Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Topics - Houshalter

Pages: [1]
1
Biology / Evolution doesn't work?! D:
« on: June 01, 2010, 08:33:19 PM »
Allright, you probably thought I was going to announce I was a creationist or something. No I just wanted to show a few cases where evolution created bizzarre effects. For years farmers bred their livestock for the most productive ones, and with chickens it was usually for egg laying. They would select for the hens that layed to most eggs. Now, years later, the desendents of chickens from these breeding programs have to have their beaks clipped and housed in seperate cages. The reason is because the chickens that layed the most eggs also happened to be more aggressive and fight their way to the top of the pecking order. A researcher tried a new approach to selective breeding on these chickens. By using group selection, after only six generations he had increased the productivity of the birds by more then double and increased the life expectancy by nearly 3 times. (source.)

Once, a swedish researcher tried to increase wheat yields and picked the biggest wheat kernels to plant each year. After only 5 generations he had a strain of wheat that produced 6 giant wheat kernels per stalk.

When insects were damaging fruit trees, farmers expirimented with using cyanide to poison them. They put up giant tents over the whole trees and pumped in cyanide, killing the insects. Plants are immune to cyanide but no animal that uses respiration can be. And in only 5 years or so they had resistant insects. The resistant insects would -- when anything startling happen -- sit very still and hold their breath for half an hour or so.

2
Off Topic / Organic Computers
« on: May 08, 2010, 11:22:29 PM »
This idea always fasinated me, but I think I have a plausible way to actually use evolution in a practical way. Allright, so the idea goes that organisms like bacteria are thousands of times more complex then an average computer. They can store terrabytes in single, short molecules, they can form extremely complex networks, etc. There was an expiriment a while back where they used a bunch of mouse neurons to power a robot. But thats not what I mean. I mean that it might be possible to evolve an organism that could do rudimentary computational tasks, and could be put with thousands of other organisms like it to create efficient biological computers if you will. How would you do that you ask? Well I'm not quite sure, but my idea is that you choose a canidate organism that looks like it could evolve the ability your looking for. Then you let it multiply. It can be stored in a small area like a film canister, or even smaller probably. The idea is to a) create a large population base to start the expirement with,  select for the fastest reproducing ones, and c)select for the ones that survive/reproduce best under the lab conditions you have provided. The faster they reproduce, the faster the expiriment will go, the better, so this is somewhat important. I immagine this would have to be pretty large scale to have any chance of success, so maybe a giant warehouse full of thousands of these containers breeding the organism. Since its impractical to have hundreds of scientists going down each aisle and testing each one manually, a machine or robot would have to do it which isn't that far fetched an idea, although it might be a little expensive to obtain one. Then you advance evolution in stages. The first stage you do what I already said, select for quick reproducion and build a large population base. The second stage you slowly introduce an electric current into the enviroment and create a tolerance, maybe even adaptions to use the current to its advantage. Then you select for efficiently using the current, picking the ones that let the current flow through the container the best. Now you have an organism that is adpated to elecricity somewhat. From here you diverge into evolve seperate componets of the computers, mabe one at a time or dividing resources to do multiple ones. I don't know much about electronics, but I think you would start selecting for properties of basic computer components. Logic gates, transistors, capacitors, retaining electric charges or some kind of chemical memory, pulsing behaviors, etc. I know that other expirements had success in evolving bacteria such as in nylonasse bacteria or E. coli, abiet those took a long time to get results. The advantage here is that your breeding it in masses and in a more industrial way, not as an expiriment. If you could "grow" a computer, then you could push the cost of GFLOPS way down.

How does this relate to Darwinbots? I don't know, I guess it involves evolution and computers, plus if it worked we could run DB on it.

3
Off Topic / Mechanical Neural Network
« on: May 05, 2010, 06:56:42 AM »
I'm not particularly fond of neural networks since their not universal turing, but in the real world they do have their uses. I was wondering if it would be possible to make a neural network out of paper. You might laugh, but I've seen some crazy things built out of paper inculding clocks and automata, plus I've seen old mechanical adding machines before. My problem is trying to find a simple way to do addition and multiplication for the weights and thresholds. Any ideas? Am I crazy?

4
Bot Tavern / How does this bot possibly reproduce
« on: March 06, 2010, 11:35:48 AM »
How does this bot reproduce? Its alga minimalis afer 5 million cycles and 30 mutations. It has the number 300 hundred and an inc but it only reproduces occasionly, but it looks like it should reproduce every cycle or not at all. I don't get how the code works to allow this.

