3 ms·

My daughter went through this in Canada; same basic concept and she did not like it and neither of us could make sense of a lot of it. My theory is that this v

My daughter went through this in Canada; same basic concept and she did not like it and neither of us could make sense of a lot of it. My theory is that this version of Math is designed by people who hate Math (English teachers). It's all about estimation and fuzzy thinking. Everything that I personally liked about Math as a kid seems to have been stripped away. In my opinion, children who want to become computer programmers, scientists, and engineers are probably not well serviced by this type of teaching. I'm not saying the old method was perfect either; mostly because if you're just teaching algorithms then there isn't much understanding about why they actually work.

candiodari 10y ago
One of the things machine learning has taught me that for many complex algorithms, I don't know how they work, and ... no-one does, and here's the great insight : it's not important, it is in fact a very good thing that I don't know what they're doing. It wouldn't make any difference if I did, and it would in fact be more limiting than not knowing. Having the ability to execute and design them is all that's important. The issue is that machine learning, at least in the modelling domain, is exceeding the maximum complexity the human mind can contain. Attempting to analyze neural networks is fun, and it's what you do as a researcher. But no amount of study is going to tell you what more than 10 neurons are doing to any reasonable amount of detail, no matter how smart you are. Machine learning algorithms suck at explaining what they're doing mostly because what they're doing is far more complex than what we do. And yes, I can build toy examples like a 100 bit binary adder where I do know, I'm talking about non-obvious cases. Therefore: rote is all that matters. You should be able to execute these things, forwards and backwards and take it's code apart and put it back together again. That's all there is to it. And if machine learning does indeed conquer the world, that's all there will be to anything.
- johncolanduoni 10y ago
  That may apply to machine learning, but not to anything you'd get in a grade school math education. Or a graduate one for that matter. Moving from rote implementation of something else came up with to figuring out how to approach problems where the steps aren't already laid out for you is a big speed bump in university education for mathematics majors. Also, unless our current machine learning algorithms just need bigger computers to take over the world, somebody is going to have to develop more. If some kind of guiding intuition isn't a part of that process, the universe will hit heat death before we build an AGI.
gohrt 10y ago
> My theory is that this version of Math is designed by people who hate Math (English teachers). You might think that, but you'd be wrong. "Precision" is explicitly one the 8 principles of Common Core Math. https://en.wikipedia.org/wiki/Common_Core_State_Standards_Initiative#Mathematics_standards https://en.wikipedia.org/wiki/Common_Core_State_Standards_In... > In my opinion, children who want to become computer programmers, scientists, and engineers are probably not well serviced by this type of teaching. Do you have any data or evidence regarding this opinion, or is it just your "estimation and fuzzy thinking"?

candiodari 10y ago

One of the things machine learning has taught me that for many complex algorithms, I don't know how they work, and ... no-one does, and here's the great insight : it's not important, it is in fact a very good thing that I don't know what they're doing. It wouldn't make any difference if I did, and it would in fact be more limiting than not knowing. Having the ability to execute and design them is all that's important. The issue is that machine learning, at least in the modelling domain, is exceeding the maximum complexity the human mind can contain. Attempting to analyze neural networks is fun, and it's what you do as a researcher. But no amount of study is going to tell you what more than 10 neurons are doing to any reasonable amount of detail, no matter how smart you are. Machine learning algorithms suck at explaining what they're doing mostly because what they're doing is far more complex than what we do. And yes, I can build toy examples like a 100 bit binary adder where I do know, I'm talking about non-obvious cases. Therefore: rote is all that matters. You should be able to execute these things, forwards and backwards and take it's code apart and put it back together again. That's all there is to it. And if machine learning does indeed conquer the world, that's all there will be to anything.

johncolanduoni 10y ago
That may apply to machine learning, but not to anything you'd get in a grade school math education. Or a graduate one for that matter. Moving from rote implementation of something else came up with to figuring out how to approach problems where the steps aren't already laid out for you is a big speed bump in university education for mathematics majors. Also, unless our current machine learning algorithms just need bigger computers to take over the world, somebody is going to have to develop more. If some kind of guiding intuition isn't a part of that process, the universe will hit heat death before we build an AGI.

gohrt 10y ago

> My theory is that this version of Math is designed by people who hate Math (English teachers). You might think that, but you'd be wrong. "Precision" is explicitly one the 8 principles of Common Core Math. https://en.wikipedia.org/wiki/Common_Core_State_Standards_Initiative#Mathematics_standards https://en.wikipedia.org/wiki/Common_Core_State_Standards_In... > In my opinion, children who want to become computer programmers, scientists, and engineers are probably not well serviced by this type of teaching. Do you have any data or evidence regarding this opinion, or is it just your "estimation and fuzzy thinking"?