4 ms·

Thank you for your thorough reply. Colleges are under fire, even from their own ranks, by people whom I believe confuse training with education, but it's clear

by dhimes 9y ago

Thank you for your thorough reply. Colleges are under fire, even from their own ranks, by people whom I believe confuse training with education, but it's clear that's not your issue. It sounds to me like there needs to be some sort of deep-dive "onboarding" program where new hires can work on a curriculum of projects and learn the SOPs of the organization. Colleges could take some of it on, of course (testability and maintainability for example), but one of the complaints I heard even ten years ago is that they can't keep up with the changes in the field. No true fundamental best-practice principles have evolved, it's largely company-dependant.

bpicolo 9y ago
It would be interesting to see a course than spanned multiple years building upon the same project with the same team, emphasizing different aspects year-by-year, but I imagine that would be pretty nightmarish on the scheduling front.
- dhimes 9y ago
  It's also a problem of how much you can ask a student to do. There was a time when the expected time to complete a "4-year degree" was approaching 6 years in the engineering-ish fields, and the tuition-check-writers got pissed and called their state legislators to put a stop to it. So adding some sort of multi-semester capstone project would have to be woven into what is already in place, but I would imagine it wouldn't entirely fit into the curriculum without increasing the student level-of-effort (meaning, all of the stuff being taught would still have to be taught, but in addition the extra bits special to the scale of the project would add to the overall workload).
  - AnimalMuppet 9y ago
    It wouldn't have to be that big. Even if part of one semester was to take a month as a class and try to add some features and fix some bugs to a codebase that the previous decade of classes had worked on, that would be an eye-opener to most students.
alistairSH 9y ago
It sounds to me like there needs to be some sort of deep-dive "onboarding" program where new hires can work on a curriculum of projects and learn the SOPs of the organization. My employer does exactly this, both within the R&D organization and within our services/consulting group. All new hires from college do a 3-4 week "boot camp" where they do all the common indoctrination stuff, from HR paperwork, to learning the shared tools, to a mini programming project. Expecting a college graduate to show up ready to contribute like a 5 year veteran is ridiculous. As the parent message says, college is mostly for education, not training. Internships and co-ops fill some of the gaps, but high quality internships are few and far between.
- AnimalMuppet 9y ago
  I know of a company that hires fresh graduates, and doesn't expect them to really be able to contribute for two years. They're in Indianapolis, though, so they may have considerably fewer problems with their employees getting poached by others before they can contribute enough to pay back the training period.
AnimalMuppet 9y ago
It seems to me that, out of ten graduates from a CS program, only one is going to go on to do academic computer science; the other nine are going to become computer programmers. CS programs are doing a poor job of preparing those nine. Now, in other areas, we have a distinction. We have physics departments that teach people theory, and we have separate engineering departments that prepare people for careers putting the theory to useful work. Well, where does the CS program live? At least where and when I went to college, the CS program was part of the Engineering department. So I think it's fair to say that colleges should take on considerably more of the job of preparing software engineers for real-world careers in software engineering. Hiding behind "we teach CS, not software engineering" is a cop-out, especially if CS is within the College of Engineering.
- TheOtherHobbes 9y ago
  There are still different levels of expertise and talent in engineering, broadly dependent on mathematical ability and skill for abstraction. This is maybe more obvious in hardware design. At the top of the tree you have people like Maxwell, Heaviside, and Shannon, who invent entirely new possibilities out of pure math. At the other extreme you have technicians who don't truly understand math or theory, but can build a circuit that will probably work if handed a cookbook. In the middle are people who can work with abstractions like DSP and filter design as long as the ground has been broken for them. They understand enough math to find their own way through a problem from first principles, but aren't creative enough to invent anything truly original. CS is more amorphous, the levels are maybe harder to separate, and it's cursed by not having a truly consistent practical mathematical foundation analogous to the applied physics that underlies engineering of all kinds. But IMO there are similar skill levels - although at the higher levels, abstraction can become a bad thing rather than a good one. The problem is that although there's math in CS, after Church/Turing - which is pretty elementary compared to most physics - there isn't anything that passes for a formal theory of problem/solution specification and computation. Without that objective basis, a lot of CS is academic opinion instantiated in a compiler. And a lot practical commercial CS is a mountain of hacks built on language and tooling traditions with no empirical foundation. Commercially, the most productive people will be middle-rankers - not so clever they'll be thinking of new applications for category theory in an iPhone app, but clever enough to be able to think beyond cookbook development.

