Some ideas about Solomonoff that might be of use: he is trying to raise induction from given deductive bases into somehow being “forwardly deductive” rather than inductive. No doubt we can refine our bases for inductive hypotheses to make them as deductively strict as possible, but ultimately if they are too deductively strict we leave no room for induction and would merely be proposing a continuation of the status quo as a process leading inevitably, deductively, somewhere.
Induction is a process from supposed real facts to hypothesized real facts, and when the hypothesized real facts eventually arise, we deduce the accuracy of our hypothesis about them. We can hypothesize backwards about the bases for the supposed real facts, to open issues about their supposed reliability as the basis for a future hypothesis, and use deduction to check again and decide again their reliability as facts, but we should have already done that before relying on them for the future hypothesis.
Hypotheses are not limited to the future, and we can hypthesize about the basis for anything to open up argument, and use deduction to be satisfied we are secure. What Solominoff is doing is identifying some important factors to deductively refine our induction when hypothesizing backwards or forwards. My concern is that creativity in working real facts into inductive hypotheses in reality can be guided by deductive refinements, but only broadly. The entire point of induction is to open up, and deduction is to close down, so I would not get too excited about Solominoff’s attempt at limiting induction to some kind of deductive certainty. All he is doing is building deductive refinements into his inductive methods.
I agree that the status quo may be a process leading inevitably statistically somewhere, and that we can try to use past deduced bases for the status quo to inductively predict its future. That is nice and secure, and perhaps a little like dominoes falling in line. In reality, data is more complicated, and interpreting it using induction (refined by deduction) is more creative. I will read Solominoff’s gudelines again, but I assume they are a product of his consideration of all the fundamental logical and practical issues that need consideration for greatest certainty, and if so, useful. However, I like looking at nature around me to see the factual complexity that needs reconciliation by hypotheses, and work on that rather than dredging too much through the basic formalisms, Facts rule at the end of the day.
The entire point of induction is to open up, and deduction is to close down, so I would not get too excited about Solominoff’s attempt at limiting induction to some kind of deductive certainty.
The first part of the quote (before “so”) seems right—induction generates hypotheses, and deduction destroys them (for loose definitions of in/deduction). One way to see Solomonoff induction is that we first “open up” as far as possible, by generating all possible hypotheses, and then use deduction to “close down”, by throwing away all the invalid hypotheses.
Could you possibly provide a simple reason why it is wrong, to let me know what to look for if I go to your links? It is fine if you have no time to provide a simple reason, rather than “this seems wrong”, but I would much prefer any reason at all or any reasoning at all. Just a short sentence would be fine to address your key point. Otherwise it appears disrespectful, like “back to the drawing board, lad” without any reason whatesoever. I am happy to argue my post above, which explains very clearly the meaning of the quote you chose. but I cannot go chasing rabbits of a decription I do not know, were I to chase rabbits. See this as a challenge Vladimir, in response to what seems a lazy reply.
Induction, creativity or any other aspect of intelligence can run on a completely deterministic computation, and in a certain sense require determinism/structure to be expressed. The aspect of intelligence and choice that feels like it requires arbitrariness or uncertainty results from logical uncertainty, from not knowing some of the facts implied by what you already know, including the facts you yourself determine.
I agree that nature might even be entirely deterministic, but only statistically in particle & field behaviour, and ways poorly understood in biology, but that is not the point. It is obvious that we are resoning to a deductive conclusion because we hope that the solution, if there is one (a Unified Theory for example) explains a self-consistent state and our path to discovering it. The entire point is that in the practical journey to this hopeful order we use inductive creativity that is as open as we need to gather and refine into a stricter theory. I think you have set up a straw man argument against me, and your answer does not seem to relate directly to the quote you provided, which is consistent with the above. It is clearly foolish to assume in advance of deductive confirmation that we will get it, which is why we open up to be creative. We frequently revisit the deduced bases for our induction by induction and further deuction to refine them as the process works both ends.
As to the side issue you seem to raise in support of the claim that I “seem wrong”, namely the appearance of choices when in fact we might be better off deductively guided to the best option, that is a non-argument. If we had that deductive guidance we wouldn’t be in a process of reasoning about options to begin with. No matter which way you look at it, we are free to create when we have no clear deductive solution, and that’s what we do when we use what you seem to call “Free Will” (a much broader issue that your identified point). Logical uncertainty is our reality until satisfied with reasoning to something more secure, as fact rule at the end of the day and we are nowhere near reconciling them across all sciences. Perhaps you should read my book, its free at my website http://www.theumandesign.net and deals with these issues. I see a lot of rigidity in your approach, perhaps leading to a misunderstanding of what is really going on when we discover things.
