I think he's operating with the assumption that all languages have the same concepts, just with different "lexical and syntactical bureaucracy." But I think the point of the advice to learn more programming languages is to expose yourself to different programming paradigms. And in order to do this, I don't think you need to become an expert in the nuances of that language, but you need to become comfortable in thinking in new ways.
If your language of choice is C, and you learn Pascal, then you haven't gained any new concepts. But if your language of choice is C, and you learn Prolog, then you will be exposed to a whole new way of thinking.
My native state of coding is procedural with a sprinkling of OO, mainly to represent families of concrete data types. I find functional programming appealing on an emotional level, and my attempts to learn how to think functionally have changed my native coding habits, even in non-functional languages.
My pragmatic advice is learn enough of a language from a different programming paradigm so you can apply that new way of thinking to your native language.
I didn't read it that way. He's seems to be claiming that getting good at a new language involves a lot of non-mind-expanding overhead---which, in the author's estimation, isn't worth the opportunity cost past one or two additional (sufficiently different) languages. I think this passage sums it up:
Aside from the immediate reasons, there's some merit to the mind expansion argument. I think being proficient in at least two languages is indeed important for boosting your ability as a developer. This resembles human language: learning a second one changes the way you think and your perception of the world. The third or fourth, not as much. But it can't be any two languages. If you know Portuguese and Spanish, your mind didn't have to expand much. Likewise, learning VB.net and C# doesn't count.
I've found that most languages I consider worth learning have fairly small overhead. The syntaxes of ML, Lisp, Haskell, Smalltalk, Prolog, Lua, Ruby, Qi and Python are small, simple and easy to learn[0]. Most of those have small standard libraries as well. Learning any one of those, assuming you don't already know a very similar language is likely to change how you think about programming. Even if you don't use that language much, it will alter how you code in other languages.
Is this worth it, from a strictly professional perspective? I have no clue. I've never met a strictly professional programmer who was any good.
[0] This should not be taken to mean that these are the only languages I consider worth learning - they're just the first ones that came to mind.
Note that I said the syntax is small for all of (list of languages) and most of those have a small standard library. CL has a simple syntax and a large standard library. By extension, so does Qi.
Sure, those are all parts of the syntax of a language, but I still maintain that CL's syntax is simpler and more regular than that of Ruby and vastly simpler than that of Java.
As an aside, the first form of the example you gave is what most Lisps use. CL, Scheme, Emacs and Goo all use this. The second, as far as I know is only used by Arc, which has a more complicated syntax than most Lisps.
My pragmatic advice is learn enough of a language from a different programming paradigm so you can apply that new way of thinking to your native language.
I think "enough" is a crucial point here. The author almost seemed to be arguing against switching to a new language every year, which I agree is ridiculous. The idea is to learn some new concepts, not to write significant programs in a new language every year. I only adopt new languages for regular work if they're the pragmatic choice.
That's exactly it. Saying that learning a new language is just a matter of learning a new syntax is like saying that an English-speaking, non-German speaker could translate a German text to English by taking a German-English dictionary and going through and translating the words one-by-one. It's not quite as simple as that.
Actually, I'd compare that to saying that two languages are the exact same except for some of the words -- that JavaScript is exactly like Java except System.out.print is renamed to document.write. Saying that learning a new language is just the syntax is more like saying that translation between English and German can be done with just a German-English dictionary and a full knowledge of the grammar.
However, just as lambdas cannot be directly translated into languages without lambdas, some languages have unique concepts. For example, the German language has a part of speech called the particle that conveys a general feeling into sentences. While the statement "Er wisst das." could be easily translated with a dictionary and knowledge of conjugation rules into "He knows that," adding a particle to make "Er wisst ja das" could be variously translated as "Of course he knows that," "He obviously knows that," "He most definitely knows that," etc.
I find such comparisons misleading. The differences between natural languages are not at all the same thing as the differences between programming languages. All natural languages have the same expressive power, or at least very close. The existence of the phenomenon of translation proves it. You can take a book in Classical Latin (according to my friend's girlfriend who is a linguist, a very terse language) and its translation into German (typically given as an example of a very verbose language) and the difference in length will be maybe 30% (and frankly, I don't even believe this difference can be accounted for by the greater expressive power of Latin; my guess is it's just that German words are longer on average.)
