I see forms as another mode in the “Thought” or “Information” line and our subjective experiences as just one finite mode in that same line.
Yeah, I can see that. I’ve read Seth Lloyd’s Programming the Universe and Gleick’s book The Information. But there’s still value in distinguishing between the two natures/ applications of information.
monstro, yeah I totally get your last post. And while I like words (see username) and respect definitions, the ability to frame Big Ideas in simple statements is something I seek out. When it’s all said and done, do I come out on the other side of this rigorous discussion with crisper approaches for how to face the day, you know?
Thankfully, you have already got your answer from the GQ thread you started:
And that is, of course, exactly what I said in this thread.*
Moreover, and very important:
You should not be so terribly eager to assume that the physicists who said they favored the MWI, and also said that randomness was irreducible, do not believe that the MWI is deterministic exactly as defined. Determinism is a very easily defined word, but the same ease applies neither to “random” nor “irreducible”. I’m not exactly sure what they mean when they talk about irreducible randomness, and in my opinion, you should not be so (seemingly?) sure of what they mean either. Depending on their definitions, they could very easily acknowledge causal determinism while also approaching randomness in personal, idiosyncratic ways.
*(I want to commend you, by the way, for opening that thread. I’ve had conversations in the past which simplified to a basic GQ style of question, and the other poster, rather than taking the effort to actually gain more information about the simple matter of fact they were literally just making assertions about, instead said something along the lines of “Well, I’m not going trust on economist on that, derpy derpy derp.” Curiosity is an important trait, and it’s always admirable to make oneself more knowledgeable about a topic rather than summarily dismissing something because it comes from a strange source. I admire that.)
I’ll return to defining “decision” either tomorrow or the day after.
I didn’t say it was being the entity.
I said it was being the entity’s mind.
Even in parens, that is not a word to be ignored.
Unfortunately, for now I’m going to have to leave this particular topic of conversation where it lies.
I actually haven’t been trying to convince anyone that it is.
Rather, I have been trying to explain how a causally deterministic universe is potentially consistent with what we perceive. Not that determinism must be the case, but that it legitimately could be the case. Many of these threads – maybe most, and maybe even all of these threads – get caught on the notion in some people’s heads that “determinism” (however strangely they define that word) cannot possibly explain events as we see them unfold in front of us.
I didn’t enter this thread to argue for a deterministic worldview. I entered this thread to define determinism in a way that might be actually clear for people, so the worst of the objections to the idea could finally be dismissed.
And I see right now that my major mistake in this thread was trying to answer your questions about whether I believed this universe was actually deterministic. I mentioned time and again that I wasn’t certain of anything, but after that, I gave my own opinion and said sure, I think it is. I pointed out it would take a much longer post to make even a bad argument in favor of that.
I have not even once been trying to make this much longer argument. I’ve just been trying to answer honestly whatever questions about determinism you offered up. And now you write here you think the universe is probably not deterministic.
I think we all knew that already.
If I actually wanted to argue in favor of determinism, each of my posts would be about ten times longer than it currently is. That might sound like a joke, but it isn’t. It might actually be an underestimate. I was trying to spare the people here the tedium of experiencing that, by giving them the lesser tedium of answering your questions honestly in a (for me) brief space. But even that seems to have been too tedious for this particular context.
I’m not going to do that anymore here. I’m not going to answer questions about qualia, or simulating a mind vs being a mind, or anything along those lines. They’re a digression that seem to be taken for an argument.
From here on, I will answer questions about the definition and implications of causal determinism only. Most importantly for this thread, this includes the definition of “decision” within a deterministic system. But most of the other stuff that is bothering people, I’m going to stop with right now. So, in an attempt to return to defining determinism and exploring its implications…
I want to be very clear here.
The MWI does potentially make “predictions” that differ from the other interpretations… from the perspective of a very small subset of individuals. What it does not offer is predictions that can be printed up in a scientific journal. One such prediction can be seen in the idea of quantum immortality.
This idea is very frightening to some people, so anyone who is a bit squeamish might want to stop reading now.
