Slashdot Mirror
Scientists create digital bug-life
berniecase wrote to us with coverage of the creation of digital life, in a computer-petri dish. The bugs succed by getting more processing time and thus living more, and reproducing more. This type of experiment has been going on for while, which sparks the debate of digital life? Is it coming? Is it already here? Will it never arrive?
The elegance of this is just great:
"You can see why Microsoft is interested in robust languages because theirs is not," Adami said.
Tee hee hee!
simulations on the creation of actual amino acids in ocean water have been going on for a while now.
if the scientific fellows figure out how simple life was created initially the rest of the stuff is a piece of cake.
signatures are for fools with hands
I read about a proposed project several years ago that would have done essentially the same thing, but allow the 'bugs' to migrate from computer to computer over the internet, executing on a portable virtual machine.
It was expected that as CPU usage changed during the day, the bugs would move around the world to the point of lowest usage (greatest resources).
The idea was to let it run for a very long time, on a lot of computers, and see what sort of bugs were produced.
I never heard anything more about this. Perhaps the thought of having random code travelling from computer to computer over the Internet did not make it politically feasibile for many to host this experiment.
-josh
Just what we need... Hello World that grows it's own control-C intercept.
I thought tax time was when Excel DID generate random code that thrived on numbers until the most robust got filed?
There's just too many one liners waiting in that article.
I downloaded life for my palm pilot a few weeks ago, and have been doing similar experiements, man those crazy digital bugs, sometimes they look like a glider and go right across the screen, other times they look like a snowflake, then they just spin over and over. I wish I would have known; I could receive grant money for this.
It seems that all this really shows is the result of intelligent design in the "bugs", and programming them beforehand with an idea of the rules of their environment. The Morris Worm competed with other programs for processor time too. I guess I just don't see the novelty in this article.
Today: bugs
Tomorrow: cyborgs
Are you listening John Connor???
I can see why Microsoft is interested, they got plenty of digital bugs to experiment with.
EJB
see subject.
hasn't microsoft been creating "digital bugs" for years? *duck*
I was a little disappointed that the authors didn't mention Tierra, the a-life system that Avida was based on. Tierra has been around since the 80s and can be found at: the Tierra homepage.
The programs that evolve in Tierra get pretty interesting, and include the evolution of parasites and virus programs. Pretty neat stuff!
has anyone ever used the Tierra artificial life software? I downloaded it years ago; hardly knew what to do with it, as I knew nothing about C.. got it to compile, but it always locked up my 386...
This article piqued my interest in the software again, and I found some info on it, for those interested...
links:
Web page
FTP Site
Documentation
GP, the "next step" in some ways, is also years old.
However, it won't replace human programmers anytime soon. Genetic algorithms and genetic programming require a lot of work.
I'm currently working on a GP system that uses XML to store human readable intermediate results (such as doing dumps of whole genertations), as well as for distribution of individuals for evaluation in huge distributed setups, and it certainly takes time.
But most importantly: It takes a lot of time to evolve complex systems, and a lot of time to specify a fitness functions that "works". If you aren't precise enough, the system will inherently exploit any weaknesses in your model, and may achieve high fitness values by "cheating" (relying on unintentional quirks in your model).
This problem grows exponentially with the increasing complexity of the problem you want to solve.
Some suggested solutions involve using GP/GA's to "police their own". You could for instance conceivably evolve GP's that rate music almost like you. It would take a lot of time for you to put together a training set. But once you've done that, you could use the music rating GA/GP as a fitness function for a music composing GA/GP.
After a while, components based GA/GP will probably be more popular: To avoid having to expose humans to the evolution all of the time (which is very time consuming), you limit the problem to evolving a decent fitness function.
In the music composition problem above, it would take you a lot less time to put together a set of examples of music you like and dislike, and evolve a GP/GA that rates them approximately like you, than it would be for you to rate the output of a GA/GP that tries to evolve composition skills from scratch.
You'd still have to check the results from the composition GA/GP from time to time, but the GA/GP you've evolved to rate music would sort out potentially pleasing music from complete crap.
GA/GP has a lot of potential. The problem is to find good ways of reducing the time spent on evaluating the programs.
Microsoft's been producing digital bugs in all of their software for years :)
-- My neighbors dog has a four inch clit.
The strange and scary part of this kind of technology is the word mutations. Mutations are by definition unpredictable. Maybe I've read one too many science fiction novels, but the thought of a totally unpredictable, completely new life form scares me.
