manual) selection, etc. etc. etc. Use it on its own, and you're likely to build CPU clocks that secretly pick up radio waves from your instrumentation.

The Entities, as described, use only evolution. That's why they're dumb.

If you use your human intelligence to examine the results of your evolutionary algorithm and reject the unusable, crazy or far too situational stuff, then you're already not using pure evolution.
Click to expand...
Click to shrink...

They might use only evolution because they have tried everything else.
 
Baughn said:
It's a tool for your toolbox, one among many. It's normally used in combination with other techniques -- simulated/quantum annealing and their many varieties, backpropagation, artificial (perhaps manual) selection, etc. etc. etc. Use it on its own, and you're likely to build CPU clocks that secretly pick up radio waves from your instrumentation.

The Entities, as described, use only evolution. That's why they're dumb.

If you use your human intelligence to examine the results of your evolutionary algorithm and reject the unusable, crazy or far too situational stuff, then you're already not using pure evolution.
Click to expand...
Click to shrink...
Nah, the Entities are dumb because they're using solely conflict as a means to try to produce what they need; but given that they're looking for creative power use, the scheme they've gone with will only give them the most vicious power users.

It's more of a misconception about how evolution works than anything. 'Survival of the fittest' doesn't mean everything fights each other to the death and the winner gets to reproduce, despite the fact that's how the Entities seem to interpret it.
 
PhDEevee said:
Nah, the Entities are dumb because they're using solely conflict as a means to try to produce what they need; but given that they're looking for creative power use, the scheme they've gone with will only give them the most vicious power users.

It's more of a misconception about how evolution works than anything. 'Survival of the fittest' doesn't mean everything fights each other to the death and the winner gets to reproduce, despite the fact that's how the Entities seem to interpret it.
Click to expand...
Click to shrink...
There's a reason lagomorphs won at survival of the fittest.

...oh god don't give the Entities ideas.
 
TikiTau said:
Destroyers in Arpeggio tend to not have mental models. There are exceptions, but thatbis usually from a bigger ship donating extra processing power.

Carriers are an entire different kettle o' fish, and get referred to as 'suppression vessels' or 'assault ships'' and apparently can do some really funky tricks with their klein fields from the little we've seen.

Amusingly, they stopped using aircraft because they're too small for The Usual Fog Tricks, so humanity can shoot down Fog aircraft reliably. It's made the carriers a bit bored, since being a carrier was still their primary purpose.

Of the two that have shown up in manga, One of them spends her time fishing while the other collects books and antique nautical navigation equipment.
Click to expand...
Click to shrink...

We're up to three of them now.

Zuikaku's got her love of Fishing
Akagi we're still learning more about
Lexington's got a love of history and has amusingly infiltrated human society as well or better than the I-40X units (and unlike the I-40X series the Americans may not know she's a mental model in her civilian guise). We don't know her human alias but it's probably decent enough to get her job... She's currently on paid vacation to do this mission she's on. Gal knows how to work her jobs. Also her emblem does appear modified with the laurels being added on as when she says she's the University of Fog and the entire teaching thing she's not lying. It's her day job.
Lex knows you're browsing SB during class. She's not going to hack your phone as it'd be rude.
 
Farmerbob1 said:
We have not seen a complete Scion. Remember that he is extremely diminished in the story. Most of his abilities loaned to humans.

Now, a Scion that is complete will have EVERY power we see in the story. Leet's ability to make ANYTHING. Tattletale's ability to intuit. Dinah's ability to predict outcomes. Number Man's ability to crunch complex problems.

It is entirely possible that the entities have already tried ALL different discrete and intuitive methods to find the answer to the only question they have any interest in.

Reversing entropy apparently does not have an intuitive solution.

So, what is left? Look for unpredictable anomalies. When you find them, study them and see if they give you useful data.

The fully assembled Space Whales might not be as stupid as many, including myself in the past, have thought.