If you wanted a very, very, very small subset of your selves in various worlds to know for pretty much certain whether they lived in an MWI universe (conditional on the very important possibility that they actually did live in such a universe), you could set up a gun pointed at your head, hooked up to a quantum event. Before you tested on yourself, you would want to make sure the equipment worked. Have the mechanism hooked up so that quantum event 1 fired the gun, and quantum event 0 did not fire the gun, and then run the event 1000 times, so that on average the gun fires 500 times (or whatever) over the course of the 1000 trials.
You could then aim the gun at your head and run it (potentially) 1000 more times.
If the first trial aimed not at your head looks something like 0111010010111010010, with the gun often firing, and…
if the second trial looks something like 0000000000000000000, so that you never seem to get shot in the head…
then that can inform the one survivor of the experience of the underlying reality. At the expense of countless other murders. This is obviously a highly unethical experiment. No one else in the scientific community could repeat the experiment in a way that you were likely to verify. But it does make an unambiguous prediction that can potentially be confirmed for a small subset of people. (It does not make a prediction that can be rejected for even a small subset of people, since things turn out very badly for every version of you, instead of merely 99.999% of the copies of you, if the idea is wrong.)
And what is even more disturbing for some people is (again, not for the squeamish) our very lives are very similar to this experiment. We walk around every day, and it’s quite possible that the gun sees a 1 for certain versions of ourselves. Every day. But that also means the gun sees a 0 for others, and that might continue for the survivors. Forever.
Now, physicists don’t tend to call things like that a “prediction” because they can’t publish a paper on it.
But there is nevertheless the distinct possibility that this is relevant to your future life experiences (for a very, very, very small subset of “you”). If you’ve ever 250 years old and thinking back on this conversation, I would strongly suggest that you change your mind on the deterministic nature of the universe.
And that brings me to “shut up and calculate”.
I do not believe that every single person who has ever spoken that phrase is narrow-minded. Absolutely not. (I have already tried to clarify this, yet you still decided to ask that sort of the IMHO question at the end of your GQ thread, so it seems prudent to clarify. Again.) But here is the point.
The world is what it is.
If it’s deterministic, then it is not remotely interested in your relative disbelief about that. The world is what it is. It follows whatever rules it follows. (Or maybe in your view, it follows whatever non-rules it follows.) However it works, I believe that intellectually curious people should have some interest in the question. We are, after all, talking about one of the fundamental questions of reality. How does this world actually work at a core level? Can we figure that out? And if we can, what answer should we accept?
There are some physicists that say “show up and calculate”. And if you ask (some of them, not all of them) why they do that, they say, “Well, I’m trying to do physics. Worrying about which interpretation is correct is philosophy, not physics.” And okay, I can roll with that. I don’t quite define those two words in that way, but okay. But then, more than once, the next sentence out of their mouths has been something like, “And I don’t care about philosophy.” And notice right there that they’re saying that the question of how the physics of our world actually works is a question that not only does not belong to science (!) but also a question that doesn’t even interest them (!!!).
To this day, that is one of the most intellectually chickenshit positions I have ever come across in my life.
I don’t listen to this sort of conversation often, which means I haven’t heard this sort of thing often. And (of course!) it’s not from literally everyone who says they like to “shut up and calculate”. But the two are definitely correlated strong enough that even that phrase – which is potentially innocuous on its own – still sets my teeth on edge.
It makes perfect sense for scientists to focus on the kinds of questions that they can share with the rest of the scientific community. In that case, something like which interpretation is true is not “scientific” precisely because the result of an unethical experiment could not be repeated by others across the world to confirm it. That’s a perfectly acceptable definition of science. (Not mine, but fine.) It’s not that these people aren’t interested in the question, or that they don’t think it’s important. Not at all. It’s that they think the question lies outside their profession as they practice it. “Shut up and calculate” makes a certain sense in that context.
But to add on top of that that they don’t even care about the question, well, there is a huge difference there.
I would hope that all of my points relating to the bare definition of determinism is clear.
I definitely made a big mistake leaving that topic and trying to discuss whether this particular world is deterministic.
