High IQ Reproductive Turtles. Matrix nightmare.

One disk. A user programs one disk and sets it into a turtle. The disk contains the instructions for the turtle to go mine, refuel itself (with a starting value of 50,000 which is more than enough for a turtle), create furnaces, create machines to support energy demand and keep growing, crafting new turtles and inserting the root disk in them, making more miners and consequently more builders, maybe even adding a random selection of A.I.s . Complete simulation of a living entity. This is the next frontier for the turtle developers. It's gonna be the next big brother for ComputerCraft servers. People are gonna watch machines grow and take over, building structures. Sounds like a fun trip. I've already seen programs capable of lots and lots of features, but I think if the developers of these programs got together, it's a possibility that soon we will see iron houses and migrating groups of turtles. Maybe, if this happens, the mod developers might even add this to world generation, who knows? Is this a crazy enough idea? Think it's ever going to happen? Are you willing to help on the A.I.?

No one will write code for you. This forum is for help on code that you have written.

Dlcruz129, on 22 December 2012 - 03:35 PM said:

No one will write code for you. This forum is for help on code that you have written.

Settle down.
I didn't ask to write it for me, did I? I asked if such idea is actually possible. I also wanted to throw the idea of an advanced turtle out there. What happens next is up to the developer. Don't think I want to take any of the credit.

marcelo8576, on 22 December 2012 - 03:41 PM said:

Dlcruz129, on 22 December 2012 - 03:35 PM said:

No one will write code for you. This forum is for help on code that you have written.

Settle down.
I didn't ask to write it for me, did I? I asked if such idea is actually possible. I also wanted to throw the idea of an advanced turtle out there. What happens next is up to the developer. Don't think I want to take any of the credit.

He's correct though, this is the wrong place to post this type of topic. It should be in general. I've already reported it so it can be moved.

Dlcruz129, on 22 December 2012 - 03:35 PM said:

No one will write code for you. This forum is for help on code that you have written.

Whah? He opened a very interesting subject and really didn't ask for anything at all. And there are many, many exceptions on that rule. I bet that this place is suited for discussion about the technical part of CC as well. I'm starting to notice how little u usually contribute to conversations.

On topic. This is quite a popular subject lately. There was even a competition on something like this: http://www.computercraft.info/forums2/index.php?/topic/4462-competition
And a similar question: http://www.computercraft.info/forums2/index.php?/topic/7219-skynet-turtles-theory
I must tell you, it's a though challenge. There are so many factors involved in this, it's hard to get a grasp of it. And I think that in order to finish this project in a feasible amount of time, you'd have to start on a lot of theory first.

And on the subject of moving this, I think it's slightly better to have this in General, but I don't think it's a big deal (look at a lot of previous posts like the skynet one). But Dlcruz definitely missed the point, he wasn't asking for any code.

It can be done, it has been done, it…well, it tends to not end up being as cool as it sounds.

Technically, feasible. Except for the part where the turtles evolve into "something more", they just become a crapload of turtles, with some kind of inventory/gps array at the center.

ChunLing, on 22 December 2012 - 05:41 PM said:

It can be done, it has been done, it…well, it tends to not end up being as cool as it sounds.

Technically, feasible. Except for the part where the turtles evolve into "something more", they just become a crapload of turtles, with some kind of inventory/gps array at the center.

I think it's at least a very difficult challenge. If it has already been done, mind sharing the topic?

I think that people tend not to post their final codes for projects of this nature. There are probably two reasons for this. First, it isn't generally useful. That is to say, a fully autonomous turtle farm tends to just wreck an area for most other purposes, by nature it doesn't give enough control to the user over the outcome. Second, it's actually rather destructive. I think that most people who finish coding/testing something of this nature decide that it not only wouldn't have a lot of value to anybody, but would kinda be a bad thing to post.

It is clearly possible, but I flirted with the idea after I had run dry of ways to make my turtles actually more useful and then decided it wasn't a very good idea pretty quickly. So I'm biased into thinking that most people that think about it long enough to have it substantially coded also end up thinking it a bad idea.

I know someone made an automated turtle factory that did 64 turtles and then decided that was already too many and didn't go further. Some pretty cool screenshots, but I had to agree with the "too many" assessment.

