Autodidacticism
raganwald.posterous.com
raganwald.posterous.com
To claim so would be to discredit my friend who explained pointers to me in high school, or my first programming boss who helped me transition from writing HTML to SQL.
While I didn't take a class from Hennessy or Patterson, Gosling, the GoF, McConnel they certainly 'taught' me from their books, as did the countless anonymous internet posters in #java, #c, and as of late, #scala.
The feedback I received from friends who reviewed side projects or the JBoss guys reviewing my patches was invaluable.
It may be fair to say I created my own computer science curriculum, or that I wasn't formally taught, but I'm quite hesitant to take on the label of 'self taught'. Unless someone grew up in a cave without the internet and emerged a programming wizard, I'd caution others too.
In some ways, the purpose of a university education is to give you the ability to learn new things. The best example in our field is programming languages. While many of us can code in several languages, we probably did so without the aid of explicit instruction.
It certainly comes down to definitions though :)
It's just that all programmers are self-taught to a degree. You more than most.
If you lived alone in a cave and independently discovered programming while there, I am not sure you are teaching yourself any more than scientists doing groundbreaking research are teaching themselves. We typically file that under research. It only said to be taught after the discovery.
The dictionary defines self-taught as learning without being formally taught. As far as modern lexicon goes, I have to side with the dictionary.
It's still a vaguely useful term, but I do think it's important to keep the wide variation along more than one axis in mind. Some people mean something like, "I read the manual and taught myself to code", while other people mean something closer to, "I studied something close to a regular CS curriculum, but outside a traditional university lecture environment".
But what material did we cover? Pretty much http://philip.greenspun.com/sql/ with some additional topics. I had seen that online book before. Why had I never made myself study the entire thing? Why had I never gone through all of the examples on my own? I don't know. I guess I didn't see the value in it. Now I do.
After displaying great disdain towards structured education for the first 8 or so years of my career, I eventually decided to go back to school for a masters. Things were considerably harder; I discovered that there were parts of math and theory that I didn't quite "get". Often I had "seen" these during my studies, but hadn't looked at it in any detail because it didn't seem interesting for some reason or the other. I survived masters (and actually did quite well thanks to some guidance from professors and being well organized)
After another couple of years of working, I decided to do a phd and proceeded to discover in great detail the list of fundamental things I didn't understand about topics I thought I knew well. It took quite a while to cover that ground back.
Overall my experience has been that the purely interest driven approach I took is extremely valuable. But at the same time it makes one prone to fundamental gaps in understanding compared to a more systematic approach. (Just prone. It may not happen). I have found it valuable to realize this. Among other things it encourages me to continue poking at the subjects I care about.
p.s. for the same reason many people have a CS degree and can't code.
A friend of mine who both has years of real world experience and a PhD in CS is fond of saying: "A 'Web Developer' is a person militantly ignorant of computer science who spends most of their career re-implementing elementary computer science."
It's sheer hubris to think that one can disregard decades of CS theory and be a good developer. You might know the syntax of a language or have a great memory for APIs, but if you don't know how to efficiently apply that knowledge, what's the point?
"Web Developers" often have a much more practical sense of how computers and the web actually work whereas many CS graduate "programmers" tend to assume that the computer and browser will behave according to theoretical principles.
Both "web developers" and "programmers" are a crucial part of the web app development process. Just because someone can't write a sorting algorithm does not mean that they can't be of value. Similarly, just because someone can write a sorting algorithm does not mean that said algorithm will actually perform in IE7.
Leaving aside the scornful sneers at those "web developers" (which read more as envy that "this chump makes more money than me and didn't spend as much as I did on a degree"), there is a strong tendency for people to assume that what they know and can do is special, while what someone else knows and can do will never be up to snuff. And there's a tendency to extend this to group membership. Witness the hate of people with theoretical CS education for people without.
In reality, there are large fields of programming which are well-trod enough at this point, or possess layman-oriented tools of sufficient caliber, that years of systematic formal theoretical CS education simply is not a necessity. Which, to be honest, is a good thing; programming is a useful skill to have in the toolbox, and restricting access to or participation in it by denigrating dabblers or self-taught amateurs is a net harm.
