If you'll indulge me to make some assertions that I can't fully justify in the moment - I think there is a common theme to our thinking, which I will call incremental satisficing in conceptual space.
Satisficing is traditionally thought of as creating good enough solutions to problems. The idea here is to apply it to the concepts we form during education. That is, maybe it's okay to give a person a concept that is "wrong" from a higher vantage point, but functional, because it can be considered just a step on the ladder to enlightenment.
My recent post on the Glitch Gremlin is an example of this. You might perhaps want to never need gremlins. Ideally, you want to simply know the optimal way to apply your capabilities to all problems, and assign the optimal amount of emotional regret to yourself and others when you make a mistake or suboptimal choice. That Ultimate Optimal solution might not involve gremlins, but the "good enough solution for now" might.
I also think the Ultimate Optimal solution might actually involve gremlins. I'm not totally discounting that possibility!
The way I'm thinking about education right now is about making sure the local minimum is the best. That doesn't contradict satisfiscing, but it will mean satisfiscing diff things.
Physics proves that incremental satisfying is possible in an especially satisfying way, where nothing needs to be thrown away. All approximations are useful in different situations AND they can be rigorously related to one another. THAT is the physics envy that all other disciplines should have in my opinion.
About time that I caught up on your posts a bit 💪 I'll probably have several comments. I'll fire them off as I think of them.
My first is on the Gorgias Problem. This problem is induced by incomplete specification + optimization. In other words, it is almost ALWAYS the case that when you specify an optimization problem incompletely, the best solution will NOT be exactly what you want.
This is a fundamental, universal problem that extends all the way to things like fitting lines to data by least squares. The "best" solution, in the sense of minimizing squared error, is almost never the true solution, because the unknown truth cannot be included as a constraint in the specification.
This is the wellspring of many strange phenomena. For example, it's long been known in machine learning research that it can be beneficial to stop early when optimizing the fit of a model to data, because fully optimizing the fit eventually sends you past (the closest approach to) what you wanted and into the regime where you're overfitting, making the solution worse while still improving the optimization criterion.
Another way to put it: if you get an answer to a question and you don't like the answer, you should check if it is a technically valid answer to the question. If it is, the question needs work. If it isn't, the student needs more context to understand, or scaffolding to construct, the answer.
Socratic questioning may play a role in discovering what context is missing.
Re "the question needs work" - maybe you are looking for a step between the answer you didn't like and the action to improve the question. The step of asking why this happened? What did you expect the student to know or value that would have led them to reject this answer in favor of the correct one?
the space I'm thinking about rn is in places where the student is often giving the right answer, but without understanding. later on, the student just can't even guess, because the pattern is too complex.
But wrt shallowness discounting in general, I'm not sure it neatly fits 'question-answer' dynamics. It's easier to appear kind than be kind. etc.
I also feel like I am just barely hanging on to your meanings, and that I understand you ~70%
There may or may not be anything profound for you to find in what I'm saying. I haven't taught in a classroom in about 13 years. I do work with my children a fair amount on academic things, especially math. I don't think I really see much of this shallowness problem, because I have many many ways of testing for understanding, which I weave into work sessions on a second-by-second basis.
I work with my kids too, and their habits of thinking are so distant from the ones of students I do remedial math with.
I think the second by second basis evaluation is a big thing. For my kids, it made it easy to instill habits where the locus of truth was in their own perspective (they don't want to please me, they want the TRUTH), and where if you don't know or understand, you just say that.
But the schooled kids I'm dealing with have horrific habits. They will slip into guessing the moment they don't know (they rarely say "I don't know", and if they do, it doesn't mean "let's find out", it means "talk at me while I don't listen"
I'm still confused how this happens at such massive scales. Is it because the teachers themselves can't distinguish between rote learning and understanding? Or don't value the difference properly? Or is there just not enough time? I'm really unsure of this.
Plato wrote a dialogue called Gorgias, named after the same Sophist. Gorgias boasts that he teaches the most important skill, that of persuasion. And Socrates starts digging in, and it becomes clear that to be really persuasive yet ignorant of the topic at hand means you aren't instructing the other person, rather 'tricking' them. He makes some analogies pastry-baking:nutrition, cosmetics:gym-trainer, rhetorician:philospher. The first of these is a 'knack', a crafty creation of something that seems good but doesn't hold up. the second gets to the real thing.
The concept of discount isn't really there explicitly, only that it's easier to learn from Gorgias how to be persuasive than it is to learn to be an expert in the actual field you want to be persuasive in. And that even then, Gorgias's rhetoric might be more effective than your real knowledge.
Good idea. Though perhaps discount rather than discounting. I thought about the term for a little bit, unsure whether you were talking about lower cost or willful downplaying.
The shallowness discount is a big problem in online discussions. Commenters by and large trade in shallowness, and content creators favor it overwhelmingly.
I'm consistently disappointed by good titles about interesting things that turn out to not talk at all about the interesting implications of the title.
If you'll indulge me to make some assertions that I can't fully justify in the moment - I think there is a common theme to our thinking, which I will call incremental satisficing in conceptual space.
Satisficing is traditionally thought of as creating good enough solutions to problems. The idea here is to apply it to the concepts we form during education. That is, maybe it's okay to give a person a concept that is "wrong" from a higher vantage point, but functional, because it can be considered just a step on the ladder to enlightenment.
