What is a GA?
Discuss.
704 thoughts on “Holding tank for general chatter about GAs”
Leave a Reply
You must be logged in to post a comment.
The Skeptical Zone
"I beseech you, in the bowels of Christ, think it possible that you may be mistaken."
What is a GA?
Discuss.
You must be logged in to post a comment.
Why hide that brilliance? Put it out for all of us to see!
Yup I just didn’t edit YOUR shit out- are you proud of that too?
And I said the mutations are random and the selection process directed them towards the goal which it does.
Better
Oh, you’re just too much the tough guy for me. I’m afeerd you’d hurt me. Just post your code.
This is a keeper too:
Wrong again, as usual- your ignorance is not a refutation oleg.
Mutations are accumulated by selection- THAT is my claim- they accumulate towards the fgoal of the GA- and that means the GA directs the mutations towards the goal.
It doesn’t direct which mutation to occur- which is what your little mind thinks i am saying.
What? I have nothing against you ben.
I was just checking to see if you were really interested and you ain’t
Selection does not care one whit for mutations. It knows nothing about them and it does not direct them. It selects fitter organisms. Mutations remain undirected throughout the entire duration of a GA. No guidance is involved.
Umm dumbass- the mutations would be what is being selected, so the selection process is all about selecting the mutations that best fit the goal.
With a GA all is directed towards the goal
Joe, I am afraid you are out of luck. No one is going to pay you.
Oh, I am very interested in seeing your GA code. Just not in person, cuz I am scared of you.
It doesn’t direct which mutation to occur- which is what your little mind thinks I am saying. 🙂
Then you have to pay me
Nice:
Wrong again, as usual- your ignorance is not a refutation oleg.
Mutations are accumulated by selection- THAT is my claim- they accumulate towards the goal of the GA- and that means the GA directs the mutations towards the goal.
It doesn’t direct which mutation to occur- which is what your little mind thinks i am saying.
But, I am poor and only have enough money to put food on my family. Are you saying you aren’t going to share your vast reservior of knowledge with us proles?
If you want free stuff you have the intertubes- all you have to do is search for what you want to know and then read.
He doesn’t have anything to show. All hat, no cattle.
Okedokee. Although could you answer one question for me? Since you know more biology than biologists, more physics than a physicist, and more about GA than people who actually use them in their work, why exactly are you wasting your time arguing on the internet with us retards? Why aren’t you out there in the real world doing ground breaking science?
This is as close to a concession as we’ll see from Joe. Good. This is in line with the standard definition of a GA.
Cambridgeshire, UK, Joe. Just tell me when you land in the UK, and I’ll come pick you up.
Joe G,
I agree with you 100% here.
This now only leaves the problem of defining “selection”.
Since you are an IDist, I believe you think that selection is focused by some sort of conscious entity, but I think it could be much simpler and still work.
Water does not pool in the lowest point due to any conscious direction but the pooling is instead simply a function of gravity.
I believe evolution works in the same way.
So … we have a process of mutation that generates novel variants, and a process of selection that causes those variants to survive more frequently and/or produce more offspring than the current variant, leading to increase in the population. We iterate this process. This is (somehow) different from ‘conventional’ evolution, even though it does the same thing.
Joe calls this a ‘GA’ – genetic algorithm – because that same technique is used in search methods in computers to find solutions to various problems that happen to be tractable by this technique – where there is a large search space and a ‘landscape’ of fitness peaks and troughs. (I wonder where they got the idea from …?)
In the latter case, fitness is determined by approach to a particular solution being sought. In the former, it is determined by survival in the actual environment in which the organisms must exist. So the ONLY thing that Joe has added is teleology – the ‘natural’ GA is ‘goal-oriented’ in a sense that lies beyond enhanced survival in the current environment. 700+ posts for that!
Q.E.D.
All the threads you show up on are far longer and fact-free than all other threads; and most of this is your insults, taunts and swearing name-calling. You have more of your crap sent to Guano.
This is not an idol “accusation,” it is pure, objectively observable fact. Count the comments.
You simply cannot prove that wrong.
Tell us Joe, did the GA in that case reside inside of the antenna?
Give us one real world example, any example at all, of a GA being totally contained and working inside of the object being evolved.
Put up or shut up time for you Joe.
And that should have been not an idle accusation. My cat seems to think so as well and keeps blocking my view of the screen.
