Learning to Swim By Reading a Book - Teaching CS in Rwanda

http://blog.nyaruka.com/learning-to-swim-by-reading-a-book

by nicpottier • 15 years ago

39 28 15 years ago

blog.nyaruka.com

MarkPNeyer2• 15 years ago

"My advice to anybody in Rwanda who wants to learn Computer Science is this: Don't attend University"

Honestly, this advice applies to anyone anywhere.

tomjen3• 15 years ago

Have you attended a university? Did you study computer science there?

If not, how can you make a comment like that.

MarkPNeyer2• 15 years ago

i earned a bs in computer science from a liberals arts college (Xavier) in 2007, and an MS in computer science from the University of North Carolina in 2009. I am thoroughly convinced i'd be much farther along in my career now if i'd never gone to college.

thinkign about this has brought back a lot of memories, so i apologize for the long post.. maybe i'll turn this into a blog post.

at the end of high school (i graduated in 2003), i spent a lot of time hanging out with defunkt (github co-founder and ceo). we dreamed of making our own games; "you be john carmack and i'll be john romero," he told me once. summer 2003, between halo (and later planetside) LAN parties with our circle of friends, i started learning directx. our plans were probably too ambitious (3d multiplayer action game) for a group of recent high school graduates with no work experience, limited knowledge of c++, and no art skills. i got to the point where i had a .x mesh of a tiger moving around a hightfield (with shading!), a simple network chat protocol and console, but when our artist left for college, i had no real hope of finishing the game.

towards the end of that summer, we started a 'consulting firm' which never really went anywhere. when school started i spent most of my time working on bullshit homework assignemnts and working as a bus boy for spending money. i was also majoring in physics and math, so i didn't have much free time. what little time i did have, though, i spent programming. defunkt showed me php and apache, and i learned far more trying in the 10 hours a week i spent programming a 'schedule the bus boys' web app than i did in the 30 hours a week i spent going to class and doing homework.

defunkt spent most of high school fucking around and having fun, whereas i spent it studying my ass off and taking AP classes to get more college credits - i started undergrad with something like 50 credit hours. defunkt didnt' get into the college he was hoping to (miami univeristy of ohio), so he started at a community college instead. he dropped out in spring 2004 when he realized how worthless it was for him. i knew he'd be wildly succesful and cursed myself for not having the courage to do what he did.

he then spent all his time trying to find programming work, while i spent all my time busing tables and doing all the stuff i had to do for the english and theology classes i was taking becuase i chose to go to liberal arts college in cincinnati, ohio instead of an engineering program at a top-rated school. i had a girlfriend at the time, whom i'd been dating since i was a freshman in high school, and i thought i was chosing love over money. :-/

defunkt found some work doing web development for a trucking company in new jersey (i'll always admire his hustle) and in spring 2005 he parlayed that experience into an offer to work for cnet in san francisco. he tried to convince me to move with him when he got the offer from cnet. at the time, i told him something like 'i know you're going to be incredibly succesfull but i dont have the guts to quiet college.' to be honest i really watned to go, but i was still in that (now very unhealthy) relationship at the time, and didn't have the courage or insight necessary to break it off.

fall 2005 was the worst period of my life. my friend was apparently doing quite well for himself in california, and i was bored and miserable in college, stuck in a relatinship i thought i had no way out of. i had had tried unsuccessfully to get programming internships, but hat no luck becuase xavier wasn't known for its cs program, and the best i could do was an IT job at a medical device manufacturer. i spent my days fixing corrupted outlook installations and installing replacement hard drives. all the while i was coding like mad on the weekends, on toy projects (a risk map generator was one of my favorites) and simple games. again, i learned more about good design and code layout working on those projects than i did in class, where i was now ostensibly learning about "software engineering".