My recent post on the Glitch Gremlin is an example of this. You might perhaps want to never need gremlins. Ideally, you want to simply know the optimal way to apply your capabilities to all problems, and assign the optimal amount of emotional regret to yourself and others when you make a mistake or suboptimal choice. That Ultimate Optimal solution might not involve gremlins, but the "good enough solution for now" might.
I also think the Ultimate Optimal solution might actually involve gremlins. I'm not totally discounting that possibility!
Satisfiscing wrt education for sure!
The way I'm thinking about education right now is about making sure the local minimum is the best. That doesn't contradict satisfiscing, but it will mean satisfiscing diff things.
Physics proves that incremental satisfying is possible in an especially satisfying way, where nothing needs to be thrown away. All approximations are useful in different situations AND they can be rigorously related to one another. THAT is the physics envy that all other disciplines should have in my opinion.
*incremental satisficing
About time that I caught up on your posts a bit 💪 I'll probably have several comments. I'll fire them off as I think of them.
My first is on the Gorgias Problem. This problem is induced by incomplete specification + optimization. In other words, it is almost ALWAYS the case that when you specify an optimization problem incompletely, the best solution will NOT be exactly what you want.
This is a fundamental, universal problem that extends all the way to things like fitting lines to data by least squares. The "best" solution, in the sense of minimizing squared error, is almost never the true solution, because the unknown truth cannot be included as a constraint in the specification.
This is the wellspring of many strange phenomena. For example, it's long been known in machine learning research that it can be beneficial to stop early when optimizing the fit of a model to data, because fully optimizing the fit eventually sends you past (the closest approach to) what you wanted and into the regime where you're overfitting, making the solution worse while still improving the optimization criterion.
This make sense. For some strange reason I don't find the description of
(specification - error) x optimization = seeming/being
To be incredibly fruitful at first glance. It feels like it should be. Maybe I'm not a mathematician?
Another way to put it: if you get an answer to a question and you don't like the answer, you should check if it is a technically valid answer to the question. If it is, the question needs work. If it isn't, the student needs more context to understand, or scaffolding to construct, the answer.
Socratic questioning may play a role in discovering what context is missing.
Re "the question needs work" - maybe you are looking for a step between the answer you didn't like and the action to improve the question. The step of asking why this happened? What did you expect the student to know or value that would have led them to reject this answer in favor of the correct one?
the space I'm thinking about rn is in places where the student is often giving the right answer, but without understanding. later on, the student just can't even guess, because the pattern is too complex.
But wrt shallowness discounting in general, I'm not sure it neatly fits 'question-answer' dynamics. It's easier to appear kind than be kind. etc.
I also feel like I am just barely hanging on to your meanings, and that I understand you ~70%
There may or may not be anything profound for you to find in what I'm saying. I haven't taught in a classroom in about 13 years. I do work with my children a fair amount on academic things, especially math. I don't think I really see much of this shallowness problem, because I have many many ways of testing for understanding, which I weave into work sessions on a second-by-second basis.
I work with my kids too, and their habits of thinking are so distant from the ones of students I do remedial math with.
I think the second by second basis evaluation is a big thing. For my kids, it made it easy to instill habits where the locus of truth was in their own perspective (they don't want to please me, they want the TRUTH), and where if you don't know or understand, you just say that.
But the schooled kids I'm dealing with have horrific habits. They will slip into guessing the moment they don't know (they rarely say "I don't know", and if they do, it doesn't mean "let's find out", it means "talk at me while I don't listen"
I'm still confused how this happens at such massive scales. Is it because the teachers themselves can't distinguish between rote learning and understanding? Or don't value the difference properly? Or is there just not enough time? I'm really unsure of this.
Could you explain the origin of connecting the problem of shallowness discount with the name Gorgias?
Plato wrote a dialogue called Gorgias, named after the same Sophist. Gorgias boasts that he teaches the most important skill, that of persuasion. And Socrates starts digging in, and it becomes clear that to be really persuasive yet ignorant of the topic at hand means you aren't instructing the other person, rather 'tricking' them. He makes some analogies pastry-baking:nutrition, cosmetics:gym-trainer, rhetorician:philospher. The first of these is a 'knack', a crafty creation of something that seems good but doesn't hold up. the second gets to the real thing.
The concept of discount isn't really there explicitly, only that it's easier to learn from Gorgias how to be persuasive than it is to learn to be an expert in the actual field you want to be persuasive in. And that even then, Gorgias's rhetoric might be more effective than your real knowledge.
Ah. Very well explained. Perhaps Gorgias Problem is the best term, with that summary in the definition.
I don't think so. You already convinced me that an intuitive term that isn't rooted in a particular is better.
Very well.
Good idea. Though perhaps discount rather than discounting. I thought about the term for a little bit, unsure whether you were talking about lower cost or willful downplaying.
The shallowness discount is a big problem in online discussions. Commenters by and large trade in shallowness, and content creators favor it overwhelmingly.
I'm consistently disappointed by good titles about interesting things that turn out to not talk at all about the interesting implications of the title.
Keep up the good work.
Stupid phone. Unsure, not on site. Ugh.
agreed. 'discount' is sharper and better.
"I thought about the term for a little bit, on-site whether you were talking about lower cost or willful downplaying."
I don't fully understand this. can you explain?
"Keep up the good work."
Thank you! means a lot!