:)
All of the life forms currently on this planet have had millions of years to evolove into a stable relationship of population dynamics. What happens when a brand new life form is thrown into the mix? Will our steady-state system be pushed into an unstable region? What then? If you reward a life form for one particular action without instilling a natural balance between risk and return, that life form could evolove into something frightening very rapidly.
Hey... I have an idea. Lets make a movie of this... it can star Keanu Reeves because he's such a talanted actor.
--
-- First post (by a female living in a state that begins with M and does not end in a vowel with a birthday that falls
Will it never arrive?
:-)
Well, since it's being developed at Michigan State, the digital representation of life will arrive with about 10,000 others, late at night, with a bottle of beer in one hand and a torch in the other hand. Then, about 30 minutes later into the simulation, digital representations of law enforcement will show up and begin showering the digital life with digital representations of tear gas. The simulation will then go into chaos with the digital life setting fire to the digital law enforcement's vehicles.
Very realistic
Tom Ray did this about 8 years ago. His results were quite spectacular: Tierra's "critters" spontaneously developed self-replication, viruses, predation, symbiosis, and more! Here's a link to the Tierra Homepage.
tamagotchi's don't adapt themselves until the most robust code wins. and they can't take care of themselves. and they're just "toys," as in, the dinky little toys made to entertain children, or bored high schoolers (as I once was.. bored, that is. still a high schooler.), and they don't feed themselves, and....
Insert mind here.
Heh this is just what we need. Windows 2000(2050?) to spawn and take over computers run by Linux. It's the final step to world domination!!!! So, who is going to develop some digital Raid?
Sounds like the "evolution" of windows.... from 3.1 to 95, 98 NT 3.5 to win2k...
Unix is user friendly... it just chooses friends selectively!!
Unix is user friendly... it just chooses it's friends selectively!!
"The proposed project will create a very large, complex and inter-connected region of cyberspace that will be inoculated with digital organisms which will be allowed to evolve freely through natural selection. The objective is to set off a digital analog to the Cambrian explosion of diversity, in which multi-cellular digital organisms (parallel processes) will spontaneously increase in diversity and complexity. If successful, this evolutionary process will allow us to find the natural form of parallel and distributed processes, and will generate extremely complex digital information processes that fully utilize the capacities inherent in our parallel and networked hardware. The project will be funded through the donation of spare CPU cycles from thousands of machines connected to the net, by running the reserve as a low priority background process on participating nodes." (Taken from http://www.hip.atr.co.jp/~ray/pubs/reserves/node1. html ).
i ntro.html claims to have taken over the concept, but I miss the distributed computing that Tierra ambitionned.
Tierra http://www.hip.atr.co.jp/~ray/tierra/tierra.html really looked awesomely exciting, but it seems to have remained confined to a relatively small group of researchers.
Avida http://www.krl.caltech.edu/~charles/avida/manual/
Weird...misread part of Deborah's post, thought that instead of of "population dynamics" she'd written "copulation dynamics"....
"How many light bulbs does it take to change a person?" --BMcC-->
I'm sick of these ai experiments called humans. They simply do what they're programmed to do. Tamagotchis running on brains.
. . . . . . .
may u!sh 2 sm!le at dz!z bad nn.!m!tat!ion
Now, if I could only see this happening in real life, that would be nice. I'm still waiting for mutations and other life forms to evlove.
That's my 1/50 of $1.00 US
JM
Big Brother is watching, vote Libertarian!!
--Justin Mitchell
"2nd Place is a fancy word for losing" --Bender (Futurama)
Great! No it is finally possible to generate decent programs that consumes all CPU cycles without performing any usefull work. MS is far behind... /Nisse Ghandie
This is akin to mathematician-novelist Rucker's concept of "boppers" a-life as I understand it. To see Rucker's home page, follow this link.
"How many light bulbs does it take to change a person?" --BMcC-->
It's only a matter of time....
Big, well-written and funny: called The Hitchhiker's Guide to Evolutionary Computation (FAQ in comp.ai.genetic). Here is the relevant extact:
t ic/ as the files: part1 to part6. The FAQ may also be retrieved by e-mail from . Send a message to the mail-server with "help" and "index" in the body on separate lines for more information.
t ec.ps.gz
Obtaining copies of this guide
This FAQ is available between postings on rtfm.mit.edu:/pub/usenet/news.answers/ai-faq/gene
A PostScript version is also available. This looks really crisp (using boldface, italics, etc.), and is available for those who prefer offline reading. Get it from ENCORE (See Q15.3) in file FAQ/hhgtec.ps.gz (the ASCII text versions are in the same directory too). In Germany, its also available from the SyS ftp-server: lumpi.informatik.uni-dortmund.de:/pub/EA/docs/hhg
ENCORE is a set of FTP sites, including
ftp://ftp.krl.caltech.edu:/pub/EC/Welcome.html
ftp://ftp.cs.wayne.edu:/pub/EC/Welcome.htm
ftp://coast.cs.purdue.edu:/pub/EC/Welcome.html
and others
Kaa
Kaa
Kaa's Law: In any sufficiently large group of people most are idiots.