A real phenomenon: there are people who never get anything done because they are constantly dabbling in the language flavor of the moment.
Another real phenomenon: some of the best programmers in the world still rely entirely on C. [Late edit: "and outperform developers in modern languages"].
I agree, absolutely. But I wouldn't go so far as to say that "dabbling in languages is counterproductive". That's like concluding that martinis are fundamentally wrong because some people are alcoholics.
I'd also suggest that of the two pathological conditions -- Overdabbling and Underdabbling -- the Underdabblers outnumber the Overdabblers by a huge margin. We just see a lot more of the latter on the Internet, because for many obsessive dabblers the whole point is to brag about it.
You need to find a balance. In the end, I think this essay is much more about that balance than its title suggests.
I'd like to see you defend your argument, that "underdabblers outnumber the overdabblers". Control for talent, please. Clearly there are thousands of VB forms developers out there; let's stick with the people working on interesting problems.
Well, if I'm not allowed to casually wave my hand and invoke my legendary army of VB forms developers, my "argument" will evaporate, as you no doubt suspected. :)
Because, alas, I can't name any talented folks, working on interesting problems, who are obvious under-dabblers. This may be because I have led a sheltered life: Too many of the programmers I know are from MIT, where until recently they inoculated everyone against underdabbling by making them take an entire course on Scheme. I tend to work on open-source stuff that's hosted on Linux, another community where underdabbling is rare and overdabbling is the danger. I read news.yc, which incorporates plenty of dabbling -- much more than I have time for, really, and perhaps more than you have patience for. ;) Finally, I myself am only a programmer because I kept dabbling in software while I was supposed to be working on physics. If I had the personality that found it easy to focus strictly on one thing, I'd be a semiconductor process engineer right now.
Of course, the statements that the original article took such exception to -- the ones like "you should learn a programming language every year" -- are largely intended for the less-talented people that have just been defined out of the discussion. And, again, that's kind of where the original author ends up: "Once you learn one language... oops, make it two languages... no, five! Once you learn five languages, you should definitely stop dabbling!"
From Peter Norvig's "teach Yourself Programming in 10 years"
"Learn at least a half dozen programming languages. Include one language that supports class abstractions (like Java or C++), one that supports functional abstraction (like Lisp or ML), one that supports syntactic abstraction (like Lisp), one that supports declarative specifications (like Prolog or C++ templates), one that supports coroutines (like Icon or Scheme), and one that supports parallelism (like Sisal)."
That's actually very similar to the point this writer makes: that learning different languages does have benefits, but that they should be significantly different. The two just differ on whether you should learn many different languages, or a few.
These days, I don't know if I can trust advice on the internet anymore. Could be they just want to send the competition down the wrong path (Norvig is commercial now, working for Google). Follow their advice, and you waste a lot of time, while they are busy making progress.
If you're trying to compete with Norvig right now, for god's sake don't waste time learning more languages. You have a business to build.
If you're willing to bide your time and you hope to develop jedi-level expertise eventually, then I think it's still good advice. But again, it will pay off only in the long term.
Likewise, if someone is beating you up after school every day, don't start studying a martial art that will take you 5 years to learn properly. Learn something quick and dirty. But if you want to learn to be a true master of martial arts, it will be a lifelong study and you will likely need to learn at least several styles.
But, as noted, brass knuckles (or a gun) will not cause you to grow up as a person. The martial arts teach patience and poise, which is almost exactly the opposite of a quick fix.
More concretely, Sig Kufferath (who basically carried the Danzan-Ryu torch to mainland USA) apparently beat the hell out of a couple guys who tried to mug him, then got them medical help and sent them flowers in the hospital. (The story may be apocryphal, but I would not have put it past him) Norvig's motives appear similar -- he is an acknowledged Jedi and has little to fear, in general, from helping others to raise their game. Having crossed paths with both guys, I feel reasonably justified in my perception.
