Comment by OhMeadhbh | Hacker News Reader

Comment by OhMeadhbh | Hacker News Reader

Back in the late 80s / early 90s, I taught non-traditional students coming in through state-funded retraining programs at UT Arlington Continuing Education and Texas State Technical Institute (now TSTC) in Waco. The idea was "here are a bunch of laid-off manufacturing workers who want to retrain as SOMETHING with high-tech." The most popular courses were of the very short "here's how to do some things with Unix" variety: how to log-in, how to do simple admin, how to exit vi, how to edit a file with emacs, Make for the perplexed, etc.

I did teach a 6 week C++ course. Mostly attended by CS students at the local college. Back in the day we didn't teach kids how to code in a CS cirriculum. They were expected to just pick it up themselves. Six weeks really isn't enough to teach how to be good at C++.

I spent a fair amount of time reading Papert and Piaget, trying to develop a theory on how people learn and how to best teach "programming" concepts. I've yet to see ANYONE (myself included) do an excellent job teaching the kind of programming employers want. I saw another comment here about how "programming" is simple, but teaching architecture and putting things together is hard. I would mostly agree with that.

But... there are some techniques that are better than others.

My experience w/ Lambda School was they tried to hire me as an instructor way back when. I talked to them briefly about pedagogy and their approach to modeling learning and how they were going to measure improvements in their cirriculum. I got blank looks. They kept harping about how they were going to iterate, but without the slightest idea how they were going to improve between iterations.

I quietly walked out the door never to return (moving to Seattle to work for Amazon instead.) It looks like I made the right call.

> I've yet to see ANYONE (myself included) do an excellent job teaching the kind of programming employers want. I saw another comment here about how "programming" is simple, but teaching architecture and putting things together is hard.

Dartmouth College once had a course called CS23 (Software Engineering) that was infamous for introducing students to "real world" development. The first half of the course was dedicated to C++, with weekly assignments like "implement this data structure class, given the .h file."

The latter half of the course was a 5-person group project with teams chosen by the instructor. The project ran 5 or 6 weeks. And with one week to go, right as the students entered finals, the professors would make a change to the spec. Every time. It was definitely deliberate.

Grades were partly team-based (which sucked if you had a bad team) and partly individual (especially if you had a bad team).

The students who went through this were smart, highly motivated, and had at least 2 or 3 other successful CS classes first. Non-majors almost never took the course.

Time after time, I saw this course turn CS majors into people who could work on real team projects. Not all of them; even in a solid Ivy League CS program some people were a menace to their group. But I'd watch my younger friends go through the experience, starting out naive and coming out with actual instincts on how to ship something.

In a lot of ways, I don't think CS 23 was always well taught. Our professor was a Lisp programmer teaching C++ and his .h files often contained serious errors. But the genius of the course design was that slightly dodgy teaching only added to the overall experience. The course was intended to be a high stakes team project under realistic conditions. Bad team assignments, bugs in the spec, a changing spec, the whole nine yards. And the students were as prepared and qualified as any students could be, going in.

But I think there's a kernel of a good idea here. Once students can program, the best way turn them into developers is to make them build and ship something as a team.

(Dartmouth also had solid theory classes in the major, but that's a story for another day. You couldn't graduate without having done a real project and having written a lot of mathematical proofs.)

> Once students can program, the best way turn them into developers is to make them build and ship something as a team.

I had a semester project like that, in a team of 7, though we didn't had the changing spec... But it was a really good learning experience

It is incredibly hard to find ANYONE at any of these "coding bootcamps" or online education platforms that has any idea about what pedagogy is (or that it even exists) to the point that I seriously think anyone that makes it out of these programs and lands a job was mostly because they got the focus and time to learn by themselves (due to being in the program) than due to the program itself.

They just needed a kick in the butt. Anyone that actually needs to learn how to program for real, won't make it out.

I joined a bootcamp once, already knowing most of the material covered. I wasn't the only one in my cohort (of 4 people) who already knew how to program.

My hope was that the bootcamp might make it easier to find a job, which didn't happen. They put more effort into instruction than into placing applicants. Not sure why; I can easily cover any material that would be covered in a bootcamp myself.