For the record, I think yours are consistently the best posts in these sorts of threads.
You
-
Are likely better read on this that I am. And
-
Might be right. I don’t know. Maybe it’ll the gist of a future WordMan thread!
monstro, as longwinded and laborious clarifying what our words mean is, there are reasons to do it. Terms are often slippery beasts that need to be pinned down into place some lest the different meanings to different users result in mere talking past each other.
Do I believe that human mind decision making is magical and divorced from cause and effect? No. And I would find a discussion about FW as defined as that to be so trite as to be of no interest whatsoever. But my background is, I think we both appreciate, very different than yours. I was not raised with a Church influence. God as a concept was not something my family cared at all about and religion to us was more a matter of culture, belonging, and traditions, inclusive of intellectual ones, and really not the source of our everyday morality and ethics. A soul independent of the body is not a concept I grew up being exposed to much other than in fiction I read. My father was a salesman who would have loved to have studied psychology as an academic pursuit if only … The pulpit thumping concepts you react against that you label FW are ones that are not in my personal experience nor as the predominant version in the exposure I have to the concept of FW as a matter of philosophy.
As a matter of personal psychology I think the concepts and terms of FW and of determinism add more confusion than clarity.
We do not need to go there to appreciate that we are all the ever-changing result of many contributing causes, some nature and some nurture, and that our sentient selves have predispositions to behave in certain ways. We do not need those terms and the confusion they bring along with them to appreciate that our volitional sentient selves are just the tip of our information and states/needs processing information systems. Nor to consider how culture and biology have interacted to create them. The best balance in each circumstance between owning the locus of control and accepting our limitations, the degree that we are empathic and able to generously look at the world through the eyes of someone else whose behavior seems out of what we see as usual norms and the degree that we feel consequences of some sort are important parts of the inputs to all of us that keep social structures predictable and working for most … does not require considerations about the reality or illusion of FW and the nature of determinism.
IMHO.
On preview I see Hellestal has posted but I do need to run. Not sure when I will have the time to read and respond but please do not think I am ignoring you!
Perhaps it has been lost in all the words and speaking-past-one-another, but I am still confused what your particular definition of FW is. I have a good handle on the definitions expressed by others, and I think I understand the basis for these definitions. But I am not clear on your definition and the process that you have gone through to get to that definition.
If someone tells me they don’t believe in determinism, they are telling me they don’t believe that actions can be pre-determined. If you do think actions can be pre-determined but not with absolutely certainty, I can get with that as long as you are humble enough to admit that our models are likely always going to miss important variable(s) because knowledge has a tendency to be incomplete. But I still don’t think an imperfect model is evidence of FW. It just means that humans lack the ability to predict what another being is going to do 100% of the time. The limitations of humans doesn’t say anything about the nature of the universe.
I once asked my father if he believed in FW and without skipping a beat, he said “OF COURSE!” My father is a learned man with a scientific bent (although he is also religiously devout). He’s socially progressive and understands (just as you do) that decisions don’t arise in a vacuum. So I’m left wondering why he is so adamant that FW exists. What function does this belief serve for him that I’m somehow able to do without.
Hellestal seems to the take the same view that I do. I actually don’t know whether the behavior of people, cats, robots, or the universe is entirely determinant. But the determinist argument makes a lot of sense and jibes with the empirical evidence that is currently available to us. When discussing biological entities at least, there is more empirical evidence favoring determinism over FW. So f I were going to go with the most parsimonious option, which I’m inclined to as a scientist, I’d have to go with determinism. That is why when people shout in my face that OF COURSE we have free will, I demand they pony up some evidence. Not to be obnoxious, but because I’m a scientist and I don’t like hearing OF COURSE.
Thanks! I do wish I had a style that was as persuasive as yours, however. I can only speak through analogy because I lack the vocabulary.
No, you’ve definitely got an audience! Yeah, some of us are kinda at sea, but we’re watching with cat food on our breath! Or something!
I wonder, though, if this very technical stuff is necessary. It seems to me that this high-end debate doesn’t really serve to advance the argument either way, but just moves a part of it away from the ordinary guy’s level of comprehension.