My apologies for drawing Norvig into this - actually I also hold him in high esteem and I don't really think he had any ulterior motives for that remark. And I do like to learn new languages.
HN sometimes makes me think what I wrote, though - a good way for the competition to keep you busy, for example by posting math riddles and stuff.
His thesis: that learning the basics of most new languages is an exercise in tedium and trivia with little or negative benefit. He instead recommends reading high-quality code, learning computer science theory, and mastering a few quality tools (not Java).
I thought this was going to be a "let's make fun of this idiot" Reddit-type post, but I was wrong.
I promised myself long ago that I would never engage a "language war" thread. Not sure if I'm actually breaking my promise, but this article was too compelling to ignore. The author states what I've intuitively known all along, but was never really able to explain.
I am proficient in about a dozen languages, but only use 3. I go back to the SPS, PL1, Fortran, and COBOL days, and would rather chew razor blades than ever use them again. Today, I use javascript on the client and php on the server; I have yet to find a problem I couldn't solve with them.
In the years in between I discovered BASIC and have found it to be the Swiss army knife of apps. Not the old Dartmouth Basic, and certainly not anything Microsoft bastardized, but there are many other versions that seamlessly integrate with relational data bases that I think are a dream. <laughter dies down>
It's actually reached the point where I "think" in BASIC, design my app, and then sometimes write in in javascript or php.
I certainly wouldn't recommend my approach to someone else, but I think that's the whole point. Use the right tool for the job and get good at it. I see no need to learn new languages just for the purpose of learning something new. If I get "curious" or want to expand my mind, I can think of several hundred app problems waiting to be solved in any language, take your pick.
Thanks, Gustavo Duarte, for the chance to put a little thought into this. Time to xfer control back to work (GO TOs not allowed).
I think he misses the important benefit of being more marketable as a programmer. I have a major problem when I interview people and they are totally focussed on one language (e.g. Java). It's not the language that ends up worrying me, it's the lack of curiosity.
For that reason I often ask people in interviews to tell me about the computer they have at home: what OS does it run and why? And I ask them what they hack in their spare time. That tends to be very revealing.
Purely anecdotal, but I went to a Boston Ruby group meeting a couple weeks ago. I didn't hang out for long, but both of the 2 people I talked to were there to hit on developers! One offered me a job in the elevator. Last week, a funded startup called me with a job offer out of the blue--they new a friend of a friend.
I think it depends on the job you're optimizing for--ruby and python are good choices when optimizing for startup jobs.
It's hard to find Ruby talent (I have a Ruby product team). But any clueful employer will hire a strong Python dev for a Ruby slot, any Lisper for pretty much any non-C slot, and even a Java dev if they rockstar their interview.
The people who have Ruby slots tend not to keyword screen.
The people who have Java slots do (or, at least, their directors do).
> I didn't hang out for long, but both of the 2 people I talked to were there to hit on developers!
+1 for using the dating metaphor for recruiting/finding a job. How often do developers get hit on in the dating world? On dates or even just talking to a girl, I find I quickly steer the topic 'what do you do' to something like 'Heard of Facebook? I want to build stuff like that'
"What is a date, really, but a job interview that lasts all night? The only difference between a date and a job interview is that in not many job interviews is there a chance you'll end up naked at the end of it." -- Jerry Seinfeld
It happened when a b-school grad tried to talk me into joining his startup. I contract sometimes, so if he was funded, I'd have "given him my email." It was the sense of having something someone really wants when what they are offering isn't something you're interested in. The predatory leer also triggers some part of the "flight" instinct.
It was also enlightening in the sense that the rejection wasn't about him (he was a Sloan grad, and a nice enough guy), or his idea--it just wasn't a match. I work on unfunded startups, I'm just busy now and didn't care for his idea.
When I had to politely repeat how busy I am, I could suddenly see how being an attractive woman who had to fend off advances could be exhausting, and how easy it would be to stop being polite if you had to do it all the time.
Even that can be misleading. I share the post's distaste for Java, but I get paid to rip apart cool applications, and lots of them are written in Java --- even more are written in C.