Vladimir, are you at liberty to confirm whether you have provided any of my various posts today with negative votes, and how many? I await your detailed reply to the basic issues I have outlined above in any event, but I read somewhere that it is common for unexplained negativity from others to be explained when asked. What is your explanation, if any?
I downvoted you because you keep plugging your book, and your comments suggest I’d gain nothing by reading it. Also you don’t seem to have engaged with the reasons for that impression, though people have tried to point some of them out. Perhaps you should start by assuming that you have no clue what we believe, and keep that in mind as you read the links someone gave you.
Vladimir, are you at liberty to confirm whether you have provided any of my various posts today with negative votes, and how many?
Your challenge here prompted me to downvote each of them myself, after a brief confirmation. The multiple advertisements for your “book” also don’t help.
I await your detailed reply to the basic issues I have outlined above in any event, but I read somewhere that it is common for unexplained negativity from others to be explained when asked. What is your explanation, if any?
Downvotes mean “I want to see less comments like this one”. Since this currently applies to every one of your comments you may consider contrasting your comments with the ones that are voted highly and seeing if there are any features worth emulating.
An attempt at formulating an explanation for the negativity might begin with:
“Your posts so far contain many critical and extremely destructive mistakes outlined in the Sequences, the core material usually discussed on this site and generally assumed to be at least partially understood by most commenters.”
However, I find myself having trouble identifying further probable causes. I suspect the main reason for this is that when I read your posts, there’s something out there in the back of my mind screaming bloody murder get me away from here arggh!… in a manner very similar to what I described in response to your post in the Welcome to LessWrong topic.
If you’re confident that this place is somewhere you want to be, it would be most prudent to read the above-linked Sequences and re-think your approach when writing comments—a personal tip would be to start smaller, make simpler, less-bold assertions, and use the feedback on those to more clearly build a mental model of the norms and standards of this community.
Otherwise, as Eliezer suggested, you’d best start looking elsewhere.
Some ideas about Solomonoff that might be of use: he is trying to raise induction from given deductive bases into somehow being “forwardly deductive” rather than inductive. No doubt we can refine our bases for inductive hypotheses to make them as deductively strict as possible, but ultimately if they are too deductively strict we leave no room for induction and would merely be proposing a continuation of the status quo as a process leading inevitably, deductively, somewhere.
Induction is a process from supposed real facts to hypothesized real facts, and when the hypothesized real facts eventually arise, we deduce the accuracy of our hypothesis about them. We can hypothesize backwards about the bases for the supposed real facts, to open issues about their supposed reliability as the basis for a future hypothesis, and use deduction to check again and decide again their reliability as facts, but we should have already done that before relying on them for the future hypothesis.
Hypotheses are not limited to the future, and we can hypthesize about the basis for anything to open up argument, and use deduction to be satisfied we are secure. What Solominoff is doing is identifying some important factors to deductively refine our induction when hypothesizing backwards or forwards. My concern is that creativity in working real facts into inductive hypotheses in reality can be guided by deductive refinements, but only broadly. The entire point of induction is to open up, and deduction is to close down, so I would not get too excited about Solominoff’s attempt at limiting induction to some kind of deductive certainty. All he is doing is building deductive refinements into his inductive methods.
I agree that the status quo may be a process leading inevitably statistically somewhere, and that we can try to use past deduced bases for the status quo to inductively predict its future. That is nice and secure, and perhaps a little like dominoes falling in line. In reality, data is more complicated, and interpreting it using induction (refined by deduction) is more creative. I will read Solominoff’s gudelines again, but I assume they are a product of his consideration of all the fundamental logical and practical issues that need consideration for greatest certainty, and if so, useful. However, I like looking at nature around me to see the factual complexity that needs reconciliation by hypotheses, and work on that rather than dredging too much through the basic formalisms, Facts rule at the end of the day.
This seems wrong, see Lawful intelligence, Free will.
The first part of the quote (before “so”) seems right—induction generates hypotheses, and deduction destroys them (for loose definitions of in/deduction). One way to see Solomonoff induction is that we first “open up” as far as possible, by generating all possible hypotheses, and then use deduction to “close down”, by throwing away all the invalid hypotheses.