the only thing i remember from the software engineering class i took was when we were pretending to come up with 'user stories' for an electronic voting machine. my friend jacob and i were supposed to be the customers, and sharon and kyle were supposed to be the developers. jacob and i said that the machine should be a regular dodecahedron that spins on its axis 'like those sunglass racks at the malls', and it should have eye recongition, a finger print scanner, and stool sample acceptor to identify voters. it's still funny in retrospect, and i guess it did teach us that customers are fucking morons, but i could have told you that without the class.

fall 2005, i was diagnosed with depression and then bipolar disorder, and put on a number of different medications. i tried to kill myself a few times and went to a mental hopistal. my life seemed to be going nowhere. i figured i'd be stuck in that relatniship the rest of my life, doomed to spend eterntiy fixing broken hard drives in cincinnati.

finally, in early 2006, i worked up the courage to break up with my girlfriend. live got a hell of a lot better.

damn this is long.. to the point!

i learned a lot of aweosme stuff in school, don't get me wrong. i can show you how to convert a deterministic finite state machine to a regular expression, and construct a NFA for ww' like nobody's buisness. i spent summer 2006 working for the NSF on approximation algorithms, so i learned a lot about that kind of thing. and there were some things i learned in school that have been really useful - we wrote an operating system my senior year, and it was a hell of a lot of work. i learned a ton about debugging and machine code.

at the end of the day though, i look at where defunkt is now, ceo of a wildly succsessful start that's basically a household name among hackers, and i'm still a nobody.

i woke up one morning when i was 14 and decided i'd better start kicking ass in school if i wanted to succesed in life. if i could go back to 14 year old me, i'd say forget abotu school. you can learn a hell of a lot faster than any teacher can teach you. learn to code by coding nonstop, and get experiecne working for other people as soon as you can. the stuff you learn in school is valuable, true, but it's not worth the opportunity cost. i had a full ride to undergrad and still feel like it was a net negative financial decision.

arethuza• 15 years ago

Give that the article recommends the constructive and practical step of downloading and working through an MIT course I think it is clear what was meant was that while Rwandan universities aren't very good at teaching CS there are plenty universities who do a splendid job.

Unless, of course, you are really claiming that all CS courses everywhere are pointless?

irahul2• 15 years ago

> Honestly, this advice applies to anyone anywhere.

A good university teaches you things in a systematic manner, and teaches you things which you won't learn by yourself. It's not that you can't - most likely you won't. Universities like MIT and Stanford have been around for a long time, and a lot of thought has been given to designing courses.

When you are in a university, your whole schedule is dedicated to learning, which generally isn't the case while learning on your own.

There are many problems while learning CS which you will need help with. There are qualified peers and teachers available who you can go to.

Universities make you learn unpleasant things - things which you won't be learning on your own. Most of the subjects require a basic investment before you learn to truly appreciate it(it's true for non cs as well - music, sports, painting etc). When you are on your own, you might very well become discouraged and not pursue it.

Most of the self taught programmers don't understand lexing, parsing, advanced datastrucutres(think red-black tree, b-tree, tries, splay list), various locking protocols, memory management etc. The reason for it is it doesn't take much to cruise through your day job, and they don't see why they have to learn this stuff. But many jobs do demand your command over these topics, and top paid programmers are proficient in these, setting a precedent.

snikolov• 15 years ago

I think I see what you're trying to say --- practical experience beats theoretical knowledge in most work you'll do in industry --- and I agree. But as someone who did an EECS program that didn't require a lot of coding, I've found that a lot of the practical specifics are easy to pick up when they readily fit within the theoretical frameworks that I've learned.

patio11• 15 years ago

That's a good FizzBuzz problem. Broadening the scope from Rwandan CS students, many well-compensated professional programmers in rich nations could not complete it successfully. I know, this is crazy, but we live in a crazy world.

And to those HNers who think that inability to FizzBuzz is a funny story made up by overly cynical neck beards: spend a year hiring.

nicpottier5• 15 years ago

Ya, it is a similar thing for sure. The problem is just accentuated here. Two of the four we interviewed had some clue as to get started. The others were pretty stuck right off the bat. And that's (approximately) the best out of 50.