Like, say we’re arguing creation/evolution, and two guys start going into the specific named genes that produce a specific enzyme, and get deeply into a debate over RNA transcription. Okay, great stuff, but does it really help anyone to reject creationism?
Cool and yes, I agree which is why I focused on the everyday implications.
Hellestal I appreciate your rigor. The opacity comes through length more than anything. These are long posts for this board.
Apologies if this derailed things. I might back out then. I agree totally and completely with DSeid that it’s important to define words like “decision” as rigorously as possible, but I won’t be able to do that in short posts. Relative rigor requires relative length.
It wasn’t my intent to choke up the traffic here with long posts. I think I just (finally!) saw a way to explain a relatively simple idea in a relatively simple way,* and I wanted to explore where that approach led. I think my style is… evocative? But it’s not terse, and there is no possible way to make it terse. That’s simply not how I think through things, and I write the same way I think.
For what it’s worth, I stand by my original statement in the thread:
In fact, I’m even more confident of that position now.
Given the length of my posts, I can even imagine a book: “How Philosophy Can Help You Make Better Financial Decisions”. Haha, how’s that for practical?
*(Not an easy idea, but nevertheless a relatively simple one. This remains an important distinction.)
If nothing about life changes either way then it is basically just mental masturbation.
Even if serial killers don’t have the will to choose not to murder, then courts don’t have the will to choose not to send them to prison.
So it’s basically just a “who came first, the chicken or the egg?” debate with no practical meaning.
You are obviously intelligent and have thought this through but it seems that you are claiming to know much more about how the universe works than anyone on the planet.
The fundamental understanding of quantum mechanics is a huge problem when making a deterministic argument.
Let me give you a very simple example that anyone can understand. For a long time there have been choose-your-own-adventure books and even simple computer programs that simulated them. Given a set of instructions (programming), the outcome would always be the same for a given set of inputs. That seems perfectly logical.
Now, what if I add something like this true quantum number generator to the decision process. They are commercially available and can “choose” whether you head East or West first but nobody, not even the best theoretical physicists in the world, can tell you why. In fact, it may be theoretically impossible to know that at all yet something is making that “choice”.
I am one of the few fundamentalist agnostics out there and, while I admire your efforts, I still seriously doubt that you have cracked the underlying decision making processes of the universe let alone how it all ties together with human consciousness.
What I think is really fun is that there might be a third way decisions get made: what if the computer program is so sophisticated that it has models of itself within itself, and observes its own decision-making process, so that the whole process involves depths of self-reference. Also, the decision-making process is divided into parallel machines, each having a different value system. Also, the system is able to model the consequences of its decision, if inexactly, by the process of imagination.
The process isn’t “random,” but some of its sub-elements do partake of randomness.
In a vacuum, my decision whether or not to cheat on my diet and have a doughnut might be predictable. But it may be purely random chance that I walk in front of a doughnut shop, exactly as they’re putting the honey-glazed bear claws on display in the window. A real-life, working deterministic model isn’t really able to correlate all the billions of little contingencies like that.
(Doggone it, I had to talk about doughnuts. Now I’m hungry! I’ve GOT to come up with a different metaphor!)
Hellestal, again, I appreciate the rigor and your posts are clear in what they are wrestling with. This thread has evolved past my attempt at a practical question and dug in a bit. I don’t think of it as “mental masturbation” (and thanks for that insightful drive-by, cj!) but the strict definitions take time.
And as for your book idea, it exists: The Black Swan, by Nassim Taleb (link to my thread about it.)
I am in no way a mod. I am just really enjoying this thread. It’s in IMHO not GD for all the discussion-not-debate reasons in that choice.
I agree with you, Hellestal is obviously intelligent and thought it through. I do not agree that he/she comes across like they are Philosopher-ier than Thou. A rigorous definition is not condescension.
You have the ability to engage at least a couple of thoughtful Determinists, given monstro’s great posts. Why not ask them in a way that promotes discussion vs. snark? You don’t have to agree with their positions, but you can make sure you understand them, yes?