Ive already made one. Its more of a Virtual Intel. though. Not AI.

Here is a video of something that is close http://www.youtube.com/watch?v=oanR7y4erbU&feature=youtu.be&t=24m1s

A horrifying vision, but more an illustration of the final result than a demonstration of a "seed" program.

I thought about turtles attacking players before, sadly turtles are mined up too quickly for that xD

You got to surprise the player, and catch them in a boat or cart.

the surprise tactic is actually fairly easy to pull off. you just mine underground and dig out what they are standing on, then if you can do it fast enough you can place a lava bucket in there. my bases all have gaps in the walls and floors between each room for circuitry, hidden sensors, tunnels, traps etc so I can have turtles roaming under my base but I doubt I would implement it. the hidden machines would get in the way

Well, you don't try to catch them inside your base. But yes, if you're going to attack a player then bombing, lava buckets, and attacking from underneath all make a good deal more sense than using the attack function. Unless you catch them in a boat…though I don't know how well that works in current versions. Needs testing :lol:/>

making a tunneling program is quite easy really. you can make vast underground networks…. tempting

ChunLing, on 26 December 2012 - 02:06 PM said:

You got to surprise the player, and catch them in a boat or cart.

cause they can't use a pick in there? :P/> and well surprise is quite useless against nanosuited people

No, it's because of how damage to a boat or cart gets transferred to a player (or used to). And of course being in a cart restricts seeing/breaking things below you.

Of course, if you're talking about killing nanosuited players, then probably nothing is much use.

you can kill them. you need balkon's weapon mod and an insane trap. it still takes a while to kill them but it works. you put them in a 1x2 block of reinforced stone filled with lava, close the top block over their heads (which for some reason invariably causes them to break that block rather than the sides), you have at least 20 dispensors all firing cannon balls at this cell as they can do damage through a single layer of reinforced stone and eventually weakening him. I have not tested the final thing that will make it work. when in lava or water you dig slow and it is already hard to break reinforced stone so it will take him a long time to break his way out… you could try making a pillar of stone and a piston system pushing it down to replace the broken blocks

anyway I just noticed that I am rambling on about one of my old projects again which is not related to turtle assassins at all. in fact you might as well just envelop him in a zapping forcefield and wait for him to eventually die…

quick question. can a turtle be broken while moving? if not envelop him in a cell of turtles which keep moving down with mining turtles beneath them to mine out a path

It's one of the key ways to stop a runaway turtle (the other's being terminate/shutdown key commands).

This would be amazing to see, also, you can have evolution within the turtles by simply creating a script that will generate random code, it'll test to see if it works, if it doesn't work then it tries again, eventually it will work. Through natural selection such as reproduction and death of turtles (by them hitting each other) they'll eventually through MILLIONS of years (maybe billions due to processing power) the turtles might develop intelligence.

i believe this is highly possible, just i would think a group of coders would have to get together to actually write it. its basically an os with many features and things for what we call an AI

AI? Classified as an OS? :)/> A bit strange.
I don't believe many people would have the patience for writing highly complicated AI. :P/> just my opinion though.

which is why its not out there yet. people have time to write all the os' in the world but not one ai yet for computers and turtles (at least to my knowledge)

Systems don't develop intelligence on their own, the intelligence (ability to predictively adapt) has to be built in.

There is debate amongst biologists as to whether a system that allows adaptation can eventually develop intelligence because of their theological commitment to the idea that intelligence can evolve naturally. But there is no debate among serious cyberneticists about this being possible. It simply isn't. An intelligent system can become more intelligent, but a non-intelligent system cannot develop intelligence of itself. This is because predictive adaptability is different from mere prediction or adaptation, so even if you already have an adaptive system, the simple addition of prediction does not yield predictive adaptation. Prediction must develop orientation in order to become effective as a directive mechanism. And the primary orientation is maladaptation, that is, the self-fulfilling prophecy of doom.

Even nominally intelligent systems frequently devolve into predictively maladaptive systems, the tendency for predictive orientation towards maladaptation is so strong. In the ordinary course of events, a simply adaptive system that started to develop predictive orientation would abandon it long before the predictive orientation could progress to become adaptive rather than maladaptive.

