> If you pull lots of kids from a regular class, they're also going to benefit from a narrower spread of abilities.
I think you’re talking in theory and not practice.
In reality, one of two things (or both) happens.
1. They pull from the “slow” class. Different places call it different things, and it’s not special ed, but it’s distinctly low levels of education. The “slow” class doesn’t benefit due to it basically serving the role of child care rather than education.
2. The “regular” class is effectively a slow class, so same as above, there is very little educating happening, just child care.
Ah, but the people getting moved into the slowest actual education benefit because they go from no education to some education, right?
No again. The modal outcome is that these threshold folks are given passing grades while learning very little, all while making the experience frustrating for the student and the teacher.
> You could fix this by training teachers better on how to actually educate slower kids effectively (direct instructional methods work very well there) but that approach is not popular either because it's seen as trivializing the teacher's job
I have seen these teachers and its potential effectiveness, but only in special ed. The biggest pay offs seem to be in elementary school with basic/fundamental literacy and numeracy.
As grade level increases, the impact of this instruction decreases substantially in terms of impressiveness largely due to limited scope.
> The more meaningful argument against this kind of fine-grained tracking is that it creates pathological incentives among teachers.
I agree with this, but not for the reasons you give.
> Every teacher is going to want to teach the "easy" classes with more skilled kids, so the bottom level of students gets stuck with very low-quality teachers and they don't get anything near a fair chance to improve - they're basically stuck at that level.
First, classes with highly skilled students are not necessarily easier. They are only easier if the teacher makes few changes to the curriculum to adjust to the class (basically lack of time or teacher laziness). Gifted classes in particular can be challenging for teachers who think that they want to teach the smartest kids because often some of the students will know more (sometimes a lot more) than the teacher about the subject. This can create very awkward moments (note that this phenomenon can also be seen at universities with students and adjuncts, sometimes even at elite schools).
Second — this is another “secret” that I think covid made less of a secret — the main function of most lower level classes beyond basic literacy and numeracy in elementary school is state-supported child care.
In reality, especially in high school, there is very little learning going on in the lower end classes. The students don’t care about the content, the students’ parents don’t care about it, and their peer groups don’t care about it. The only people who care are the teachers and admins due to standards states have set and funding tied to testing to those standards. There is essentially no product-market fit to use terms that most HNers might understand. The important thing is that the kids have a place to go while the parents work. These are terrible classes to teach.
The actual way to fix this, imho, is to meet the students and their families where they are at. Specifically, introduce them to skill sets that they may actually find useful and/or interesting. It would be ideal to pair this up with co-op working opportunities. Note that this system largely exists in Japan (where they do tracking), so this is not all just theoretical conjecture.
There are a few problems with this system.
One issue is that this type of education can be difficult to create and maintain, because the content could vary widely from school to school — a school in rural Iowa would probably focus on different skills than a school in Brooklyn. This also would make it difficult to measure, so only the local community would really know if it was working (that’s ok, imho).
Another issue is that this type of education done properly is rife with socioeconomic and race issues. For example, if you are teaching a student sales in a coop environment where they actually do sales, it’s probably prudent to teach them something like register shifting so that they can adjust their language to their audience. I will leave it to the reader to figure out why that is a minefield in the US (but probably shouldn’t be).
Apologies for the wall of text, but I think that this is an interesting topic that is poorly understood by many/most highly educated Americans. This is especially true of a lot of HNers who seem to have largely experienced schooling from a middle/upper-middle class perspective (specifically, a largely college-bound school/community).