That they don't code at all in school is just inexcusable.

enry_straker3• 15 years ago

I suspect that the sample size is too small to really draw any reliable conclusion.

Having said that, you are on the ground. You see and talk to them. You definitely have a good idea of what you are talking about. Me, i'm just a guy half way around the world feeling pretty bad about it - and hoping that this is not true.

You touched on so many points that i endorse and recommend heartily. Programming is coding. It's hands-on. People have to sit in front of a machine from day one, hour one of their learning. People learn from writing crappy code, and debugging crappy code till the code starts to work. People learn from making their own mistakes. People should learn that programming cannot be memorized and repeated ad nauseum on exams.

I remember the kind of folks i interviewed when i was starting a company in india in 1995. I was horrified at the nature of rote memorization in many schools and collages. I made a decision not to hire any one with a CS degree as i felt that i would have to spend too much time making them unlearn bad habits. I then decided to go out, hire students with no programming background, teach them coding from the ground up. Teach them to solve problems one step at a time. Teach them the joy of debugging. And they turned out to be great.

I suspect that every country, region, city, place might have to go through an evolution of programming culture, but the internet and it's legions of programmers gives a great way for beginners to short-circuit their learning process.

Let us know of ways we can help, as well.

ido• 15 years ago

Hrm, maybe I was just lucky but when I was looking for a student intern earlier this year I got several dozen replies, and even tho many of these were pretty poor fits I still ended up having to choose between several good candidates.

The one who ended up doing the internship was top notch (better than I expected a 20 year old student to be) & I'm pretty sure at least 1 or 2 others I filtered out (only had 1 position to fill) would have done a good job as well.

irahul2• 15 years ago

You sure can't learn CS by only reading a book - the keyword here is only.

Following through good books is an excellent way to learn programming. I don't believe anyone who has followed through K&R would have any difficulty getting through these fizzbuzz questions.

The analogy with swimming is flawed. You don't need to read books to learn swimming(or maybe there is some theory involved - I don't know), but books are a vital part of a CS education. Good books teach you a lot more than what you learn by randomly typing.

An hour in the library saves a week in the lab.

nicpottier5• 15 years ago

Hrmm.. Maybe, maybe not. I'm certainly not making the argument that you shouldn't read, if anything, my argument there is don't attend university if your professors have no idea what they are teaching, which is the case here.

Obviously, those who write books usually have a pretty good idea of what they are talking about, so gaining knowledge from them is a great strategy.

And yes, there is a belief here in Rwanda that you can learn CS by ONLY reading books.

But to your point, I'm not sure why you rank books so highly, when in isolation they clearly aren't enough. The average student probably won't grock worse case runtimes by only reading about them. Having a lecturer explain it and give examples will help, but actually implementing some sorts and seeing how fast they run on different size data sets is really what is going to drive it home.

But for a great many programming jobs, I would actually make the argument that you COULD do them without reading a single book. Granted you probably won't be building the next map-reduce, but you can still be effective.

Learning how to find solutions online, read documentation (does that count as a book?), and debug, as well as fundamental algorithms are the big skills.

enry_straker3• 15 years ago

I beg to differ.

When one first starts out, it's ok to just sit in front of an interpreter or a compiler and start trying out small programs, make mistakes, learn from them and go on.

But one vital quality of good developers is being able to read code, understand good code, borrow good coding habits and using them in your work. Many excellent books are available where what shines through the pages is the author's experience in writing good code, explaining the logic of how he or she arrived at it, and then solving small problems which might come as exercies at the end.

University degrees, learning on the job, etc will only get you so far ahead, but to leap ahead in your productivity, it's best to learn from the real giants in the field.

Programming is a field that constantly changes. Reading, Applying, and learning new skills, and new techniques is part of the skill-set needed to survive. Great books are a complement to hands-on experimentation in this regard and cannot be over estimated as more and more learning is often self-learning since it's difficult to always find good mentors.

viblo• 15 years ago

I think its quite interesting that the example chosen to display how bad theory without practice can be is swimming (which is an excellent example). Im currently working in the Philippines with setting up an development team and here some of my local co-workers have had theoretic swimming in school. They got a grade, but cannot actually swim! Got a bit surprised when I heard it for the first time.

toumhi• 15 years ago

I was there a few months ago. While tourists were swimming, Filipinos were all floating with life jackets.