Okay, a little window before running to a morning meeting. Hopefully that won’t prevent some brevity!
Hellestal
-
Yes, the view from a transcendental perch outside and not interacting with all that has ever been and will be in all of the infinities of branching universes, of the universal wavefunction, would in the MWI be deterministic. And the view from every point that exists within all the infinities of multiverses would always and must be of an indeterministic reality.
-
I can agree that we can leave the discussion about being the experiencing self within the transcendental Laplace supercomputer to the side for now.
-
I’ve no debate with a statement that a causally deterministic universe is potentially consistent with what we perceive, at least at a Newtonian level, and in the MWI only for a value of universe that includes more than what is my understanding of “universe” but for all possible multiverses in their unified waveform totality.
-
The quantum immortality thought experiment does not in fact produce any falsifiable predictions to the observations of any non-transcendental observer that would differentiate it from reigning models. For the sake of this thread I’d request any further discussion on that, if so desired, be moved to the MW GQ thread.
-
More debate on “shut up and calculate” is outside the scope of this thread as well.
My apologies if it has been less than clear.
I’ll back up some. “Decision” literally comes from the roots meaning “to cut off” and I have proposed means cutting off some possible actions as the result of consideration. Now let us ignore the word “possible” at this point. Consideration could be without sentience. The E. coli considers the concentration gradient of its food supply and then cuts off either run or tumble. It does not use sentience as part of the process. When a sentient intelligence uses the tools of its sentience as part of the consideration process and as a result takes an action, cutting off other minimally perceived as possible choices, as an act of volition, it experiences Free Will.
Free Will is that experienced sense of volition after consideration, which occurs, to my way of thinking, as a necessary consequence of a sense of self in a sentient intelligence, and may even be the prime driver for that sense of self to exist.
Now I am willing to go with calling Free Will “real” as it is a real experience. I think therefore I am and I decide therefore I can decide. And I cannot argue too much against calling that “an illusion” … like our complete sense of self and all our perceptions ultimately are … but again, as per my very first more concise post, if it is an illusion (because choice is not actually possible) then it a necessary one.
Is that less opaque?
“Consciousness” emerges via decision-making - it is a set of potential inputs until an act of volition collapses them to a specific action.
At a purely conceptual level, it sounds similar to what happens at a sub-atomic level, when quantum potential waves collapse to a single “particle” when an observer engages.
I find a form of that weird, paradoxical “pragmatic transition from potential to specific” at the heart of the Big Questions I read up on. I think that, past metaphorical use, there is something in common across the various paradoxes here.
Yeah we need a Keanu Reeves smiley.
Anyway my personal point has never been so profound. It’s just that as soon as we start talking about “decisions” we start implicitly ignoring (not denying) determinism. An “agent” is that which decides from among the possible. “Agency” is the ability to, the freedom to (if you will) make choices between possibilities. When we as sentient intelligences experience agency, act as agents, we are experiencing FW.
Personally I can’t get too excited about randomness at the base. So what if the world is not really deterministic? Random outcomes no more implies Will than does determinism. And if there was some magical something that was not of the identified stuff that was “me” causing things to happen? To my thinking if that magical somethings exists then it is real too, not magic, just some real, previously unidentified, and poorly characterized force that we can label as the M-force, and is also part of any causal chain.
If you’ve not read any Doug Hofstadter you should. You are coming up with similar thoughts to what he expouses. The classic Gödel, Escher, Bach: An Eternal Golden Braid is a very fun read but he makes the point of his thesis much more succinctly in his 2007 I Am a Strange Loop (a review here). Most briefly put the thesis is that sentience/consciousness emerges precisely from those self-referential recursive always updating nested levels of loops that are information systems which include their own constantly changing selves as members of the sets they are monitoring and changing in response to changes in …
From that SciAm review:
You’ll enjoy it I think.
I have definitely GEB, and my opinions here are, to some degree, informed by it.
I had not heard of “I Am a Strange Loop” and will definitely obtain it! Thank you for pointing it out to me!