Has anybody succesfully compiled the source of the Beagle-X11 part of Tierra?
TIA
Sendy
GNU guru and mainframe hacker
So is Microsoft giving money to this noble cause because some forward thinking person there thinks open source might stand a chance in the very long run, and that the only way to beat socially evolving programs would be to hit them with human independent genetic programming?
So we'd be programming against a wave of IMPlike bots spewing new versions of office etc all the time?
Perhaps open source people will then see the benefits of the biological side of programming, and a whole new chapter of the history of programming will begin.
Then again, why is MS giving the money over? would there be other reasons? Am I being paranoid?
Let's say tie in the training process in a distributed effort? Where everyone has the critter on the screen to interact with when you are really bored according to a bizzare distributed training schedule.
I wonder how such a multi-parented baby will behave... mmm...
"You can see why Microsoft is interested in robust languages because theirs is not," Adami said.
Damn....Talk about biting the hand that feeds you. Jeez!
"Why can you randomly change instructions and these things tend to survive?" Adler said. "If you went in and did that with Excel, the results wouldn't be good."
Yeah! Take another bite of that hand... heh
Possible applications might include tougher operating systems, programming languages, applications and virtual machines.
Great. Just what we need....Microsoft Windows with a bad temper. "WHAT? You deleted parts of my REGISTRY? Watch this Blue-Screen, Asshole."
Or how about a C compiler that decides it wants to play a joke on you, and deletes a block of your code.
Don't bother fighting them, because the suckers will adapt, or mutate or some crap, and piss you off even more. (I'd hate to see them put that stuff into MS Bob. That would be scary.)
-- Give him Head? Be a Beacon?
-- Give him Head? Be a Beacon? :P)
(If you can't figure out how to E-Mail me, Don't.
So will this software be banned in Kansas? Or will A-Life pages just have to put disclaimer on the front page sending people in Kansas to yahoo or something.
Conscience is the inner voice which warns us that someone may be looking.
-- H. L. Mencken
If you can't tell whether its alive or not, maybe at that point its alive.
Don't forget that even the most convoluted and brilliant Life configuration is only a speck next to the obscenely complicated miracle that is human life. We've got a loong way to go.
~ Give me 101 plastic soldiers, and I will conquer the world.
I submitted this a few days ago, with a couple interesting links:
Here is the researchers web site.
and Here is a page related to the Nature article which includes source code for the experiments. They ran this on a Linux Beowolf cluster.
But seriously, does this thing, when evolving, write new human readable code or new binary code? The idea of software with no source code known to or readable by anybody is more than a little bit troubling.
I see even classic Slashdot is now pretty much unusable on dial up anymore.
Project Von Neumann is apparently some effort to create and artificial life game. Dunno much about it, doesn't look like much coding has been done. Take a look for yourself.
"Microsoft impressed with research"
:)
Welp, I guess we can expect Microsoft Bob 2000: Intelligent Life to be released sometime in 2001!
Never underestimate the power of stupidity in large groups
LOAD "SIG",8,1
LOADING...
READY.
RUN
What's going to happen when some high school teacher in the great state of Kansas decides to bring this up to his class?
is here. Interestingly, he's a Microbiology, Zoology, and Crop & Soil Sciences professor, and not in the EE or Computer Science departments.
Apparantly, Michigan State has formed a Computational Biology Group since I was a student there. The group looks heavily weighted towards natural science types, with only two computer professors, Dr. Pramanik and Dr. Punch on board. I learned both classical AI and GAs at MSU from one of Dr. Punch's classes -- he's a very good professor, intelligent, a good teacher, and an all-around nice guy.
Thought you could...
A long time ago in a far away galaxy I was able to compile Tierra in a Sparc. Never tried the Windows version. The Unix stuff compiled and run fine out of the (then) very long ftp session. Eventually the sysadmin started to pay attention to my use of disk space and cpu cycles and told me to stop. Pity. :)
From tiera-3.12 (1992) documentation:
The C source code creates a virtual computer and its operating system,
whose architecture has been designed in such a way that the executable
machine codes are evolvable. This means that the machine code can be
mutated (by flipping bits at random) or recombined (by swapping
segments of code between algorithms), and the resulting code remains
functional enough of the time for natural (or presumably artificial)
selection to be able to improve the code over time...