> the bootcamp might make it easier to find a job

Did a few stints at a couple boot camps as an instructor. Many other students thoughts that too because it’s half the boot camps’ marketing efforts.

But it was a big lie both times. They are friendly with a few startups who are looking for cheap labor at terrible conditions and they only have a handful of positions. « Career services » in the worst boot camp was an intern that changed every six months because you can’t have a longer internship and then they’d have to pay that person an actual salary. Swap for a new free intern and voila you can market your « dedicated career and placement service for graduates only »…

A despicable industry.

Those who actually make it are rare and had grit, luck, and sometimes previous knowledge. Most graduates stay unemployed a very long time, or underemployed if they’re lucky.

Don’t go to a boot camp. Use the tuition to sustain yourself while you’re learning using free resources. The ROI is much better.

If you insist on attending a boot camp, find graduates and ex instructors on LinkedIn first and talk to them to get a sense of what it’s like. Don’t just throw thousands at a shady organization without doing your own research.

> Don’t go to a boot camp. Use the tuition to sustain yourself while you’re learning using free resources. The ROI is much better.

Well, there was no tuition. Much like other bootcamps, they offered an ISA. And since the only value I expected them to provide was job placement, that seemed to make sense. So I received nothing of value from them and paid nothing for it.

But this leaves the question open of why they thought this made sense. They had to spend money providing the bootcamp to me. What were they hoping to get out of that?

Never wanted to do a bootcamp because of all the red flags around ISAs, and the often-remarked fact that there are almost no VCs or CTO's who understand pedagogy and the difference between that and growth hacking.

The thing that has been the most helpful to me to figure out what I should be learning and how my self-study time should be directed was a memo I got from a friend who worked at Google, explaining how their interview process works and what they expect SWEs to know before they can be hired (this was before that information was widely known outside the company; since then there have been some good write ups from former employees).

Could you say more, or even link, to proper methodologies for measuring iterations of your curriculum?

I'm a very amateur teacher, but have received great feedback on my "teaching style",

which I've refined based on two things: how often I get puzzled looks during a technique, how often I get thanked for using a specific technique.

I'm curious what this looks like in the big leagues - how this "scales".

Thanks for your time.

Hi DANmode, I'm obviously not OP, but I am a teacher who has gained a broad understanding of the field from practice and research.

One thing about feedback this is counterintuitive is that, students are generally very bad at understanding what actually helps them learn. When I say learn I mean gain a conceptual understanding and procedural fluency in whatever they are learning that sticks with an ability to appropriately transfer that knowledge beyond the context they've seen it in.

So unfortunately when a student says "that was great!" research has found that to be negatively correlated with learning.

If you want to delve into a study that talks about this and what effective teaching looks like in the long-term I really recommend this study done at the US Air Force Academy [0].

It is the closest thing I have seen to a gold standard study in education where a lot of research is dubious at best:

Look at what Carrell and West handed us with this study:

- 7 year study at the US Air Force Academy - 10,534 students - 421 faculty members - 30 core courses, all standardized (math, science, social science, humanities, and engineering) - Random assignment of students to professors in initial course and follow-on courses

[0] https://www.jstor.org/stable/10.1086/653808

I've got an email in my profile if you want to discuss!

The study set up is a bit unusual.

Consider three ways to teach:

A) Do what's best for the student's long term learning. ("deep learning".)

B) Optimize for the student doing well on a specific type of test (e.g. AP Calculus AB). ("teach to the test")

C) Optimize for the student doing well on a specific set of questions known in advance.

In the study, instructors had all three options.

In most high stakes testing, instructors have only options A and B.

The study seems to conflate B and C.

But A and B are more similar, than are B and C.

It would have been an interesting study if the instructors hadn't had access to the exams beforehand.

(Although typical US university high stakes exams are designed by course instructors who can do (C), I don't see this as relevant, as instructors can more easily help their students do well by making the tests easier.)

> Make for the perplexed

Chefs-kiss

That’s me every time I start a new project ;). Thanks for the chuckle, and please publish a book having that title.

Andy Oram's Managing Projects with Make is my goto reference, but there's a market for a make cookbook, I bet.

Reply on news.ycombinator.com