I agree with all that but that’s why I think you should question whether the concept itself even makes sense.
When even when we imagine a magical universe with whatever rules, or lack of, and we still can’t see a way for free will to exist then already we get a bad smell. And indeed, on closer examination, we find FW is incoherent maybe to the point of being self inconsistent.
I want to second DSeid’s views of these books.
Gödel, Escher, Bach is a sparkling, resplendent, dazzling trip into these ideas. I Am a Strange Loop is more prosaic, but also much more directly to the point. It’s more explicitly argued and carefully explained investigation into the question of consciousness. Both are definitely worth it.
Honestly, my feeling is that Taleb’s earlier Fooled by Randomness is a better book than The Black Swan. (Although it’s been a while for both…)
In the unlikely chance that I tried to write my own thing, it wouldn’t look much like either.
Returning to the definition of “decision”, there are actually two basic questions here I’ve been asked.
-
How can “decision” be defined in the context of a deterministic system?
-
Is it really important to have such a definition?
I’m going to skip Question 2 here. It’s a bigger issue than the first and much harder to discuss. I’m going to focus solely on how we might determine what a decision could possibly be in the context of a deterministic system. I also want to mention that it’s best to try to avoid the etymological fallacy here. I don’t particularly care about the roots of the word. Manure is derived etymologically from “work done by hand”. But it’s work-done-by-hand to believe that manure still means that.
In order to define “decision” in the context of a deterministic system, we want to build up from our intuition. That is, we don’t want to be arbitrary. We want some definition of the word that could seem to apply in exactly the same sense that we seem to use the word in everyday contexts.
And frankly, I think this is fairly straightforward to do. It just takes some working out.
I want to go back to the notion of those programmed worlds, but this time, I want to talk about the tasks that are faced by the programmers. First is the programming group that is tasked with building the “physics” of the game world itself. We can imagine the shape of this world as some sort of square- or hex-based grid, with “agents” in the world occupying spaces in the grid. It can be a torus, where the east and west attached to each other, and likewise the north and south (so that travelling around can send you back where you started). There might be weather patterns, terrain features (forest, grassland, ocean, etc), and those weather patterns might evolve with the input from a pseudo-random number generator with a given seed.
It’s one of the funny things about the words “random” and “pseudo-random” that the former is ill-defined, and the latter extremely well defined. A pseudo-random system with a fixed seed will do the same thing every single time. It is deterministic, in other words. Based on the perspective of any agents within the weather system, the progression of weather patterns should be a bit unpredictable. This is not because the weather system is genuinely random. It’s deterministic. But it’s deterministic based on hidden variables: a pseudo-random series of outputs that agents within the weather system have literally no access to.
In addition, the game world might have a few proper beasties. These won’t be real agents. They’ll just spawn pseudo-randomly, and walk around and attack anything they see.
After the world is created with its base rules, we can look at the much harder problem of the other programming teams who are trying to define (i.e. program) the agents that inhabit this world. Imagine that there are many, many teams of programmers, each one of them with some slight bit of knowledge about how the world works. Their agents can WALK north, or WALK south, and so on. The agents can see several grid spaces away from where they’re standing. They can see the terrain features, any GOLD that is on the ground, and also other agents that are walking around.
The programming teams for the agents have the task of having their agent have the most GOLD (or whatever) when the simulation runs down. One way of doing this might be to program their little mechanical robots in the same way the beasties are programmed. Just a string of if-then sequences. If you see GOLD, pick it up. Otherwise, walk in this pattern until you see gold. That might actually be effective for this particular game, but I want this hypothetical to be more interesting than that, so we won’t say that those simple if-then statements are actually “agents”.
To be an agent that makes decisions requires something more. It requires that the agent be constantly building a map that correlates with the territory (the game world) in which they are embedded.
I want to emphasize here that this is much, much, much tougher task. I’m going to try to speak generally about what this sort of agent would look like, but I’m not actually going to try to “program” it. I’m just going to point out the various problems that would face the programmers when they tried to create a proper agent that was capable of making proper decisions within the context of this game world. What it might have, first, is some prior notion of how the game world should work: it’s okay to ATTACK bunnies, but it’s a mistake to ATTACK dragons (until the character levels up or whatever).
