Hacker News Reader

Comment by alecco | Hacker News Reader

Dropped out out of frustration with ridiculous emphasis on calculus and extremely poor algorithms teachers. I don't regret dropping out, I regret wasting a couple of years fighting that.

But almost 15 years later I now find algorithms teaching in top universities completely out of touch with reality. Hardware changed drastically in the last decade and it seems they just gave up. It's turning into a cult of big-O theory with even tournaments with problem sets specifically picked to validate the books. Don't you dare point them to real algorithms, you'll get booed in a sec. This becomes worse in places like Google or fields like AI.

It's very sad, I'd love to see somebody taking the torch and teach concepts based on reality (measuring). Like they did back in the 60s. Mainstream sucks nowadays.

View on HN

You dropped out of college because of a "ridiculous emphasis on calculus" yet you are able to track and form opinions on the quality of "algorithms teaching in top universities"? I doubt it.

I find your lack of faith disturbing.

Agreed. Calculus as a bridge to Real Analysis is a fundamental part of a real math curriculum.

Thanks for the ad hominem.

I like calculus and math in general. But undergrad CS emphasis should be information theory, algorithms and discrete mathematics. They never used calculus to analyze algorithms, not even once. I've only seen it used on books (Knuth or Rivest et al.) In the majority of universities CS is still dominated by the same teaching body from math or engineering.

Since now we have videos of classes at top universities, there's no benefit of the doubt anymore.

I admit my comment was harsh, but not ad hominem. It was a statement of fact: given your background as you laid it out, I would not trust your judgment about the quality of algorithms instruction in any university. If you offered me an appraisal of how well some university is teaching algorithms, I would seek a second opinion.

Calculus and other basic math is just part of general numeracy and a university that failed to teach it, even emphasize it, is a failure, or a trade school.

I really respect your decision to drop out. I am a senior and I decided to take my last semester off and do a long internship. I am lucky to have found a good company, but that fight you had is still going on in my head. Most of the problems I solve at work are new to me, and nothing from my university is relevant, including the algorithms class. For instance, I have never used git in school. In fact, nobody has even mentioned it. Just think about it for a moment. How important is something like git or its equivalents, as compared to outdated theory bound knowledge?

Sounds like you guys should have gone to technical school or majored in CIS or something. Learning how to use version control is not in the scope of a Computer Science curriculum.

One time I showed my 1st commercial web system to a teacher and he was so quick to say 'but this is not CS' it was sad.

I never said it was supposed to be CS, I just showed the system... His response had nothing to do with anything... His words were just a quick self-defense he seemed very used to use, in a pretty destructive way.

It took me some time to know University was doing no good to me, but now I'm past detox and I'm glad I dropped out.

I don't care if 'it is not CS', I learn whatever it takes to build awesome stuff.

It would not kill anyone to offer one class, just 3 credits out of the 120 in an undergraduate career, about the practice of creating software. FWIW even my CS minor (30 credits) had such a course.

The course was structured around a single group project. We were assigned to a "customer" in the department, and we were responsible for soliciting (and formalizing) requirements, developing a design, implementing it, and testing it. Revision control was not optional, and we were taught how to use it. Our grade was based upon how well we completed the task, how well we presented ourselves to the instructor and "customer", and how we were assessed by our fellow teammates.

No one suffered any loss to their theoretical background, the universe didn't explode, and everyone left the class a better programmer than before. Since a CS degree is still, rightly or wrongly, considered the sole requirement for many programming jobs, I think it was time well spent.

Actually it should. Else those team projects would not make sense. I was asked to work in teams repeatedly, and I see this recurring in a lot of classes I stumble upon on the web. Try it yourself, pick a popular computer science class that gives the students the ability to work in teams and try to find a reference to version control.

Git was built using said "outdated theory bound knowledge".

This. There is a difference between learning to use today's cutting edge technology, and learning the principles that will enable you to build tomorrow's.

Man, dropped out because of calculus? Calculus is awesome! It powers most of the physics that are of interest to the lay person.

To each their own, I guess.

One of the reasons I dropped out was also that the people there were out of touch how things worked "in the real world". It was just the same curriculum as it had always been (minus the things that were cut to appeal to more students who otherwise couldn't cut the mustard).

I was legitimately afraid that I would waste my time on things that would largely be irrelevant outside academia in the sense that I had to factor in a couple of years to learn how you realistically work and create something of value.

Algorithms are such a goddamn exciting subject, and it's a shame when someone manages to make it boring or unintelligible. I'd love to have studied CS at a good American university, but with visa rules and the slim chances of admission to a good college, I guess I have to find my own way - even though getting a work visa is probably going to be semi-impossible.

So it goes.

It's worse. The joke "MongoDB is web-scale" reminds me of this growing cargo-cult full of cocky CS graduates.

Edit: thanks for the anonymous downvotes validating my point; dissent must be suppressed.

I'm guessing CS dropouts are just as vocal in support of the nosql of the moment.

Let's assume for a moment the majority of nosql supporters are CS dropouts. That doesn't say the majority of CS dropouts are supporters of nosql. You are falling for the base rate fallacy. That's exactly the kind of highly relevant concepts I've never seen taught for CS. Thanks.

While we are at it, probability theory should have a much stronger emphasis than calculus for CS. It's only seen on graduate CS on most universities. In my uni, we had 2.5 years of calculus and a miserable .5yr stats/probability course, wich was mostly stats and it was one of the less emphasized subjects. Why is it like that? I bet because engineering (and mainstream math) focus is somewhere else.

whats the definition of web-scale?

It's a vague term, my guess is they mean datasets like Google's.

Do explain how and why so many CS graduates hold on to cargo-cult notions?

Rather than reflexively attacking your viewpoint, I'm instead curious to see if I can understand it.

I'm most curious about what you mean when you say "real algorithm". Can you offer an example?

Reply on news.ycombinator.com