It’s possible to have that argument and we had it multiple times in the past in discussion in which Brillyant participated. He is here saying that the position that “CICO is a bad control model” is anti-advice. He seems to consider that it’s a good enough control model to allow him to lose the weight he wants to lose.
I’m not sure on what it specifically means the word “anti-advice”, but if it’s along the line of excluding possibly useless models, then sure, it’s anti-advice but it’s still useful.
He seems to consider that it’s a good enough control model to allow him to lose the weight he wants to lose.
Yeah, but if it only works for him and a few others, and not for everyone else, can you still say that it’s a good control model? The argument from Brillyant to me seems like: any sufficiently analyzed control model is indistinguishable from a physics model. Which is true, but useless. What I want to know, and the added value of a control vs physics point of view, is which and where are the hidden knobs and levers that controls intake and consumptions. Brillyant named some, I named others. I think we can simply dissolve the question by saying:
CI = willpower + feedback from exercise + feedback from previous meals + feedback from fat cells + genetic predispositions + environmental factor + unknown unknows
CO = willpower + feedback from previous exercise + feedback from NEAT + feedback from food + genetic factors
absorption defects + unknown unknows
I’m not sure on what it specifically means the word “anti-advice”, but if it’s along the line of excluding possibly useless models, then sure, it’s anti-advice but it’s still useful.
The paragraph you wrote has the potential to make a reader believe they have less agency about weight loss and thus be less motivated to do the straightforward actions that the CICO model recommends.
While you claim to steelman you don’t provide any arguments for which you believe that isn’t the case and why believing no-CICO would be better for someone who wants to lose weight.
The paragraph you wrote has the potential to make a reader believe they have less agency about weight loss
Well, if that’s true, then we as rationalist should embrace that.
While you claim to steelman you don’t provide any arguments
Because the argument is complex and because I’m not sufficiently invested. I was suggesting a possibility in the landscape of possible counter-arguments.
why believing no-CICO would be better for someone who wants to lose weight
This is straightforward: because some people with low agency might obtain better results acting on other inputs, as per Viliam iron deficiency.
I take, “bad control model” to mean, “it explains weight in terms of cico but the phrase cico does not tell you about the hard step of making your brain go along with it (the control model)”.
I agree with that, but I would also suggest that even a bad control model is useful compared to terribly wrong models claiming to be right, for we know this model is wrong.
A while ago a good friend asked me what he could do to increase his typing skill. I didn’t give him the straightforward advice of using a type training program but I talked to him about the promises of Dvorak. He didn’t take any action, didn’t increase his typing skills or switched to Dvorak.
Adding information reduced his impulse to take action. On the same token, it’s not simply about comparing CICO against other wrong models but simply about having a person who wants to lose weight being committed to a model and doing what the model prescribes.
It’s possible to have that argument and we had it multiple times in the past in discussion in which Brillyant participated. He is here saying that the position that “CICO is a bad control model” is anti-advice. He seems to consider that it’s a good enough control model to allow him to lose the weight he wants to lose.
I’m not sure on what it specifically means the word “anti-advice”, but if it’s along the line of excluding possibly useless models, then sure, it’s anti-advice but it’s still useful.
Yeah, but if it only works for him and a few others, and not for everyone else, can you still say that it’s a good control model?
The argument from Brillyant to me seems like: any sufficiently analyzed control model is indistinguishable from a physics model. Which is true, but useless. What I want to know, and the added value of a control vs physics point of view, is which and where are the hidden knobs and levers that controls intake and consumptions.
Brillyant named some, I named others.
I think we can simply dissolve the question by saying:
CI = willpower + feedback from exercise + feedback from previous meals + feedback from fat cells + genetic predispositions + environmental factor + unknown unknows
CO = willpower + feedback from previous exercise + feedback from NEAT + feedback from food + genetic factors absorption defects + unknown unknows
CI—CO = weight gains.
Is this better?
The paragraph you wrote has the potential to make a reader believe they have less agency about weight loss and thus be less motivated to do the straightforward actions that the CICO model recommends.
While you claim to steelman you don’t provide any arguments for which you believe that isn’t the case and why believing no-CICO would be better for someone who wants to lose weight.
Well, if that’s true, then we as rationalist should embrace that.
Because the argument is complex and because I’m not sufficiently invested. I was suggesting a possibility in the landscape of possible counter-arguments.
This is straightforward: because some people with low agency might obtain better results acting on other inputs, as per Viliam iron deficiency.
I take, “bad control model” to mean, “it explains weight in terms of cico but the phrase cico does not tell you about the hard step of making your brain go along with it (the control model)”.
I agree with that, but I would also suggest that even a bad control model is useful compared to terribly wrong models claiming to be right, for we know this model is wrong.
A while ago a good friend asked me what he could do to increase his typing skill. I didn’t give him the straightforward advice of using a type training program but I talked to him about the promises of Dvorak. He didn’t take any action, didn’t increase his typing skills or switched to Dvorak.
Adding information reduced his impulse to take action. On the same token, it’s not simply about comparing CICO against other wrong models but simply about having a person who wants to lose weight being committed to a model and doing what the model prescribes.