Think of users modifying source as sort of a nearly continuous trickle of small changes and the developers making changes (adding features, etc.) as sort of a discrete large jump. (I forget the analogous terms in evolutionary theory.)
The closed source model doesn't allow these small mutations - changes are only made in the large jump fashion, or by products dying because users discard them.
That's why Microsoft is interested in this. Since they're a closed-source vendor, they want to automate the "many eyes making bugs shallow" part without opening their source.
Isn't that just Terra? Still seems quite old to me... (And the comment about Microsoft's 'language' anot being robust doesn't help their credibility. What is Microsoft's language, anyway?) - Rob
There have been experiments of this type going on for more than a decade, including Darwinist programming whereby certain algorithms that are more efficient survive. Since this is all a digital SIMULATION we must remember Searle's caution about such experiments; a digital simulation of life by AI practitioners or biologists is not life anymore than a digital simulation of a tornado by meteorologists is a tornado. "No funny tag-line
just = (My)Opinion.toCents();
Now AI didn't kept his promises. Even the achievement of decent chess playing (and in general board game), was mostly achieved by brute force (one still can't make a decent program with a 8bit 1 Mhz and 16 KB).
Nowadays, there seems to be a comparable hype with respect to "neural networks", "genetic algorithms", "genetic programming", "fuzzy logic", "simulated annealing", etc... Now, all these methods are very good at spotting a good solution in huge research space, they are also very limited when it comes to anything relative to intelligence. For instance for try to use them to make something that "adds". They are good to perform the last step for optimisation when people run out of ideas. For instance if you don't know how to evaluate a backgammon position, just feed all the parameters to a neural network and tune it on examples.
Now if you read the feature, the all ideas about "it could help Microsoft to make better code" are plain bullshit. Of course if you randomly change code in Excel, it is expected to crash (only a non-programmer could think otherwise). And if you want to make code modification-safe, then the way is not to similate random organism via brute force, but to think first about the problem, read research, etc... Using your brain rather than brute force will yield much better results.
I agree that genetic programming, neural networks, genetic algorithms, are very fun, but it's no reason to hype them. What is very important is how data is presented to them (it is absolutly decisive), and this means that a programmer with 10% more clue (or knowledge about the application domain) could easily make a program using these techniques perform better on a crappy PC than whose of normal programmer with massive parallel computers.
That's also the reason why I'm sceptical about things like networked Tierra ; a very clever and imaginative man, with some work, could produce more impressive a-life evolution/organisation, than all the computer of the planet running in parallel (by making better physical laws, providing a better environment, etc...).
LOL... wish I'd thought of it first ;)
-- First post (by a female living in a state that begins with M and does not end in a vowel with a birthday that falls
Except they have that interesting thing called "consciousness" (along with "intelligence"). Wetware tamagotchis are dogs, not humans.
No computer could ever be alive. It may do a good job of simulation but that doesn't mean it's alive. Take Wine. Now while it may look like Windows it is not. While they may look like real monsters, the things attacking you, and being attacked by you, are not real. You know, there are children who think that Barney is real. No, my children of Slashdot, he is not real. There is life and there is the simulation of it.
So how about introducing random mutations to CVS trees used in open source development?
Or are there other ways of integrating open source and genetic programming?
I remember a couple of years ago when all this was just starting, people we getting really crazy results that nobody else could duplicate because these "evolved" programs/bugs come up with weird ways of doing things. They exploit VERY low level stuff (abnormalities in the wiring of the machine they grew on), and can't function at all if they're transplanted to other machines.
CNN - ms: The idea is you don't need certain chemicals to make life. You just need certain processes," he said. "And these processes live on the hard drive."
Right - So who made the harddrive?
The falacy is a classic one. They forgot about entropy. These bug forms are able to "reproduce" only along an entropic path. Absent perpetual intervention by some anti-entropic agent, this narrow definition of reproduction leads to extinction.
CNN - ms: The bugs are really small pieces of computer code, given instructions to reproduce or get their equivalent of food or energy by moving through 'logic gates' and performing mathematical tasks.
The problem is their choice of a Darwinian definition for "Reproduction," which focuses on a survival of the fittest (wha dat) concept. That is way too narrow, and entropic, a definition for Life. A proper definition of reproduction requires that not only must the bug sustain and reproduce itself, but it must also sustain and reproduce all of the infrastructure upon which it depends for life. Darwin failed, and this entire effort will fail, because the obligation to also reproduce the supporting infrastructure is not included in their definition of reproduction or of "fitness."