That's true across all the verticals. Plenty of 2.0 apps are Java --- including large swaths of Google. Careful about myopia. In the end, the problems you work are are far more important than the language, unless you suck. I'd much rather work on a parser/compiler in Java than a report generator in Ruby.
I think the article would be received more warmly if it were put into the perspective of a company project. I.e., don't implement this new product in XYZ just because I want to know/learn XYZ and it's all the rage right now. That would be a waste of time.
I'm a full-time Java programmer now because it pays the bills. But you'd better believe I'm learning other lanaguages on the side, because it's pretty clear that Java 1.) has some serious limitations that other languages don't have, and 2.) is falling out-of-favor as a web development tool to more flexible, less statically-typed, more functional languages.
It's not so much to "open my mind" as it is to keep myself marketable after Java's impending demise.
> It's not so much to "open my mind" as it is to keep myself marketable after Java's impending demise.
As someone who is starting to step out of the comfort zone of Java, I agree about marketability. However, Java is not going to disappear, it's firmly entrenched. It's like COBOL. I'm not sure how Sun pulled it off exactly but Java in it's 13 years has spread itself in a lot of corporate systems/applications/backends (some of which may be COBOL). Ruby on Rails and other minority languages are the minority in some companies precisely because the majority don't like the minority ("RoR? We don't use that"). However, there are some rogues in companies who are trying to bring RoR in discretely (I'm trying to be one - developing business cases for RoR adoption on small projects)
Completely agree you shouldn't choose a language just because you feel like learning a new one but the article ignores the design implications of choosing one language over another. I just don't think it is a particularly complete or well reasoned argument and is a border line rant.
Waste of time? Quite a lot of fun things could be considered a waste of time. Learning new languages can be fun. People learning one language and sticking with it for life is why COBOL stuck around for so long, and Java seems to be going the same way.
EDIT: I just read the article. Maybe it's just that many of the languages I've been learning have had little syntax and no operator precedence quirks (or any operator precedence whatsoever). I also noticed the article made the same comment about COBOL and Java that I did. I really should read these things first. Anyway, I guess as a high school freshman I'm still at the point where learning new languages is really the best I can do.
> Maybe it's just that many of the languages I've been learning have had little syntax and no operator precedence quirks
This is a good point. It is probably not a good use of your time to learn Java and then C# (or similar), or Perl and then Python and Ruby. Sure, there are differences, but basically the languages in the same groups are the same.
If you go for the more "academic" languages, you'll get more return on your learning investment. Recently, I've learned Lisp (this is what started me down the "wait, learning new languages isn't a waste of time!" path), Smalltalk, and OCaml, and I've picked up a lot of good ideas to port to my primary language. (Otherwise, my background before that was C, Perl, and Haskell. All good languages to know :)
By all means, play around with different languages when you're getting started. I think it's really important to be exposed to different languages as part of your education, be it undergrad or (for you as it was for me) junior/high school.
But even in your situation, I'd say balance languages with other knowledge that is important. It's easy to play with languages because it's fun, but there's a lot of other important stuff to learn, some of which might make more practical difference to you.
But by all means, learn and have fun. If you're getting a kick out of languages, have at it, especially if you've got plenty of time.
> If you're getting a kick out of languages, have at it, especially if you've got plenty of time.
Plenty of time? No, not the school I go to. What I do have is the syslab, which is full of Linux boxen and a lot of smart people, especially during lunch. Attached to the syslab is a small library (formerly storage closet) of programming-related books, which contains, among other things, multiple copies of ANSI Common Lisp by Paul Graham.
Learning new languages teaches you what you know and most importantly don't know. Coming from a Java working background, I find Ruby on Rails pleasantly abstruse and remarkably concise (which makes it hard to learn, coming from the land of boilerplateness).
I have a long commute on the train and I find that when I am trying to debug/learn - (learning Ruby on Rails) - the time flies by like that. If that isn't a benefit (making time fly by when you're having fun learning), I don't know what is.
Learning RoR is fun because I like personal challenges. If it were easy, it would be boring. It's like a new way of thinking, not just programming. And like jjburka said, it's like taking a vacation from the land of Java.