Code: [Select]
dist

 start
 300 or


 start
 inc


 else

5
Off Topic / Micro evolution vs Macro evolution
« on: February 20, 2010, 12:08:57 PM »
I recently dicovered that there are supposedly two types of evolution. Anti-evolutionists use these to dispute evolution. Now I know both of these are technically outdated terms and that even if they weren't we've observed both of them in real life. But have you ever seen Macro evolution in Darwinbots. I've seen the bots tweak their behaviours but I've never seen them radically change them or add brand new behaviours. By the way, Micro evolution is just an organism changing slightly while macro evolution is an organism becoming an entirely new species. In sexually reroducing organisms this usaually means it can now longer produce viable offspring with the species it diverged from.

6
Interesting behaviour bots / Sybots V1 (Houshalter)
« on: February 19, 2010, 11:20:59 PM »
Allright this was inspired but not plagerized from Sammeh but I still should give him some of the credit. The idea is to force two different bots to work together or die. Sybot1 can hunt but can't reproduce. Sybot2 can reproduce and help Sybot1 to reproduce but it can't hunt. Sybot2 seeks out Sybot1s to tie to. You can make the bots more balanced by giving the bots extra abilities. For example, if Sybot2 is becoming useless because Sybot1 has learned how to reproduce on its own, you might have to change Sybot2 to give it extra ablities like venom, poison, shell, etc, so that it will not be completely useless and therefore will be selected for.

Sybot1:
Code: [Select]
'the important feeding gene
cond
*.eye5 0 >
*.refeye *.myeye !=
*.in1 77 !=
start
*.eye5 40 <
*.up 10 add .up store
not
*.refvelup 1 add .up store
true
*.refxpos *.refypos angle .setaim store
-1 .shoot store
stop

'set my out1 to 88 so sybot2 can find me
cond
*.out1 88 !=
start
88 .out1 store
stop

'show im not tied
cond
*.numties 0 =
*.out2 33 !=
start
33 .out2 store
stop

'show im tied
cond
*.numties 0 >
*.out2 22 !=
start
22 .out2 store
stop

'turn if i see nothing
cond
*.eye5 0 =
start
314 rnd .aimdx store
stop

'turn from allies
start
*.refeye *.myeye =
*.in1 77 = or
314 rnd .aimdx store
true
stop

'share the spoils :)
cond
*.multi 1 =
*.nrg *.trefnrg !%=
start
41 .sharenrg store
stop

'share the waste :(
cond
*.waste 100 >
start
1 .sharewaste store
stop


Sybot2:
Code: [Select]
'find a sybot1
cond
*.in1 88 =
*.in2 33 =
*.numties 0 =
start
.tie inc .tie inc
.refvelup .up store
stop

'store 77 in my out1
cond
*.out1 77 !=
start
77 .out1 store
stop

'also to help in the search
cond
*.numties 0 =
*.eye5 0 =
start
314 rnd .aimdx store
stop

'dump waste
cond
*.waste 100 >
start
100 .shootval store
-4 .shootstore
stop

'store 2 in readtie
cond
*.multi 1 =
*.readtie 2 !=
start
.readtie inc .readtie inc
stop

'turn away if i see something
cond
*.eye5 0 !=
*.multi 1 !=
start
314 rnd .aimdx store
stop

'collectively reproduce
cond
*.trefnrg *.nrg add 6000 >
*.multi 1 =
start
2 .tienum store
.repro .tieloc store
50 .tieval store
50 .repro store
stop

'manage body
cond
*.body 20 <
start
20 .strbody store
stop

Bugs:
1. Sybot1 easily becomes a Big Bertha.
2. Sybot2 seems to have a diminishing population, its only role is for reproduction anyways.
3. There is some weired glitches with the tie physics that screw up reproduction and a few glitches that cause just strange behavior, like the entire dna of every Sybot2 to not execute. The biggest glitch I can find is that the multi bot sharing isn't working for any explainable reason.

7
Biology / 300,000 birds
« on: January 25, 2010, 06:48:02 PM »
Have any of you ever seen this?

8
Off Topic / Go back in time...
« on: January 24, 2010, 01:22:52 AM »
This is wierd. You know how google.com is always listed as a user. Well this site was also listed as a user, so i was curious and typed it in to my url and went here. Apparently this site has been making copies of the darwinbots homepage since 2006. Kind of interesting to see what you guys were doing back then.