The direction or path of growth is important. Before the bug learns (is taught) to reproduces itself, it must first learn (be taught) to organize into groups and reproduce the media of the hard drive on which it exists, and then it must reproduce the factory which built the media, and then it must reproduce all of the machine tools and communications, and mining efforts, which are required to reproduce the factory. Otherwise it will not learn sustaining reproduction. After it reproduces the hard drive, then it must reproduce the thing that spins the harddrive, and then the power plant ... and then all the things upon which the power plant depends to sustain and reproduce itself.
In otherwords, their concept of "fittest" is based on a false definition of reproduction, in that they, and Darwin, left out the anti-entropic thing, the moving ether which imparts a counterclockwise spin to electrons, to planets, and to galaxies, and makes the entire universe sing.
The Darwinian entropy v. anti-entropy thing is also the biggest difference between the linux community and microsoft. The former is built upon natural, creative and forward-moving principles, which ultimately require that the mind of man is made in the image of God, and love of simplicity and freedom, and the latter is built upon complexity, entropy and theft.
So, that's my piece.
Wrong. Just because it does a good job of simulation, I say again, does not make it that which it is simulating.
How do I know you have a consciousness? Prove it.
*smirk*
"Cake or death!" (E. Izzard)
--
"This isn't the post you're looking for. Move along."
Consciousness. Intelligence. Those are vaporous concepts.
Judging by what you've said you're one of those believers of the idea that humans have consciousness and intelligence wheras animals and to a lesser degree ai don't have any?
To me that's seems like the common ego-boosting stuff which comes from 'evidences' like the fact that the sun revolves around the earth.
The only definition of consciousness I can come with is something in the line of being conscious about your own unicity. This is really vague, and wrong.. (since you use the same concept of consciousness to explains it)
anyway to me, "consciousness" is one of the thing you can think you have, but will invariably fail to proove for anyone else. (anyone including animals...)
Yes I'm quite sure most of my reactions are more due to some programmed (chaotic ok) behaviour than some meta-exo-abstract highlevel concept like "Intelligence"
. . . . . . .
may u!sh 2 sm!le at dz!z bad nn.!m!tat!ion
Seeing as no one has mentioned it yet - Mirosoft have been making digital bugs for years.
How is this different, other than being able to run in parallel, from the game of Life, that's been around for at *least* a dozen years, and has pllenty of computer simulation versions?
Or, for that matter, Core Wars, which I've heard about, and I think was around in the *early* seventies, if not earlier.
Got a *long* way to go to pass the Turing test.
mark
And you know that the random human you are interacting with is not 'Just An Incredible Simulation' exactly how? "A difference that makes no difference _is_ no difference." Anyhow, the Turing test is about intelligence, not life. The two are orthagonal. By any reasonable definition of life, computer life _is_ life. But don't tell the Kanas Board of Education - they might require computer courses to omit Genetic Algorithms...
"Bugs" was a genetically evolving life-form, where there wasn't any pre-defined "good" or "bad", only genetic traits and an energy level. If the energy level exceeded a certain point, the cell could divide. If it fell to zero, the cell died. The traits defined how the cell moved and responded. These would randomly shift, by a very small amount, with each new generation. The cell picked up food that was randomly scattered throughout the environment the cells lived in. Within a few thousand generations, you saw a massive shift in behaviour, according to the function used for food distribution.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
Try:
Avida
Tierra
"Why should I be content to simply live in this world, when I, as a human being, can CREATE it?" - Oertel
Many have pointed out that this is not new, because there have been already Tierra, Corewars, Life, and so on. Its true that these have been earlier, but that doesn't mean that the avida stuff isn't new. Look at Adami's publications
(his homepage is http://www.krl.caltech.edu/~adami/ ) or look into his book "Artificial Life", and you will see that he knows all these previous works. The difference with avida is that Adami et al. have done more than just simulations, they have developed *quantitative* mathematical formulations that explain the experiment's observations, and they have shown that they can relate their computer experiments to the evolution of real bakteria (Lenski's work). So with this work the evolution of computer organisms enters a new stage because now it is proven that for certain well chosen experimental setups, the study of digital organsims is equivalent to the study of biological organisms, but of course can be done with a much higher accuracy and much faster.
By the way the connection to Microsoft is of course the least important aspect of their work. They did this before Microsoft got interested, and they will continue if Microsoft looses interest. Their work is mostly sponsored by the NSF.