Hi Joe,
Long time lurker, first time poster. I know you’ve got lots of posts to reply to, but I just wanted to check that I’ve got the right idea about your proposal. This echoes Alan’s post upthread a bit too, I guess.
First off, here is my understanding of how a GA operates, using the antenna GA example:
We start with a population of circuits. The individual circuits in this population reproduce, and their offspring are subjected to a limited number of random mutations. Each individual offspring circuit is then tested to see how well it performs as an antenna. The ones that make the best antennae are permitted to reproduce, their offspring receive new random mutations, and the cycle continues until hopefully we get an optimal antenna circuit.
Does this match up with your understanding of how the process works?
To me, this is pretty much how I understand biological evolution to work too. The main difference is that instead of being tested against how well they perform as antennae, the only test that individuals undergo is to see how often they can produce viable offspring.
I know you don’t accept ‘Blind Watchmaker’ evolution, but is this close to your idea of how it is proposed to work too, or do we have different ideas?
Finally, I want to make sure I understand your GA proposal (which I’ve just realised you could call a ‘Joenetic Algorithm’, seeing as you get to name it!).
Your population consists of individual strands of DNA inside a biological organism. These strands reproduce (inside the organism? Or between generations of organisms?) and their offspring are subjected to random mutations. The strands of DNA are evaluated for fitness (by the organism? By the environment?). I’m not sure what fitness means here though – is it how much it benefits the organism?
Sorry for asking so many questions – just want to make sure I understand your idea.
Thanks!
BINGO!
I only needed ‘threaten with a meet up” to complete a line in “Joe G Dipshit Bingo”!
But you give so much of your time for free to stupidity…
Fudge! I was only a “nature operating freely” away.
I’ll try to summarise what I’ve got from Joe’s ramblings so far, as well as my own thoughts and (admittedly limited) understanding of GAs.
1. Joe uses his own private definition of “Genetic Algorithm”: “a search heuristic designed to solve a problem”, I shall call these “Joe algorithms” (JAs) to avoid confusion. My own understanding is that GAs are simply in silico applications of models of population genetics used to explore fitness landscapes.
2. According to Joe GAs need not work upon populations, and the term “population” is imprecise. My understanding is that all GAs work on populations of strings (“genomes”) that are evaluated by a fitness functions. Each string can be mapped to a particular configuration of the “organisms” (“phenotypes”), the fitness function may evaluate the strings directly or indirectly via their phenotypes. Thus, there wouldn’t be “imprecise” elements such as “populations of circuits”, but rather populations of sets of circuit parameters specified by the strings.
3. JAs have specific goals, determined by functional specifications rather than specific “solutions”. The goal of one of the GAs from the SciAm article is determined what an antenna “has to do”. When asked what would biological organisms “have to do” under that analogy, he replied “they have to survive and reproduce”, which is rather quaint as the best known way to do that is to let the environment set the fitness function, emulating natural selection.
4. First, JAs work by directing mutations in response to environmental pressures, but later, we’re told that mutations are random then directed by cumulative selection by the GA (within a single cell, within a single generation? where’s the population of mutations to select?). The dynamics of this is unknown, but one should think they are different from NS, as Joe says that NS can’t do anything.
5. Contradicting point 3, Joe tells Thorton JAs have built-in “targets”, perhaps even linage-specific targets. The interplay between the deployment of the built-in targets and the adaptive response mechanisms remains unknown.
6. The JAs are supposedly located in and transmitted by the cytoplasm and/or cell membrane (Pivar’s paper is presented as supporting this, but besides being probably the worst developmental biology paper ever published, it actually says nothing about the transmission of anything by the cell membrane), and they have no material basis. They are just “information”. Real GAs wouldn’t work within cells because the population size is always equal to one.
7. Joe has no model for a JA, and can’t even provide a dummy example of what a JA could be like. In the lack of a model, he can’t claim to have evidence that fits his model.
8. Joe does not seem to posit direct input from the designer during the execution of the JAs.
9. JAs are goal-oriented (points 3 and 5), and that’s an indication of ID by definition. However we’re not provided a single concrete example of such goals. It is not clear either how can Joe find out the goals of JAs. To claim that JAs have goals by definition and that’s all we need to know is fallacious, resembling the old “every creation needs to have a Creator”. Joe has not discovered how JAs work from any evidence, he just made them up.
10. Another reason why JAs imply ID is because they are linked to the designer in their origin. Joe however doesn’t know any specific JA and lacks any general model for them (not to mention he can’t even support their existence), so it is unclear how could he know anything about their origins. If JAs do not require the input of a designer during execution and cannot be shown to require a designer in their origins, the designer element is completely superfluous and unsubstantiated.