It would require some outside influence on the system, to prefer predictively oriented systems despite their being maladaptive, while also encouraging them to develop towards predictive adaptation. This exterior influence itself would have to be predictively adaptive (or under the control of a predictively adaptive process) to the extent that it was functionally capable of distinguishing and promoting development towards predictive adaptation despite the intermediate predictive maladaptive stages of development, which an adaptive selection procedure would terminate.

:)/> However, you're missing my point. I said "natural selection such as reproduction and death of turtles" it may develop intelligence, this means through Mutation of some sort meaning the code will just randomly change. :P/> The chances of intelligence rising from that is extremely little, however, I'd find it amusing to watch something weird spiral off of "Evolutionary programming" using random variables.

As stated, the process you describe cannot produce intelligence.

The debate among biologists (and associated specialties) really breaks down to those who believe that the impossibility of intelligence arising from non-intelligent processes implies that humans are not intelligent, but only subject to a biologically imposed delusion (like a computer that has been programmed to claim to be a person), and those that do not accept this.

There are various sound reasons for rejecting the idea that the lack of any possible biological basis for intelligence means that humans are not intelligent. It is entirely reasonable to take the position that only some humans are intelligent, for reasons that have nothing to do with biology. Of course, there are plenty more unsound reasons than sound reasons, particularly when sampling the reasons of a group of humans.

If you make the turtles intelligent in the first place, capable of realistically imagining the results of their available courses of actions and choosing actions based on which results are the most "beneficial" (a prerequisite concept that you'll have to program into them somehow), then you can go a bit further and make it so that their intelligence improves (so that their imaginations become more realistic based on past experience falsifying previous expectations). But you have to put them over that threshhold first. A system that is not already intelligent is not going to become intelligent without it's development being directed by an intelligent system.

ChunLing, on 30 December 2012 - 12:09 AM said:

As stated, the process you describe cannot produce intelligence.

The debate among biologists (and associated specialties) really breaks down to those who believe that the impossibility of intelligence arising from non-intelligent processes implies that humans are not intelligent, but only subject to a biologically imposed delusion (like a computer that has been programmed to claim to be a person), and those that do not accept this.

There are various sound reasons for rejecting the idea that the lack of any possible biological basis for intelligence means that humans are not intelligent. It is entirely reasonable to take the position that only some humans are intelligent, for reasons that have nothing to do with biology. Of course, there are plenty more unsound reasons than sound reasons, particularly when sampling the reasons of a group of humans.

If you make the turtles intelligent in the first place, capable of realistically imagining the results of their available courses of actions and choosing actions based on which results are the most "beneficial" (a prerequisite concept that you'll have to program into them somehow), then you can go a bit further and make it so that their intelligence improves (so that their imaginations become more realistic based on past experience falsifying previous expectations). But you have to put them over that threshhold first. A system that is not already intelligent is not going to become intelligent without it's development being directed by an intelligent system.

You are incredibly good at typing a lot, but very poor at explaining yourself.

If you make the turtles intelligent in the first place, capable of realistically imagining the results of their available courses of actions and choosing actions based on which results are the most "beneficial" (a prerequisite concept that you'll have to program into them somehow), then you can go a bit further and make it so that their intelligence improves (so that their imaginations become more realistic based on past experience falsifying previous expectations). But you have to put them over that threshhold first. A system that is not already intelligent is not going to become intelligent without it's development being directed by an intelligent system.

That's where genetics come in, in both software development and biology. Genetics is not an 'intelligent system'.

Which is exactly why it cannot produce intelligence absent manipulation by an already intelligent system. The precursor development stages of intelligent systems are predictively maladaptive by nature, thus an unintelligent system like genetics would eliminate them.

As for explaining myself, that was not my intent. I'm explaining why non-intelligent systems cannot develop intelligence on their own.

Technically if you were to randomly mutate code, with an infinite number of turtles you'd eventually end up with an intelligent turtle. (Under the assumption that it is indeed possible to program an intelligent turtle)

You mean under the assumption that it is indeed possible for intelligence to arise from a process that is neither intelligent nor under intelligent direction.

I've seen the Internet, so I do have some idea of just how good the evidence is that intelligence arose from a random process…granting the assumption that's what produced humans.