It surprised me that, in a country that is basically a cluster of small islands and where you can only get around by boat, nobody can swim.

Fargren2• 15 years ago

Going on a tangent here.

While programming is certainly the most marketable skill of a computer scientist, computer science is not about programming. Computer science is the mathematical study of computation; it's the study of processes and of information. It is possible to be a great computer scientist without knowing how to program.

This doesn't excuse the university of Rwanda: programming is possibly the best tool we have to teach CS these days, and it is a skill that most computer scientists should grok. But I do think a CS degree's value is not only measured by how good a programmer it means you are: CS is a scientific field, not engineering.

mckoss• 15 years ago

Lacking tangible skills, it's hard to evaluate whether someone is a "great computer scientist".

Can you be a "great artist" if you can't even draw? How about a "great architect" if you don't know how to build a house?

Possessing the basic skills in your field of study, I think is a prerequisite for acquiring the advanced skills and the ability to contribute to the state if the art.

Fargren2• 15 years ago

There are other tangible skills a computer scientist should have, other than programming. The ability to demostrate graph properties, solve computability and complexity problems, explain and develop algorithms are all things that don't require programming at all. Some of my best teachers at college are not great because of their programming skills, but because they are great at other areas of CS.

enry_straker3• 15 years ago

Though i am inclined to agree with you i do think this distinction might be appreciated in the context. What i mean is, when a person is starving he doesn't really care if his food is from a 2 hat or a 3 hat restaurant.

Having said that, i view programming as a tool which is to be used for solving problems. But i would expect a CS major who wants to work in the industry to be really proficient with his or her tools, and that includes programming.

A CS Researcher might be able to get away without any knowledge of programming in a university setting, but could be out of place in the real world unless he focuses on a speciality which is in high-demand.

I once had the experience of sub-contracting work to one of the best universities in my country and found that both the students and the professors involved with the project were not familiar with coding. Thankfully i could train them on coding with minimal effort since they had a background in rigorous mathematics.

zokier• 15 years ago

I think the major problem here is that CS and programming are too intermixed in education (and thus in the industry too). Uni degrees (imho) should educate scientists who study their field and do research.

If we continue on the swimming analogy: You don't actually need to know how to swim to research swimming. You just need to know how to gather data (by studying swimmers), and how to analyze that data to draw some conclusions.

I think the mixup is reflected in the comments here on HN too, people going to study CS in hopes to learn to program, with no intention in becoming a scientist. And then they are disappointed when the CS curriculum is trying to train them to be scientists.

imho the whole computer related education needs a complete re-engineering, from primary schools to universities. We need to assess what kind of people we need (scientists, engineers, mechanics, professionals, common people, etc?), and what skills they need to be efficient at their future jobs. I think it's quite clear that the concept of having CS and throwing everything computer-related there is not working very well atm.

saturn• 15 years ago

Well, I can never resist implementing these interview questions! Here's my answer:

  def strip_dupes string
    res = ''
    string.each_char do |char|
      next if char == res[-1]
      res << char
    end
    res
  end

Any improvements?

update: and can never resist reducing them to one-liners either.

  def strip_dupes str
    str.each_char.inject('') {|r,c| r << c unless c == r[-1]; r}
  end

nicpottier5• 15 years ago

You're hired. Though we are mostly a Python shop, I'm sure you'll adapt. Sorry, no relocation package though. :)

candeira• 15 years ago

Yeah, I couldn't resist it either. Mine:

def de_n_plicate(word): return ''.join(c for i, c in enumerate(word) if i == 0 or c != word[i-1])

Learning to Swim By Reading a Book - Teaching CS in Rwanda

Related Stories