Evolution style research might be their only hope.
Click to expand...
Click to shrink...
Its a bit like giving a monkey an auto shop.
 
DestinyPlayer said:
I guess that you could say that evolutionary algorithms are good for finding the starting point, and then it's better to manually go to hell with it with occasionally running another algorithm to see what you might've missed, and then manually trimming it down and improving it.

After all, evolution takes a ridiculous amount of time for simple things, while scientific method allows for much more efficient means to an end but also depends on a solid foundation.
Click to expand...
Click to shrink...
Eh, you'd be surprised. I don't really feel like trying to dig out my old research, and have even less idea how I'd demonstrate it to you, so I'll allow you to assume I'm a liar if you wish, but I know from experience that evolutionary algorithms are actually quite fast compared to "manual" or other "directed intelligent" methods of solving certain classes of problem.

Baughn said:
It's a tool for your toolbox, one among many. It's normally used in combination with other techniques -- simulated/quantum annealing and their many varieties, backpropagation, artificial (perhaps manual) selection, etc. etc. etc. Use it on its own, and you're likely to build CPU clocks that secretly pick up radio waves from your instrumentation.

The Entities, as described, use only evolution. That's why they're dumb.

If you use your human intelligence to examine the results of your evolutionary algorithm and reject the unusable, crazy or far too situational stuff, then you're already not using pure evolution.
Click to expand...
Click to shrink...
Simulated annealing is a predecessor to swarm intelligence and evolutionary algorithms. It rarely is superior to a particle swarm optimizer, which performs roughly similarly well to a basic genetic algorithm. And if your design is exhibiting behaviors you find undesirable, you've probably not defined your reward function properly. None of the techniques you listed would "avoid" the problem you raised.

No, EAs are not the be-all and end-all solution, but for extremely complex problems, they work very well compared to anything else we have.

Now, if you really want to consider the Entities, their problem is that they're not engaged in "evolution." They're engaged in serial guess-check-try, losing most of the power of parallel experimentation which actual EAs thrive upon.
 
SirKaid said:
Contraception doesn't help the problem. So they stop having babies. Eventually there's the heat death of the universe after they've eaten everything and they have to cannibalize each other again like they did on their home planet. That's a direct fail state.
Click to expand...
Click to shrink...
It does stave off the "We've eaten everything" for a bit though, given that they'd no longer be multiplying exponentially.
 
SirKaid said:
And in return for staving it off a little they get to have less data to play with when it comes time to pool resources in order to try and find the solution. Rather a poor trade for an Entity, don't you think?
Click to expand...
Click to shrink...

Problem is that they're looking for a solution to a problem that has no solution. Short of being able to create energy out of nowhere and nothing or make an actual perpetuum mobile or create or move to an entirely new multiverse, it is impossible to stop or escape entropy and the heat death of the universe. It's like a human desperately looking for the solution to the question of: "how do I turn into a Tyrannosaurus rex before I die?". It's just not possible.
 
Laurelin said:
Problem is that they're looking for a solution to a problem that has no solution. Short of being able to create energy out of nowhere and nothing or make an actual perpetuum mobile or create or move to an entirely new multiverse, it is impossible to stop or escape entropy and the heat death of the universe. It's like a human desperately looking for the solution to the question of: "how do I turn into a Tyrannosaurus rex before I die?". It's just not possible.
Click to expand...
Click to shrink...
The Entities aren't actually trying to stop or surpass entropy, or at least not only that, but are looking for a solution to the problem of what happened last time they exhausted all resources. They once consumed all resources on their home world (including all parallel dimensions), and then ended up in a war that destroyed 99.9999% of their species, where they ended up consuming each other in an orgy of violence. This period scarred them and now they travel the cosmos to find a solution to the problem of what happens to them when they eventually consume all resources again. Reversing entropy is perhaps one solution, by making sure that end never comes, but they are largely looking for a way to just not end up killing each other trying to stay alive at the end.
 