I’m sure I’m missing many things. Especially since I started writing this in the morning and haven’t refreshed the thread yet (the last post I read was OM’s on March 3, 2012 at 5:01 pm).
Not from Joe G. One never misses much.
This again comes back to the Fundamental Misconception of all ID/creationists; namely that it’s all “spontaneous molecular chaos” down there, everything in the universe falls all apart and therefore there must be “information” pushing atoms and molecules around.
“Information” pushes nothing around. ID/creationists have not and can not elucidate any mechanism that does that. Joe G’s use of GA appears to be another version of “information” pushing atoms and molecules around. Simply asserting it does not prove it; it simply reaffirms that Fundamental Misconception.
The laws of chemistry and physics regarding the condensation of matter come from actually taking matter apart and studying the phenomena and the rules. These are the rules that apply; not “information.”
Genetic Algorithms are written to mimic nature. Insofar as our understanding of nature is correct, the algorithms pop out the kinds of results we see actually occurring in nature.
There is also an objective measure of Joe G’s effectiveness at articulating concepts; and that is contained in the numbers of comments he puts up in every discussion he pushes himself into.
Take the number of his comments that actually articulate the correct meaning of any concept; in other words, take the number of his comments that actually perform the function of articulating a concept. Then divide that by the total number of comments he makes.
Now take the negative of the logarithm to base 2 and we get the number of bits of “Functional Misinformation” from him.
Try it; it comes out very high.
I asked: “what IS the meaning of a mathematical expression or procedure concerning a group of organisms being located inside or outside these organisms?”
JoeG answered: “How would you locate a GA outside an organism that has to control the inside of an organism?”
You seem to think that this is somehow a follow-up question to my question. It clearly is not. Please read the question again, Joe. It is: what does it mean that a mathematical expression or procedure concerning a group of organisms is located inside or outside these organisms – i.e. *located* anywhere in relation to these organisms? And then please try to answer it.
Guano Alley, judging by the steaming pile of Joe Guano that our blogmistress will have to shovel out when she gets back.
LOL, that would be an interesting exercise, but I’m afraid I don’t have the stomach to read again all this dreadful thread (and that would still be missing all the stuff that was moved to Guano!).
I think the concept of information, –plain information–, is interesting by itself and would be worthy of an own thread. I know of an ecologist who describes ecology as the study of the flows of energy, matter, and information in the biosphere. In my not so well worked out views, information would be merely an abstract representation of a pattern. When we observe the entities or relationships between entities are involved in a regular process, as we observe nucleotide sequences are involved in transcription, we construct a system of representation out of the relationship between the observed pattern and the output of the regular process. Thus, information, as well as functions, would belong to the realm of descriptions rather than nature itself.
I suspect this is all part of a ploy from Elizabeth to get tonnes of free fertilizer.
Wow- you are wrong, right off the bat- I am using the standard definition of GA as out forth in wikipedia, and every pkace else- as far as I can tell.
Dawkins’ “weasel” and “Evolving Inventions” are exactly what I am talking about- finding definite solutions- see the first “termination” in the wikipedia article.
2- I explained the population confusion
3- The point is to find a solution to get to the goal- that is pretty well accepted in real-world GA land. Again see wikipedia
4- I have ALWAYS said the mutations are random but are cumulatively selected towards a goal- again see “weasel” and “Evolving Inventions”- and again if you could step up and demonstrate NS is up to the task I wouldn’t be having this discussion
5- What?
6- You are wrong as the population in the cell consists of DNA, RNAs (many different types) proteins and amino acids- at the least
The cell would be the computer
7- Provided- also just look at the way the internet works- your email doesn’t always travel the same exact path when you use the internet to send it- the signal doesn’t care what path it takes as long as it gets to the proper destination- it is an ever changing environment with the non-random selection making sure everything gets to the right place.
8- Set it and forget it
9- Exactly and the goals would be 1) just to pump out proteins in the first place and 2) to get new molecules and configurations when required
And what is the option- just get the right chemicals together and badda-bing, badda-boom, ribosomes appear and just start chugging out proteins?
Again any evidence in support of your position and my GA is not necessary- so step up at any time.
But anyway-
What I said:
1- You need a target-> the goal-> what it is you are trying to achieve. No need to write an algorithm if there isn’t a problem to solve-
“I wrote an algorithm”
“What does it do?”