Consider, for example, how many people doing quite successful and worthwhile things today got their start writing BASIC on their home computers a decade or two ago, and then consider how many of them and how many of those successful and worthwhile things we'd have if they'd all listened faithfully to Dijkstra's bile.
I am both self-taught, and have a CS-type degree. I've walked both sides of the street on this issue, at times I've thought I wasted 4 years at uni, and at others I've been glad I sat through at times dry and disinteresting lectures, when I get to apply that theory in the "real world."
- decomposing my code into patterns, abstractions, without going overboard,
- documenting stuff so that I will understand or other people will understand things in 6 months time,
- writing code on a team,
- having a good project workflow to handle source code versioning, releases, packaging
- knowing how to deploy things to a multitude of servers
- debugging caching problems, optimizing a DB for performance, understanding runtime performance of javascript,
- dealing with customer expectations,
- dealing with deadlines, unexpected events, etc...
- dealing with legacy sourcecode of hundreds of thousands of crappy bullshit.
Even in the aspects that "could" involve more theory, like optimizing a DB for performance, I find that theoretical aspects have pretty much nothing to say. There may be some CS involved on the patterns / abstractions level, like which data structure to choose or how to structure a communication protocol. But to be honest, I don't really have the time and the need and I just resort to whatever library people think is cool and stable. And that works pretty well.
None of the things above is simple, and pretty much none of it is taught in CS. And that's what makes a good developer I think, not whether I'm able to bang out a petri net or a neural network.
I'm a little bit sensitive to this. This morning I received a 55 page 'requirements document' from my management-appointed senior software engineer (whose background is in forestry). It amounted to little more than a very complex relational database schema and some algorithmic pseudo code. I then spent 4 hours via phone and email trying desperately to explain why those weren't requirements. His forceful rebuff: his approach had worked in the past, so it should work again. I didn't blame him - he just didn't know what he didn't know.
I've worked with some damn good programmers who had little knowledge of computer science or software engineering. They've been awesome at cranking out some (even clean) code, and they're some of the smartest people I've ever met. But, software is hard, the breadth of knowledge is vast, and just being a good programmer doesn't cut it.
Some people believed the Ruby on Rails "magic" was the worst violation of software engineering to ever exist, while others felt it was the greatest advancement in programming to date.
No viewpoint is wrong, because there is no such thing as right.
"2 + 2 = 5"
"do mutlti-threaded reads and writes to a shared mutable data structure and you absolutely never have to worry about race conditions, clobbering or corruption"
do you still think all viewpoints are equal and none are right?
Take, for example:
做多线程读取和写入到一个共享的可变数据结构，你绝对不会担心竞争条件，会破坏或腐败
Maybe you can grok that, though chances are that you cannot. It is still just as valid for someone to believe it is an efficient method of communication though, even if you disagree.
Generally accepted patterns can be useful, but it takes people pushing that envelope for us to evolve.
Just read Moonwalking With Einstein, which touches on this.
Seems like this guy Dr. Ericsson in Florida is the go-to expert on mastery in general.
As working programmers I do think it's easily in our interest to spend time learning how to acquire new skills as well as we can.
Moonwalking With Einstein: http://www.amazon.com/Moonwalking-Einstein-Science-Rememberi...
Dr. Ericsson: http://en.wikipedia.org/wiki/K._Anders_Ericsson
In the University the education has to be focused on teaching what is needed, not what an individual wants to learn. I did not like databases, but I took 2 db courses. I did not like compilers but I took 1 course just focused on this topic. And both topics are useful, and somebody working on CS need to know its basics. This cannot (and should not) be decided by an individual sorely based on his interest or needs. A basic layer of knowledge is decided by experts on the field that consider that ANY xxxx (put here whatever career you want) must have.
What I felt in the fields I am an autodidact is basically a huge lack of basic knowledge. How am I supposed to decide whether the book A is better than the book B for the topic I am trying to learn? Do I need to know something before getting into this other topic? How I decide the path I need to follow to reach a certain knowledge? Those questions are (almost) always solved with somebody guiding you through the process of learning.
There is something that has always shocked me: is that most autodidacts cannot accept that their education might be partial and the lack of base (generalizing here) might affect their adaptation for new projects or their speed on learning new techniques.