MobiusC said:
The Entities aren't actually trying to stop or surpass entropy, or at least not only that, but are looking for a solution to the problem of what happened last time they exhausted all resources. They once consumed all resources on their home world (including all parallel dimensions), and then ended up in a war that destroyed 99.9999% of their species, where they ended up consuming each other in an orgy of violence. This period scarred them and now they travel the cosmos to find a solution to the problem of what happens to them when they eventually consume all resources again. Reversing entropy is perhaps one solution, by making sure that end never comes, but they are largely looking for a way to just not end up killing each other trying to stay alive at the end.
Click to expand...
Click to shrink...

Well, they'll still run into the same problem doing that, in that they're looking for a solution to a problem that has none. Vast though it may be, the multiverse is finite and unless energy or matter can be generated or created out of nothing, it always will be. So it is inevitable they'll run out of resources because, well, that's the very basic principle upon which the very reality within which they exist was built: it's finite. Unless they find a way to become completely self-sustaining in a way that neither consumes or releases energy and matter, but then they'd be a living perpetuum mobile which to my mind seems impossible to exist. Even if it did, I doubt they'd find it through their cycles, since their cycles are defined by consuming precious finite resources.
 
Laurelin said:
Problem is that they're looking for a solution to a problem that has no solution. Short of being able to create energy out of nowhere and nothing or make an actual perpetuum mobile or create or move to an entirely new multiverse, it is impossible to stop or escape entropy and the heat death of the universe. It's like a human desperately looking for the solution to the question of: "how do I turn into a Tyrannosaurus rex before I die?". It's just not possible.
Click to expand...
Click to shrink...
FTL is impossible. All evidence points to the complete and total impossibility of anything ever moving faster than light. It simply can't be done.

Entities do it casually.

You can say that reversing entropy, or evolving into perpetual motion machines, or whatever solution is proposed to the "let's not cannibalize ourselves again" problem, is impossible. As far as the Entities are concerned, they'll only consider it impossible if they use the entire multiverse as fuel for computation and can't find a way to do it.
 
SirKaid said:
FTL is impossible. All evidence points to the complete and total impossibility of anything ever moving faster than light. It simply can't be done.

Entities do it casually.

You can say that reversing entropy, or evolving into perpetual motion machines, or whatever solution is proposed to the "let's not cannibalize ourselves again" problem, is impossible. As far as the Entities are concerned, they'll only consider it impossible if they use the entire multiverse as fuel for computation and can't find a way to do it.
Click to expand...
Click to shrink...

FTL is impossible in that you can't exceed 300.000km/s, but even then science admits that in theory you can work around this if you contract the space-time so that you're "technically" crossing it faster than lightspeed would allow. Basically, sort of what Vista can do with her powers. The problem is creating something that can actually do that in practice. But it's still possible in theory.

That's still far removed from "make something out of literally nothing".
 
Laurelin said:
FTL is impossible in that you can't exceed 300.000km/s, but even then science admits that in theory you can work around this if you contract the space-time so that you're "technically" crossing it faster than lightspeed would allow. Basically, sort of what Vista can do with her powers. The problem is creating something that can actually do that in practice. But it's still possible in theory.

That's still far removed from "make something out of literally nothing".
Click to expand...
Click to shrink...
Eh, while there are some cheap ways that can technically create FTL travel, they are not exactly the kind of ideas that are a good idea to test.

Something most people don't realize is that the Speed of Light is not some arbitrary value, rather it's the concept that causality itself has a maximum propagation rate. Imagine you had a button and a light, where regardless of where you place the button or light, the light illuminates if and only if the button is pressed. According to physics, if the button and light are one light year apart from each other, the light will illuminate exactly one year after the button is pressed.

tl,dr: While there are some ways FTL may be possible, the sheer condition of having a piece of information end up a light year away less than a year after it was "transmited" would break just about everything we know about causality. And probably make Thermodynamics throw even more of a fit.
 
Reader mode
You must login or register to reply here.
Remove ads
Back
Top