Could you possibly provide a simple reason why it is wrong, to let me know what to look for if I go to your links? It is fine if you have no time to provide a simple reason, rather than “this seems wrong”, but I would much prefer any reason at all or any reasoning at all. Just a short sentence would be fine to address your key point. Otherwise it appears disrespectful, like “back to the drawing board, lad” without any reason whatesoever. I am happy to argue my post above, which explains very clearly the meaning of the quote you chose. but I cannot go chasing rabbits of a decription I do not know, were I to chase rabbits. See this as a challenge Vladimir, in response to what seems a lazy reply.
Induction, creativity or any other aspect of intelligence can run on a completely deterministic computation, and in a certain sense require determinism/structure to be expressed. The aspect of intelligence and choice that feels like it requires arbitrariness or uncertainty results from logical uncertainty, from not knowing some of the facts implied by what you already know, including the facts you yourself determine.
I agree that nature might even be entirely deterministic, but only statistically in particle & field behaviour, and ways poorly understood in biology, but that is not the point. It is obvious that we are resoning to a deductive conclusion because we hope that the solution, if there is one (a Unified Theory for example) explains a self-consistent state and our path to discovering it. The entire point is that in the practical journey to this hopeful order we use inductive creativity that is as open as we need to gather and refine into a stricter theory. I think you have set up a straw man argument against me, and your answer does not seem to relate directly to the quote you provided, which is consistent with the above. It is clearly foolish to assume in advance of deductive confirmation that we will get it, which is why we open up to be creative. We frequently revisit the deduced bases for our induction by induction and further deuction to refine them as the process works both ends.
As to the side issue you seem to raise in support of the claim that I “seem wrong”, namely the appearance of choices when in fact we might be better off deductively guided to the best option, that is a non-argument. If we had that deductive guidance we wouldn’t be in a process of reasoning about options to begin with. No matter which way you look at it, we are free to create when we have no clear deductive solution, and that’s what we do when we use what you seem to call “Free Will” (a much broader issue that your identified point). Logical uncertainty is our reality until satisfied with reasoning to something more secure, as fact rule at the end of the day and we are nowhere near reconciling them across all sciences. Perhaps you should read my book, its free at my website http://www.theumandesign.net and deals with these issues. I see a lot of rigidity in your approach, perhaps leading to a misunderstanding of what is really going on when we discover things.
Vladimir, are you at liberty to confirm whether you have provided any of my various posts today with negative votes, and how many? I await your detailed reply to the basic issues I have outlined above in any event, but I read somewhere that it is common for unexplained negativity from others to be explained when asked. What is your explanation, if any?
I downvoted you because you keep plugging your book, and your comments suggest I’d gain nothing by reading it. Also you don’t seem to have engaged with the reasons for that impression, though people have tried to point some of them out. Perhaps you should start by assuming that you have no clue what we believe, and keep that in mind as you read the links someone gave you.
Your challenge here prompted me to downvote each of them myself, after a brief confirmation. The multiple advertisements for your “book” also don’t help.
Downvotes mean “I want to see less comments like this one”. Since this currently applies to every one of your comments you may consider contrasting your comments with the ones that are voted highly and seeing if there are any features worth emulating.
The best part is the paragraph on the bottom of page 66.
Thank you for that.
It’s gone. What was it?
An attempt at formulating an explanation for the negativity might begin with:
“Your posts so far contain many critical and extremely destructive mistakes outlined in the Sequences, the core material usually discussed on this site and generally assumed to be at least partially understood by most commenters.”
However, I find myself having trouble identifying further probable causes. I suspect the main reason for this is that when I read your posts, there’s something out there in the back of my mind screaming bloody murder get me away from here arggh!… in a manner very similar to what I described in response to your post in the Welcome to LessWrong topic.
If you’re confident that this place is somewhere you want to be, it would be most prudent to read the above-linked Sequences and re-think your approach when writing comments—a personal tip would be to start smaller, make simpler, less-bold assertions, and use the feedback on those to more clearly build a mental model of the norms and standards of this community.
Otherwise, as Eliezer suggested, you’d best start looking elsewhere.
All of your recent comments have been downvoted. I suggest pursuing other forum opportunities online—LessWrong is not a good fit for you. Goodbye.