by dhimes 9y ago

bpicolo 9y ago

It would be interesting to see a course than spanned multiple years building upon the same project with the same team, emphasizing different aspects year-by-year, but I imagine that would be pretty nightmarish on the scheduling front.

dhimes 9y ago
It's also a problem of how much you can ask a student to do. There was a time when the expected time to complete a "4-year degree" was approaching 6 years in the engineering-ish fields, and the tuition-check-writers got pissed and called their state legislators to put a stop to it. So adding some sort of multi-semester capstone project would have to be woven into what is already in place, but I would imagine it wouldn't entirely fit into the curriculum without increasing the student level-of-effort (meaning, all of the stuff being taught would still have to be taught, but in addition the extra bits special to the scale of the project would add to the overall workload).
- AnimalMuppet 9y ago
  It wouldn't have to be that big. Even if part of one semester was to take a month as a class and try to add some features and fix some bugs to a codebase that the previous decade of classes had worked on, that would be an eye-opener to most students.

alistairSH 9y ago

It sounds to me like there needs to be some sort of deep-dive "onboarding" program where new hires can work on a curriculum of projects and learn the SOPs of the organization. My employer does exactly this, both within the R&D organization and within our services/consulting group. All new hires from college do a 3-4 week "boot camp" where they do all the common indoctrination stuff, from HR paperwork, to learning the shared tools, to a mini programming project. Expecting a college graduate to show up ready to contribute like a 5 year veteran is ridiculous. As the parent message says, college is mostly for education, not training. Internships and co-ops fill some of the gaps, but high quality internships are few and far between.

AnimalMuppet 9y ago
I know of a company that hires fresh graduates, and doesn't expect them to really be able to contribute for two years. They're in Indianapolis, though, so they may have considerably fewer problems with their employees getting poached by others before they can contribute enough to pay back the training period.

AnimalMuppet 9y ago

It seems to me that, out of ten graduates from a CS program, only one is going to go on to do academic computer science; the other nine are going to become computer programmers. CS programs are doing a poor job of preparing those nine. Now, in other areas, we have a distinction. We have physics departments that teach people theory, and we have separate engineering departments that prepare people for careers putting the theory to useful work. Well, where does the CS program live? At least where and when I went to college, the CS program was part of the Engineering department. So I think it's fair to say that colleges should take on considerably more of the job of preparing software engineers for real-world careers in software engineering. Hiding behind "we teach CS, not software engineering" is a cop-out, especially if CS is within the College of Engineering.

TheOtherHobbes 9y ago
There are still different levels of expertise and talent in engineering, broadly dependent on mathematical ability and skill for abstraction. This is maybe more obvious in hardware design. At the top of the tree you have people like Maxwell, Heaviside, and Shannon, who invent entirely new possibilities out of pure math. At the other extreme you have technicians who don't truly understand math or theory, but can build a circuit that will probably work if handed a cookbook. In the middle are people who can work with abstractions like DSP and filter design as long as the ground has been broken for them. They understand enough math to find their own way through a problem from first principles, but aren't creative enough to invent anything truly original. CS is more amorphous, the levels are maybe harder to separate, and it's cursed by not having a truly consistent practical mathematical foundation analogous to the applied physics that underlies engineering of all kinds. But IMO there are similar skill levels - although at the higher levels, abstraction can become a bad thing rather than a good one. The problem is that although there's math in CS, after Church/Turing - which is pretty elementary compared to most physics - there isn't anything that passes for a formal theory of problem/solution specification and computation. Without that objective basis, a lot of CS is academic opinion instantiated in a compiler. And a lot practical commercial CS is a mountain of hacks built on language and tooling traditions with no empirical foundation. Commercially, the most productive people will be middle-rankers - not so clever they'll be thinking of new applications for category theory in an iPhone app, but clever enough to be able to think beyond cookbook development.