I don't want to pretend that I am not learning RoR to be more marketable, and it's a joy to learn RoR, to build little RoR apps that I can use from my iPhone that make my life less stressful/easier.
Because I don't have 2-3 years of experience, I realize that I won't be able to pass a rigorous RoR interview. Especially those at interesting startups. My workaround to this might be by developing one or two Ruby GEMS that add value to the ecosystem (going to RailsConf '08 will help me get my bearings)
The article's not bad. His main focus, and he states it, is not to constantly learn new languages for the sake of profit. As a hobby cool, but not for profit. I can agree with that. There are all sorts of topics you could learn for fun and profit, and investing in a new language is not necessarily the silver bullet to mad coin.
Who cares though? I think as long as we're constantly learning, whether it's new programming languages, economics, or snowbaording, it makes life interesting. New skills always seem to come in useful at some point. Make the investment in learning things that interest you and it'll always pay off. Don't let someone else tell you what NOT to learn. That's jut silly. Heck even learning COBOL or FORTRAN would benefit you with a little insight, and might be fun in a retro sort of way.
Yet you profess to think in BASIC, and FORTRAN is very, very similar to BASIC.
I don't follow the metaphor about COBOL. Although it's syntax can be burdensome, the language is very efficient for the same reasons FORTRAN is: static typing, no dynamic memory allocation, efficient file I/O and the best compilers available. COBOL programs are not necessarily flexible, but the are fast and efficient.
Also COBOL has correct "accounting math", something that almost no other language has as part of the core. In fact I can't think of another language specification that implements fixed-point math as part of the core: usually it's part of an external library and is cumbersome to use.
I guess this is why I never post in language threads, we can go on and on all day. No one is right or wrong. Anyway, I'll try to address your comments.
There are so many different versions of BASIC; it's possible we are visualizing very different versions. I am accustomed to using INFOBASIC dialects as in JBase or IBM'S U2 line. Everything numeric is represented as strings (integers) including dates, times, and decimals, so there is no "typing".
As for efficiency, I used to benchmark like crazy, looking for the slickest algorithms and the best implementations of compilers and I/O routines. Not any more. Hardware has gotten so fast that I am more concerned with human time, not machine time. (Naturally, anything that runs 8 million times per second BETTER be fast.)
These days, I generate most of my code, so the verbosity of COBOL (if I used it) shouldn't be an issue. But it is. I have to drop down to the source so often and I work on the code of others often enough, that I am most concerned with getting the whole job done with the LEAST code. This is where "juniors" fall down the fastest (IMO); there's too many places for things to go wrong.
I know what you're thinking, if this is the case, why not learn even more advanced languages and frameworks with even less code. I dunno.
I guess I've "settled into" what works best for me (as everyone should). Is it the best? Probably not. Am I open to better ways? Probably. But not today. Too busy.
I think that the writer is deliberately being controversial just to get a rise out of the readers. Having said that - he does make some valid points.
The bottom line is that you have to use the right tool for the job: it could mean using what you know, learning something new, or even inventing something totally new and original.
I get the impression that the author didn't think through his argument before he wrote it. He starts off saying one thing and proceeds to contradict himself. Really there are advantages and disadvantages to everything. Spending too much time learning new languages and never master CS basics is clearly a waste. However, never learning a new language, or more importantly, never programming in a fashion or environment that is unfamiliar to you, can lead to an insular way of thinking. "If all you have is a hammer, all problems begin to look like nuts," or something like that. Its a matter of being a little bit selective and aggressive with your choices. If a new language is too easy, its probably not teaching you much.
Is this about a refusal to learn new things, though? He does, after all, point out towards the end that good programmers use multiple languages - so whilst a refusal to learn new things is a bad thing, isn't a bit of prudence in how to budget your time a good thing?
True he does qualify his initial assertion near the end of the post by saying that to be marketable you need to know multiple languages. I guess what bothers me about this piece is the sense I get that the author is writing from the perspective of optimizing ones personal profit. He is basically saying the only reason to learn multiple languages is for competitiveness in the job market and even then you should learn as few as you can get away with as well as possible.