“It’s an algorithm, stupid.”
“How do you know when its done?”
As the algorithm chugs along it keeps checking for any match to those specs.
2- You need to figure out a valid starting point- those initial conditions- one way is to determine what it is minimal you can do, without any algorithm, to get as close to the target.
3- You need the proper resources that the algorithm can use to get from starting point to the target.
4- Then there is that algorithm or algorithms that, from the initial conditions and the provided resources, some of which can be by-products of the algorthim(s), can produce the desired solution.
That should be it- once you do all of that and hit “go” it is hand’s off for the designer(s).
And a look at wikipedia says on genetic algorthms
Well if wikipedia is an example of the real world then so far so good, we are in agreement.
Allan- if you had any evidence that blind and undirected could do anything, we wouldn’t be having this discussion- ya see just producing more offspring doesn’t create new proteins nor new protein complexes…
This from the same guy who wrote
What’s the point of including amino acids on this list? They are fixed building blocks. Mutations don’t change amino acids themselves.
Joe G: 2- I explained the population confusion
You rather exposed your population confusion.
My point was in the last sentence, read that again
Then you explain yourself very poorly.
“What?” indeed.
GAs are generally made to sort entire genomes. In a diploid genome, you have only two alleles to choose from, does that mean that JAs produce homozygosity? How would that work in haploid genomes? You actually could select RNAs and then use reverse transcription, but in that case the variability would be due transcription errors rather than mutation. For proteins, variability would come from transcription, translation, and post-translational errors (again, not mutations), but, more importantly, you’d need reverse translation, which is not known to exist.
I can’t get anything from that. You need to make explicit your analogy to biological systems in order to have something of a testable model.
The way you’re using the word “goal” there is not analogous to the goals of the antenna algorithm. There you had a specific set of functional specifications. If your goal is actually movable due the fluctuating conditions of the environment, what you need to do is to let the environment set the fitness function, like NS.
There isn’t a problem to solve. Nature doesn’t try to solve problems, it just lets things happen the ways they can happen. Then we find some parts of nature interesting, and anthropomorphise the situation thinking about them as “goals” and wonder about their origins as “problems”. There is a difference between creating an algorithm for your own goals and using algorithms to model natural regularities.
I notice you didn’t answer point 10.
Hey joe, book a flight to Portland, Oregon and post your arrival day and time and flight number on your blog. I’ll pick you up at the airport. You won’t chicken out, will you?
Quotations from Joe.
2- I explained the population confusion
You rather exposed your confusion.
3- The point is to find a solution to get to the goal- that is pretty well accepted in real-world GA land. Again see wikipedia
The point was in the last sentence. Read that again.
4- I have ALWAYS said the mutations are random but are cumulatively selected towards a goal- again see “weasel” and “Evolving Inventions”- and again if you could step up and demonstrate NS is up to the task I wouldn’t be having this discussion
Then you explain yourself very poorly.
5- What?
“What?” indeed.
6- You are wrong as the population in the cell consists of DNA, RNAs (many different types) proteins and amino acids- at the least
GAs sort entire genomes. If JAs can use a single genome as a population, they’d have to choose between x alleles, where x is the number of homologous chromosomes, and then use some mechanism to produce homozygosity with the right allele (within a single cell cycle? we’re still not clear about that). You could select RNAs and then use reverse transcription, but all this time you’ve been talking about mutations and RNA variability is produced by transcriptional errors and post-transcriptional modifications, not mutations. The variability in proteins comes from transcriptional and translational errors, and post-translational modifications (again, not mutations), but more importantly, as Oleg rightly said, you’d need reverse translation, which has not been shown to exist, in order to make that work.
7- Provided- also just look at the way the internet works- your email doesn’t always travel the same exact path when you use the internet to send it- the signal doesn’t care what path it takes as long as it gets to the proper destination- it is an ever changing environment with the non-random selection making sure everything gets to the right place.
I can’t get anything from that. Where can we take a look at the genetic algorithm of email? In order to have something resembling a testable model, you need to make explicit your biological analogy.
Continuation of my last post.
8- Set it and forget it
Great!
9- Exactly and the goals would be 1) just to pump out proteins in the first place and 2) to get new molecules and configurations when required
So the fitness function is effectively set by the environment or what?
Again any evidence in support of your position and my GA is not necessary- so step up at any time.
Your position wins by default, how very nice for you!