This agent would not just be walking around blindly until it found GOLD to pick up. Every step it took would update its information. Every meeting with a bunny would update its notion of how likely it was to survive that encounter. While the agent might reasonably run away from a bunny early in the simulation, as it becomes more “skilled” with a higher level, it might reassess the costs and benefits from fleeing.
This is a huge, huge difference from the walk-around-according-to-this-pattern-and-pick-up-GOLD-if-you-see-it algorithm.
What the agent now has is not just blind instructions for navigating the territory. This is a multi-layered creature now. FIRST: It has a constantly updating map of the territory, which is correlated with the information it receives from navigating that territory. And for proper decision-making on a higher level, an important part of this map is how the agent itself fits into the game world. It’s not just a string of information that matches the territory. The agent itself is part of that string of information. The map isn’t just a plain map, but has a big red symbol ‘you are here’ in the middle of it.
(Our internal maps of the universe we inhabit are not just soul-less pieces of mental parchment. In all of our conceptions of the universe, we have a tag also that includes where we fit inside of this universe.)
SECOND, the agent has some sort of objective function that it’s trying to accomplish, like gathering the most gold. The previous if-then style of robot did not have this. It was programmed to pick up GOLD if it saw gold, and maybe to ATTACK rabbits if it saw a rabbit, but that was just a mindless trigger. This objective function is much different. The idea here is: max E[gold]. The agent is trying to maximize the future expected value of gold, based on its knowledge of the world around it. And so the very existence of an objective function requires a third feature for the agent.
THIRD, the agent must have some sort of optimization algorithm in which it tries to compute the action that is most likely to maximize its objective function. This is to say that the agent must “consider” the action most likely to get it what it wants, perhaps through calculus or through some computationally cheap approximation of a calculus-based optimization algorithm. It will weigh internally the likelihood of getting what it wants in order to determine the action that seems most likely to maximize its objective function, given the continually updating information about the world that it has available.
That is decision making.
There’s one other point I want to note.
When the World+Agent1+Agent2+Agent3+etc. program is run, it will run in the same way every time. It will be a deterministic system. It will run the same way every time. But early in the simulation, observers might see Agent5 run away from a rabbit early in the simulation, and subsequently attack rabbits late in the game. This is all part of the agent solving its objective function based on new information from a continually updating and evolving world.
But there is not any actual necessity for there to be any “observers” to this game. There might not be any visual representation for people to see. The terrain might be marked by unevocative labels like T0025B. The beasties might not be dragons and bunnies, but rather B7423C4. The programmers, instead of maximizing the GOLD for their little robots, might be seeking to maximize X. And there might be literally zero graphic art for any of this that a casual bystander could understand on an intuitive level.
Nevertheless, the world will play out the same way every time. When the program is compiled, it will all be a sequence of ones and zeroes. To someone who sees nothing but that output, the whole thing might make less than zero sense. This is very important! The series of ones and zeroes has an exact and precise relationship with the “code” as described above. There really are agents. This is to say: there really are objective functions, maps that become more correlated with the territory, and optimization algorithms to maximize the expected objective function, given the limited information that exists. All of this exists. We know it exists, because we can easily imagine the sort of code that would manage this sort of thing. (This is NOT an easy programming task. But nevertheless, it’s well defined and easy to imagine this sort of thing coming about.)
A person who was looking at the machine code, at all those ones and zeroes, would not necessarily have the slightest idea what the program was trying to accomplish. It might all look like gobbledygook. But the ones and zeroes are just another expression of the original higher level code – isomorphic to that original code. Just because a person might look at the ones and zeroes and not recognize anything does not mean that nothing is happening.
In the many-worlds interpretation, this universe is just the deterministic evolution of a universal wavefunction. An outside observer, looking at the deterministic evolution of that universal wavefunction, might not recognize that there are many worlds being simulated by the deterministic evolution of that function.
The view of the algorithm from the outside might be very different than the view of the algorithm from the inside.