Following the structure of a university in careers like CS or engineering should be a must, and new methods of education are stepping on the learning process by helping autodidacts to be directed into a clear path of what-is-needed.
This is rapidly becoming a thing of the past:
http://ocw.mit.edu/courses/electrical-engineering-and-comput...
I'm 3 years into a software engineering degree and in all honesty, I'm doing it for the piece of paper. Last semester, I attended about 10% of the classes, studied by myself and still managed to get good grades. I wouldn't recommend it to people who aren't passionate about the subject though.
One thing I hate about the 2 universities I went to (Montreal, Hong Kong) is that many teachers (un)knowingly discriminate against students who don't attend classes. This often takes the form of giving hints about the exam during tutorials, giving privileged information about which topics will be evaluated and which won't. In Hong Kong, I have even seen teachers give participation grades or require a minimum class attendance in order to pass (this happened in computer engineering/science!).
You are no longer an autodidact and you are studying the topics the college is telling you to follow. Some of them you would have never consider if you were to chose, some others you would. But in the "high level" (and probably in the future) you would appreciate the starting point that college has given you.
PS: I agree completely with you regarding attending to classes, even though I would have homeworks/exams to help retention of the material.
Learning by yourself should obviously be encouraged, but it's often difficult to know what thing to learn about next in order to ensure that the new knowledge integrates well with what you already know and (pragmatically) helps you solve the problem at hand.
To give a specific example, I'm lucky enough to sit next to our chief infrastructure engineer at work and he has helped me learn a huge amount about Linux over the past three years. He hasn't done this by giving me "formal" instruction, but instead he's pointed me in the right direction when I get stuck with something.
For one thing, a teacher cannot shove anything down your throat. It is always the student who learns and in that sense, everyone is an "autodidact". This is not to belittle a teacher who's role in directing a student, keeping her on her toes, challenging, questioning, probing, explaining, etc. which are all valuable. The thing is, a formal teacher is not the only source of those things. Our parents and peers do those things too.
You learn, you get good at something ... anything. In today's world, feedback from the community on what you need to get good at and whether you're getting good at it seems adequate to the point that if you know how to make use of that, you can learn stuff efficiently and effectively. You might say more of us are becoming "allodidacts".
The third point is teachers also continuously learn. Peer learning is how scientific progress happens, for instance. So if you do continuous learning at all, you're likely to do it for a much longer fraction of your life than the period that you're "taught".
Reality directs your learning.
With or without a mentor you are still being forced to work in a particular direction. Do you read a book for your own benefit or for your employers or for both? Who appraises you? Are your colleagues impressed by what you've learnt?
Unless you live on a desert island you probably have constant feedback.
Isn't it part of being self-taught to find mentors - in books, online, and if you're lucky, in person?
This is not some strawman I want to knock down. From the article:
"...there is more to being a good student than being good at learning. One of the responsibilities of a good student is to seek out excellent teachers. In the Wikipedia article on Autodidacticism, I find this paragraph:
"Autodidactism is only one facet of learning, and is usually complemented by learning in formal and informal spaces: from classrooms to other social settings. Many autodidacts seek instruction and guidance from experts, friends, teachers, parents, siblings, and community."
He misses the second sentence of his Wikipedia quote. He argues that you must sometimes become the student.
Any excellent student knows he must eventually teach from his corpus of knowledge to cement what he knows, and pit it against new, hungry students.
Even I - the autodidact - knew to communicate what I learned, test it in the "field of battle," wherever I could find that.
For me, it was all before I reached "higher education." So by the time I had professors, classmates, and TA's, I had already developed a highly effective method of autodidactism.
As far as selecting the best teaching material, I win. I can pick any used copy of any books that I feel are necessary, and they are cheap. Textbooks with exercises are a must. If I was shut off from the rest of the world, the sheer volume of college textbooks I physically have a copy of would keep me busy for years.
You don't learn from the teaching, you learn by doing. If you teach yourself and you never use what you have learned, done 5-10 of the medium to hard problems, and then tried to put together what you have learned, you will fail.
In the end, we agree. You should NOT be the sole teacher, unless you plan on just living and working for yourself. Perspective is a hard thing to get from yourself.
It's more than just a separation, it's an inherent power-asymmetry.