1- You need a target-> the goal-> what it is you are trying to achieve. No need to write an algorithm if there isn’t a problem to solve-
There isn’t a problem to solve. Nature doesn’t solve problems. It lets things happen the ways they can happen. Then, we find some bits of nature very interesting and anthropomorphise the situation by picturing those bits as “goals”, and we wonder about their origins imaging them as “problems”. You are confused because you think one “discovers” algorithms in nature, but we rather use algorithms to model natural regularities.
Joe G,
I note there was no reply to point 10, which was the critical part of your design argument.
Hey joe, you said that your “industry” uses genetic algorithms. What “industry” are you currently working in? What genetic algorithms does your “industry” use?
When it comes to burden of proof you’ve said, “The burden is on anyone making a claim.” Well, joe, you claim that your “industry” uses genetic algorithms, and that you have used and written genetic algorithms, and that GAs are inside organisms, and that “CSI” can be measured, and that an intelligent designer designed and created (“origins”) all living things, and a LOT of other claims. Let’s see your proof, joe.
Oh, and you flaked out on answering this question:
Why do volcanoes, rocks, rain drops, snowflakes, human toe hair, fossils, ear wax, tooth decay, asteroid impacts, super novae, birth defects, insanity, death, albinism, floods, tornadoes, balding, pimples, and freckles occur, and do they have a goal and are they directed toward it?
By the way, joe, do ghosts have an internal (or external) GA that directs them toward the goal of being a ghost? After all, you did say:
“As I said the evidence for ghosts means there is evidence for non-human agency.”
Here:
And does allah have an internal (or external) GA?
But just two days ago you claimed
and yesterday you said
So you ALWAYS said mutations are random, except when you said the designer was controlling and directing them.
Go ahead Joe, stick that other foot in your mouth too.
I’m sure I’ve heard that somewhere before …
There is a great deal of evidence, but unfortunately any time anyone attempts to discuss it with you, you wriggle and writhe like a baby attempting to avoid a spoonful of pureed peas. Won’t! Shan’t!
Aha! Nice strawman! (do you like the phrase? I picked it up off an internet troll). Mutation makes changes, not selection. Having more offspring merely promotes a change that has already occurred (initially in one organism) against the current variant. Of course, whether or not mutation can produce new proteins or protein complexes is a valid question. And the answer is: yes, it can. Now, I confidently predict that you will find something wrong with this. In fact, noted biochemist Casey Luskin may already have made the appropriate soothing noises – nothing to see here, people. But as far as I can tell, the mutations made here were random. They were, admittedly, done in a lab by people with brains, and the proteins had no biological context. But they had no way of knowing how to achieve their goal (functional protein), so they just let random variation create the input, and a non-predictive selection process pick the survivors.
Whether or not one functional protein can convert into another (after duplication) depends on the amount of function in sequence space, which this lab, using similar techniques, has found to be surprisingly high.
And this (from the actual paper) is rather cool in the present context –
Design – even using a GA based on Darwinian principles – is less powerful than the Darwinian process itself.
Hey joe, do you remember saying this on UD?
“Also archaeologists are saying things about ancient astronauts- and for good reason.
People and scientists are also investigating ghosts and the paranormal.”
Do you think that ancient astronauts or ghosts or some other paranormal thing of some sort is responsible for designing and installing GAs in humans and all the other living things on Earth?
Do you think that humans and ancient astronauts mated and had viable offspring, and do you think that modern humans have some traits, features, genes, DNA, or anything else from those aliens?
Do you think that humans and ghosts can mate and have viable offspring? If so, will you please describe what their offspring would look like and be like?
And one more question for now: Can ghosts hurt people or animals?
Joe’s usual riposte to this sort of thing is words to the effect of “What about the ribosomes! Ya can’t have proteins without ribosomes! ID!”
But when yo try to engage in a little discussion o nthe evolution of ribosomes, he slinks away with a curse or two and a couple of “obviously’s” and “your position has nothing”‘s
Allan – nice bluff- if your position had any evidence you would present it- and what GA is based on Darwininian principles? Darwinian pronciples do not include a goal.
BTW Allan – variation, ie mutation is part of natural selection.
What you can’t do is demonstrate that blind and undirected processes were responsible for any new proteins- that is the problem- all you gave is question-begging scenarios.
Hey dumbass- the GA does the controlling/ directeing FOR the designer.
oleg- I explained that to you- do you really think your retarded problems reflect upon me?