That sounds like the argument of a technology major to me rather than a comp. sci or engineer. He totally ignores the fact that many languages are based around the same core principles and if you know those well learning a language is like learning a new word from the dictionary it just happens naturally when it needs to be done.
You misread the article. He's not arguing against learning new things. He's arguing against learning the specific new things you want him to; he'd rather learn business or security.
I've read this article three times over now, trying to find a way to agree with it. I'm sorry, but since this guy thinks learning Dvorak is a better use of his time, I'm afraid I'm going to have to write him, and this terrible article off.
i just read up to here (2nd parapraph):
"The theory is that by learning a new language you "expand your mind" and become "a better programmer". Right. By that kind of argument we should all be eating LSD (don't)."
You should finish it. By the end of the essay, the guy has successfully argued himself out of half of his original thesis, and has laid the seeds for finishing off the other half. That's one of the good side effects of essay writing: You can start out arguing one thing and eventually find yourself having to argue something else.
One of the things the guy missed is there in plain sight in his essay:
You need to get to know the libraries, struggle with the environment, find or tune a text editor, look for tools, and so on. Most of the effort has to do with housekeeping chores, which are surely not mind-expanding by anyone's measure. If you hope to be productive in the new language, things are even bleaker: proficiency has less to do with the language itself than with the myriad technologies you must master to use it effectively.
Unfortunately, everyday productivity is a lot of housekeeping chores, one after another. While your higher mind is soaring with the architecture astronauts, the rest of your mind had better be editing text efficiently, and running top and netstat, and administering the version control system.
And more than half of the point of learning a new language is to force yourself to experiment with a different tool ecosystem. The point of learning Ruby isn't just to learn Ruby: You get to learn about Rails and Capistrano and the obsessive use of test-driven development and, in my case, git. (If I'd been a kernel hacker, I'd have known about git. I presume the Rails folks learned about git because some kernel hackers, seeking to learn a new language, wandered into the Rails community.) You want to be forced to experiment with different tools because sometimes they're better: git is just a much better tool than Subversion, and don't get me started on CVS. Even when the alternatives aren't better, you become a better technologist by playing with them. If I'd never played with a Java IDE I would have no chance of understanding how Java programmers think. If I'd never actually deployed both Rails and PHP I wouldn't understand why PHP deserved to live. (Yes, it's filled with stupidly broken parts, but it's so easy to deploy, and it's fast, and it's not that hard to comprehend.)
The other thing this guy doesn't seem to understand is that there's a difference between tourism and living in another country. When you "learn a new language every year" that doesn't mean you have to become a pro in your new language. You can visit, kick the tires, read a couple of introductory books, write some simple code, read some code to something more complicated, and then move on. Yes, that means you're a dilettante -- but it's better to be a dilettante than to blindly plug away at the same thing for a decade.
I think really what he is arguing is the marginal benefit of learning a new language diminishes with each new language. Which is probably true if you only learn the syntax and then move on. But thats not what software development is ,the language is a tool that a culture uses. Using your example with Rails , alot of its culture (MVC , TDD , etc) seem to be being adopted in other communities. I would argue that the reason is someone had a extended vacation in Hotel Rails liked what they saw and then brought it home. Finding out how different people do different things and adapting them to your needs is valuable in all contexts.
In real MVC, you want to put everything related to your business rules, storage access, etc. in your model. Then you wrap the View and Controller around that as a very thin web <-> model glue. That's all a web framework should do, web-enable your model classes. Things that should not happen in the Controller include touching the ORM (or SQL queries), validating input, any business logic, etc. Just pass the request to the right model, see if the model did what you asked, and display the success page or the error page.
If your language of choice is C, and you learn Pascal, then you haven't gained any new concepts. But if your language of choice is C, and you learn Prolog, then you will be exposed to a whole new way of thinking.
My native state of coding is procedural with a sprinkling of OO, mainly to represent families of concrete data types. I find functional programming appealing on an emotional level, and my attempts to learn how to think functionally have changed my native coding habits, even in non-functional languages.
My pragmatic advice is learn enough of a language from a different programming paradigm so you can apply that new way of thinking to your native language.