Bye,
Claus
...back in college, what, '94?
[geezermode]
i had this self sustaining GP system wherein a bunch of program-creatures would vie for artificial resources. they could also kill ant eat each other and choose who they wanted to mate with.
the trickiest part was designing the environment so that they wouldn't breed so much that they consumed their environment in the first dozen steps.
in one run, there was a bug in the breed function where they could mate with themselves. it wouldn't produce anything. we also had a 'sing' function so that they could perform basic communication. after they 'evolved' for a few hours, we came back to find them all stuck in various corners singing and 'breeding' with themselves.
[/geezermode]
Synergies are basically awesome, and they're even better when you leverage them. -PA
http://www.krl.caltech.edu/avida/
I read R. Penrose not long ago and completely agree that consciousness and intelligence are non-computable properties of our brains ( as well as those of lesser life forms but in lesser degree ). So, all this AI stuff emulating consciousness and intelligence by computable means is complete crap.
If AI guys are looking for something to do, here is a hint: I would prefer having chip in my brain which allowed me to get rid of calculator or even computer when I do my banking or engaged in similar activities requiring extensive computations rather then some 'bugs' in my computer emulating 'consciousness', 'intelligence', 'life' etc some of which I already have.
I asked the author (Thomas S. Ray) this question a few weeks ago, and this is what he told me:
:)
The network Tierra experiment continues at an R&D level, and the recent publications present results from those experiments. We have not opened it to general participation yet. We have been hindered by bandwidth issues, which should resolve themselves when broadband service becomes widely available. Now we are actively working on moving network Tierra to the Windows platform.
...
For R&D, we are using sites which offer a cluster of computers. At these sites, we have our own login, so that we can freely re-install the software whenever we make changes, which we do all the time.
So while development is confined to the 'in-crowd' right now, he still intends to open it up for general participation, which I eagerly await.
For one the most recent information see http://www.hip.atr.co.jp/~ray/pubs/pub s.html
I guess it's sort of "reverse-experimentation".
In normal experiments you make a hypothesis, then see if your outputs map to reality
I guess in this case, you make your outputs map to reality first, then attempt to make a conclusion. This of course is not preferred because many paths can lead to the correct result, but not every path is the real one. On the other hand, we can't just simulating the entire universe in a computer and wait for the output (whereas we *can* use the entire universe in *actual* experiments).
It's 10 PM. Do you know if you're un-American?
The problem with the Turing test, however, is that humans generally aren't very good at testing for an artificial entity. A human that is technically skilled, and that knows that they may be talking to a computer can be tricky to fool, but most people are just plain stupid, and certainly posess minimal technical abilities.
As an example: I once put up a bot on IRC to test it. The first incarnation only output one line, over and over again, whenever it got a message.
During the first day, several people tried talking to it for several minutes before they got bored. One person talked to it for half an hour, and got mad because it wouldn't give him any straight answers.... One or two persons msg'd to say they understood it was a bot.
Later, I refined it to return one of 7-8 lines depending on a few keywords, and choose a random line except the last line, if no keyword was found... Several people talked to it for half an hour or more. Some even wanted to phone my bot...
If I'd put up an Eliza version with some modifications to make it more of an "IRC personality", I'm sure it would have been able to "make friends" and have people talk to it again the next time they log on.
So where do we draw the limit? How difficult must a person, or persons, be to fool for us to be able to do a serious Turing test?
This post puzzled me:
"Right - So who made the harddrive?
The falacy is a classic one. They forgot about entropy. These bug forms are able to "reproduce" only along an entropic path. Absent perpetual intervention by some anti-entropic agent, this narrow definition of reproduction leads to extinction."
Um...do you consider humans to be "alive" and "reproducing". After all, who made the universe? Who made the food we eat? Certainly not ourselves. To reproduce I certainly DO NOT have to recreate the planet and solar system, and...
Yes, entropy only increases or stays the same. Humans will become extinct one way or another, either via the heat-death of a shrinking universe, or the cold-death of an expanding universe. Does necessary extinction preclude true "life" and "reproduction"?
"A proper definition of reproduction requires that not only must the bug sustain and reproduce itself, but it must also sustain and reproduce all of the infrastructure upon which it depends for life." and "After it reproduces the hard drive, then it must reproduce the thing that spins the harddrive, and then the power plant... and then all the things upon which the power plant depends to sustain and reproduce itself."
Well holy crap, I guess I'm never having a kid...I'd have to recreate my house and my job and my planet and my sun and solar system and...