9
Bug reports / What is this thing
« on: January 23, 2010, 06:21:38 PM »
I was running a sim when the strangest bot ive ever seen appeared out of no where. The sim i was running was with semi-zerobots. I used a program to generate them. They started as 256 lines, each line with 40 random numbers between 1 and 1000. Anyways after 13,400 cycles, after the bots had begun to reproduce and went through 2 to 3 generations, this thing appeared. Its apperently tied to 8 other bots and i have no idea how. I thought a bot could only have up to 3 ties. Even if it can have more, what are the chances a 2nd generation zerobot has advanced tie feeding genes that let it track down 8 other bots and tie to them, not using the same tie number twice. Its not its offspring it somehow managed to harden ties to either. Stranger yet, i checked its philogeny and apparently its related to a veggie, even though the veggies don't sexually reproduce (so it couldn't have somehow impregnated a random bot with a .sexrepro gene). It changes colors frequently (probably because frequent mutations), and when its off the screen the ties disapear (maybe their not ties at all, but what are they?)
PS: any way to insert pictures better?

EDIT:  Oh this is important to. The dynamic costs aren't working either. Wierd. Occasionly it adusts the costx by several decimal points and then does nothing for a long time even though sensitivity is on max and the population is 3 times larger then what it is set for.

EDIT2: Now its inside of a veggie. It still has all those bots on a leash. Is it even possible to get inside another bot . I thought the bots collided not passed through each other.

10
Suggestions / Hyperspeed Mode
« on: January 22, 2010, 08:31:50 PM »
The no. one constraint for evolution in darwinbots is speed, closley followed by population size. Actually population size might be the biggest constraint, but the only reason you can't have lots of bots is because it slows it down, hense speed is important. I recently came across conways game of life, which im sure most of you are familiar with, and somehow i came across this. Apparently theres away to make conways game run really really fast. I was wondering if something like that would or could apply to darwinbots. I don't really understand it though. Its kind of beyond me with all the technical terms.

11
Bot Tavern / Is there something wrong with this
« on: January 11, 2010, 07:43:30 PM »
 Look at this gene
Code: [Select]
start
*.reffixed 0 =
*.refnrg 10000 > and
*.eye5 40 < or
*.refshell 0 > or
*.memval *.refmulti 100 mult add *.refkills 2 div add *.refpoison 1000 div add *.refshoot add *.refeye add *.reftie add *.avgfitness <= or
*.totalbots *.totalmyspecies sub 100 > and or
157 .aimdx store
not
*.refxpos *.refypos angle .setaim store
stop
It <you>should</you> make the bot turn when theres nothing worth looking at and set its focus on the bot in focus otherwise. But no matter what I try it just wants to turn around endlessly  . If your wondering what kind of bot this is for, its for a sheperd bot im creating. By the way, .avgfitness is a custom variable which refers to the average fitness of the last ten bots hes seen.

Edit: somethings wrong with the post system, I tell it to underline a word and it puts you's on both sides of it in <> signs then it auto spell checks the you's into you's.

12
Suggestions / Pwnage shot
« on: January 09, 2010, 06:30:22 PM »
Would it be possible for the bots to have a different type of shot like a -9 shot that instead of hurting a bots memory location or taking energy/body from it, but would just do damage to the bot being shot at with no benifit to the bot that shot it. It could do more damage to a bots shell than a normal shot or also deal damage to a bots slime level as well. Maybe it could limit the amount of energy another bot could use on a cycle rather than actually taking energy from it. Just a crazy idea. In real life most organisms kill or do some kind of damage to another organism before they actually eat it.

13
Newbie / Sexual reproduction
« on: January 02, 2010, 08:33:40 PM »
This is a noob question since im sure someones tried something like this before, but what if we sped up evolution by putting all the best human created bots together in a simulation and forcing them to sexually reproduce. That way all the best genes would mix together and with survival of the fittest the best combination would survive (hopefully). You would just have to get a bunch of bots and change their .repro to .sexrepro and make a veggie that would target bots and fire -8 shots at them. You could make the costs low so the mutant hybrids would have some chance at survival. Mutations would hopefully modify the genes so they work together with each other better, and since its sexual reproduction only, the best genes would be spread through the population quickly. Just wondering if anyones tried anything like that and what the results were.

14
Bug reports / Can't save any sims
« on: December 21, 2009, 11:33:30 PM »
I had version 2.44.03 and I couldn't save or autosave any simulations or load other peoples either. All these crazy -p folders started poping up in strange places so I deleted them all and got verson 2.44.04 and it didn't change anything. I did a search on my computer for -p files and I found a 853 of them. However only a dozen of them were those crazy folders and most of them were already in the recycle bin. Anyways the problem isn't the -p folders, I can live with that. I just want to be able to save and load sims.

Pages: [1]