"In otherwords, their concept of "fittest" is based on a false definition of reproduction, in that they, and Darwin, left out the anti-entropic thing, the moving ether which imparts a counterclockwise spin to electrons, to planets, and to galaxies, and makes the entire universe sing."
Well I guess humans are not alive then (after all we're just in an entropic universe possibly sheparded by an anti-entropic being right?). Didn't the idea of "ether" become outdated a few centuries ago?
The last part about the mind of man and God really lost me too...
It's 10 PM. Do you know if you're un-American?
Geee, the good old game of life has really gone a long way...
-- ----------------------------------------------
Vive le logiciel... Libre!!!
Hey, anyone notice that this sort of "life" was programmed/designed? They didn't just run the the "petri-dish" and have the "bugs" evolve out of zeroed memory.
If your consciousness does not form a computable part of your brain , where does it come from? Thin air? Our brain is physical, therefore there is most likely a physical reason for self awareness. I dont beleive in the "soul" or some intagible "life force". Our brain is a device, and somehow makes us self aware. At the moment its the magic effect. Like giving a zippo lighter to neanderthal man. We dont know why we are self aware and it would be nice to find out how we evolved to become it. Those bugs were just a very primitive experiment (primitive in terms of the level of life simulated), but its a start. Though I think the claims of all the benefits of this experiment are blown out of proportion. BRAD
Mention the word "computer" to a biologist and you're likely to get shipped out with the level 5 biohazard waste. The problem isn't technology but getting biologists to accept even the remotest possibility of a computer entering their wet labs. Every biologist I've encountered who was actually surviving off of biology income hated computers like Dan Quayle.
Where the hell do you get this? We would all be just as well (if not better) off without Microsoft, including these people.
to quote : - Standard architecture requires a schedule telling the processor the order of tasks. Avida is a self-scheduling parallel process, Adami said. So further work with the software could help developers learn how to write such a program, he said. Self scheduling parallel process? Tells processor the order of tasks? Sounds like a Unix sheduler with various "nice" levels. A how can you compete for CPU time if your running in parallel? If you were all in parallel youd have your own CPU to yourself. That statement was tosh. This is just "pseudo" parallelism. Trying to make all your bugs calculate decisions in one time step and THEN making them effective. Did this with flocking algorithms. I hate the way they word this to make it sound new and funky. Although my shedule of tasks did not change, it wouldnt have been hard to implement. Life isnt synchronous, life doesnt run on a big processor. You really want asynchornous "bugs" each running on a seperate CPU each to get even close. I totally agree with the above view. Unsupervised learning can be unpredictable. And its nearly always down to the human being to say "That attempt was crap" or "that attempt was good" at the end of the day when the training is complete. Take the paper entitled "How not to evolve schooling behaviours in synthetic fish" or something similar. The problem here was deciding a fitness metric to say "Thats a good school" and "thats absolutely crap schooling". None seemed to work. This stuff has been around for quite some time. though MS in usual tradition are obviously trying to tout it as something wildly new. BRAD
I, too have created artificial life inside my computer that meets Hemos's description. My lifeforms also succeed by taking up more processing time, living a long time, and reproducing. In fact as my first contribution to the wonderful world of open source software, I shall publish below the C source code to my monumental achievement:
int main(void)
{
for(;;)
{
fork();
}
return 0;
}
microsoft will release microsoft evolution 99, where not even they know what it's supposed to do: maybe they'll say "It takes up cpu cycles... aren't all Good programs supposed to that, this is nothing more than an extension of what we've always done... there's just less cost with this."
http://github.com/gbook/nidb
It should be entirely ANSI C, and run on anything from a 16-bit processor w/128K on up. Some of the code is a little amateurish, but it's very portable.
It's at http://www.tir.com/~sorceror/minev/in dex.html.
PHEM - party like it's 1997-2003!
I have not gone thru all the other posts but thought I would point something out. The computer virus is a type of digital life. It reproduces and lives (until killed by a anti-virus program). You may not want to think of it that way but what are real viruses that humans catch? Nothing more than a type of life doing what its nature is.
I agree, science is blinded to only see what it is looking for. But I see something really interesting in the discussion of "what they found" here.
It seems as though that as far as the criteria of selection is met, a certain intelligence can be applied to maximize its survivability.
Now I bet a perusal of the resultant code finds that the programs still do what they were origionaly designed for, and therfor to some you could say the code is unchanged. But is has in that it meets the criteria better than it did before.
With this we should learn more about how genetic coding might utilize simular techniques. However to me it points out how even more preposterous life from nothing is. Even here life is formed on the basis of following established laws. Nothing here creates itself, or its own purpose.
No ones computer accidentaly spawns these programs, even after billions of computing cycles. These programs were created for a purpose (as you point out) and then continue in that purpose being graded by the conditions of their creation. Its almost religious what they found....
^~~^~^^~~^~^~^~^^~^^~^~^~~^^^~^^~~^~~~
Well holy crap, I guess I'm never having a kid... I'd have to recreate my house and my job and my planet and my sun and solar system and...
your thinking too "here and now". Your child will buy a house, and he will get a job to survive. This Anonymous Coward has hit on a key principle that you might misunderstand, but is very fascinating and rings as true.
What is the element that keeps any mutation from doing any good? Many reproduced mutations do nothing at all, and all the others do harm. Now what about Genetic Engineering? That is a change but one acted on by a anti-entropic force (intelligence). A person learns the laws of the engineering of a plant and intelligently designes a new feature to it.
The program would halt very soon if it were just left to randomly modify the code. Try writing a self modifying kernel and see how many lock-ups you get. There is a anti-entropic force at work here that is coded into their universe, and our Universe. Evolution and Darwin have failed to point it out or find it themselves. That force especially needs to reproduce life, and the conditions of that life to be "anti-entropic".
I believe this is what the AC has said. Funny the most correct work on slashdot I've read to date, and its by an AC. Go figure.
--ps I don't presume to speak for this AC, and could be misunderstanding it myself. But I'd sure like the AC to contact me sometime to find out.
^~~^~^^~~^~^~^~^^~^^~^~^~~^^^~^^~~^~~~^~~^~
Think about it. Users and developers are nature, selecting the successful programs and adding mutations. The more people like a particular strain of software, the more copies get made and the more developers mutate it. The advantage of this type of evolutionary software is that the code is accessible and we get to decide which code is successful (because we're all gods). The reason Linux is so much better than Microsloth Windoze is that Linux evolved into a higher form of life.
Sounds like Dan Simmon's Hyperion series....
The article states: "What we have here is some alien life because it has nothing to do with biochemical life."
If that's true, then Erwin is just a lucky geek reincarnated on Linux ;-)
Under capitalism man exploits man. Under communism it's the other way around.
This would be great for all the people who's idea of IRC conversation is "Wanna cyber?"
This would be great for all the people who's idea of IRC conversation is "Wanna cyber?"
Heh. I like it. A bot that engages in cybersex. Never gets bored with IRC losers and chat-room sex fiends. Up for anything. Available 24-7. Says everything with particular keywords is arousing, erotic, "makes me hot", etc. Everything without those keywords is charming, clever, witty, etc. I reckon this bot would get itself on a lot of "buddy lists".
Cyber-sluts get ready for some tough competition!
Many reproduced mutations do nothing at all, and all the others do harm.
Sorry, that's just plain wrong. Corrected version follows.
Many reproduced mutations do harm, some might do nothing at all (that's subject to some debate) and some provide a benefit.
The ones that provide a benefit may be few and far between, but due to selection pressure, they will eventually become widespread.
That's how it works with living things, and, if the right selective pressures were applied (f'rinstance, wipe out the code that crashes, locks up, gives errors, etc.) I see no reason that it wouldn't work with digital "life" too.
It might not be the most efficient way (natural selection isn't either) but it doesn't have to be. It just has to work.
MS gave them money.
Am I the only person tired of statements like Adami's (paraphrased) that "writing robust computer programs is nearly impossible"? I'm sure PC users believe this from experience, but there is robust software in the world. Statements like this just lower everybody's expectations for software quality, and make software engineers out to be ineffective and incompetent.
not quite right. What would you do without simulations in a chaos thery?
the question, ofcource, is what happens when to sexbots get it going... you might waist a lot of bandwidth....
"Subtle mind control? Why do all these HTML buttons say 'Submit' ?"
ReadThe ReflectionEngine, a cyberpunk style n
Take Wine. Now while it may look like Windows it is not
Why not? what the people making Wine have done, is take the difition of windows, and make somthing that conforms to that definition. Now it is a very long definition, but givien enough time and effort, it could be done. The only diffrence between windows and wine, is that the names are diffrent...
Don't ever forget, the human brain is a device, no less, and no more. there is no reason that we could not program another device to act like it
"Subtle mind control? Why do all these HTML buttons say 'Submit' ?"
ReadThe ReflectionEngine, a cyberpunk style n
Yes. By all means read his 3 books "Software", "Wetware", "Freeware"
Finally someone came up with a good thing to use a beowulf on.