Last week I’ve got such a message from my blog:
Sent: Thursday, June 29, 2006 1:17 AM
To: Eldar Musayev
Subject: (Random Thoughts and Hints on Software Development) : Hello Eldar!
Can you please tell me what I should learn to be a good software developer? Thank you.
Hm-m-m, tough question. No, seriously. A lot of people proud of being asked about such fundamental thing would gladly go ahead and give a ton of … well, you understood what I meant. Anyway, I did not want to do that. If I give an answer, I want to give a real answer, at least as good as I can. So, I postponed answering until I can find the time to give my thoughts on the subject. And, please, understand, these are just my thoughts; there is a lot of other very qualified and decent developers who may have a different opinion.
“Sometimes it’s almost as good as sex”
So, IMHO the first qualification for a good developer is that he or she should enjoy writing the code. As one of the famous hackers of the past said before “Sometimes it’s almost as good as sex”. Actually, I suspect that this is exactly the sentence that made him famous. But anyway, that’s what I am talking about. Let me explain why.
First, software development is a very demanding field. It’s not a 9 to 5 job by definition. Of course, you and your manager should try to keep it 9 to 5 (meaning not policing you to come at 9am, but rather trying to prevent you from staying late and sustaining yourself on company-bought pizzas while fixing the code around the clock), but – alas! – that’s not always possible. You simply cannot do a good job if you don’t really love to write and handle the code. Otherwise, a couple of milestones or releases and you will hate the job. So, you simply have to enjoy this work to be able to do it long enough. Sure, some do that long enough without, but we are talking about good software developers, right?
Another reason is learning. This is a profession where you have to learn, learn, learn without any breaks forever. Software development defies scientific management of Taylor invented early in the XX century to destroy trade unions of skilled workers like arsenal workers. The very idea of the “scientific management” may be shortly expressed as “look at what those smartasses do, decompose it into simple steps, hire a bunch of cheap dumb guys, teach them the same steps, fire smartasses.” In late 50s “the father of American corporate management” Peter F. Drucker discovered a completely new category of workers whom he called “knowledge workers”. Knowledge workers defy the power of “scientific management”. The problem with knowledge workers is that they know what to do themselves. What’s worse, if you try to decompose what they do, then by the time you teach a bunch of dumb guys the same thing, they already do something different. They know what’s needed, you – the manager – don’t. That’s a challenge for both the manager and the worker. Software development is the prime example of knowledge workers. As a software developer, you have to learn all your life non-stop. There is no break, no finish line, that’s a race to the horizon. Try to learn all your life the things that you don’t like, and you may end up in a mental institution. And if not, you’ll hate your life for sure. You have to love writing code.
There is also a third reason. Did you ever met a person who thinks that he is a great writer, but writes a louse fiction. He annoys his friends and relatives, editors, and simply cannot get where he wants. He does not improve. Beside being oblivious to his own shortcomings, such guy has a really hard time getting an informative honest feedback on his job. Friend and family lie to him, saying that he does a great job just to be nice. Editors simply reject his manuscripts. How can he learn? We, humans, need immediate feedback to learn effectively. That’s how our neural circuitry works and learns.
Fortunately for us – software developers – computers are much more critical and at the same time more willing readers than whom you can find in a literature field. Sometimes mere milliseconds are required to get your bugs in your face, when your code executes. This is not always pleasant, but it provides the necessary continuous feedback into the neural networks of our brains getting them into the shape to become able to create programs.
Computers are ruthless in exposing defects in your craftsmanship, they don’t care for your ego, they don’t let you save the face, they simply don’t care: they execute exactly what you ordered, and if you ordered a wrong thing, that’s what is going to happen and that’s what everybody will see. Actually, this is very good for us, this helps us to perfect and learn. “Everything that does not kill us, makes us stronger”, but unless you really love writing code, this kills your will to continue.
That’s why you have to enjoy writing the code. That’s why enjoying writing the code is the first requirement for being a good developer.
Of course, everything can be at extreme. A typical hacker enjoys writing code a little too much. Computer and coffee is all he wants in his life. Those guys can never become good software developers because beside the desire to write code, you also need a desire to write good code, a code that other people will use (and hacker doesn't care for other people). Fortunately, most people pass this phase and come to their senses, often at the same age when raging hormones start to balance better and find other channels for exit.
Of course, beside the attitude, there is simply a technical side, skills. You have to learn a set of basic knowledge and skills to start. However, this is not that hard question. Just look at computer science curriculum of a good university, that’s it. Those professors are experts in the area, they know what to teach a student. Don’t invent a bicycle, just look at what they teach.
Of course, college differs form a college and a university differs from a university. Some make a better job, some are not. I am far from endorsing any particular university or a college, but in my humble opinion University of Waterloo in Canada makes a superb job teaching their students computer science and software development. There are other universities in United States and Canada that make as great job teaching their students. Examples include MIT and University of Alberta. If you want a more complete and worldwide list, check ACM competition site here: http://icpc.baylor.edu/icpc/Finals/ And, of course, I am personally fan of my own Alma Mater – St.Petersburg State University in Russia, that ranked 6th in 2006, but was first on multiple occasions in previous years.
Anyway, once you like the field, learning is much easier.
More to it…
Still there is more to it. I hardly can give a complete list but here are a few examples:
– ergonomics, usability, perception psychology… Should this button be red or gray? How much space menu should take on the screen? How many items can you put into a scroll box? Why actionable control should be visually different from indicators? These are all not the questions about the technical side, but about human ability to handle it.
– What is the main language for the programmer? English (or whatever you use in the office). Communication skills are very important part of software development. Negotiation skills – depending on a team – may be of an utmost importance. Sometimes, either you can convince the team to go your way, or those who cannot write code but skillful at negotiations will have it their way, and you will be stuck with fixing impossible bugs and infinitely patching your code (those guys usually lack the design skills too).
– Psychology, empathy and investigation/interrogation skills. How do you find out what exactly the customers expect from your programs? Even if you have a real customer in front of you, people in sales, marketing and executive positions (who often represent a customer for you) are not cured with the brutality of computers that train you to tell truth, truth and only truth, because any lie (bug) immediately blows into your face, even if you believed it yourself. Actually, they are often trained by superiors and peers in rather opposite things. Sometimes you can meet a customer who can vividly articulate what he wants, and that’s your lucky day. Unfortunately, in many cases you have to dig out the truth, which is not easy. And that’s in the case when you have a customer in the first place.
– And what if you don’t have customers yet? What if you develop a new revolutionary product that nobody (including competition) have thought about before? Of course, you have a business concept by that time, but how do you fill the blanks? A serious help may be marketing – by the way, marketing guys, at least at Microsoft, are great: they are smart, know what they are want and have a sixth sense at what the customer needs. But your marketing guy cannot fill all the blanks too. What it boils down to is that you have to have tons of common sense to make the right decisions. And some marketing knowledge does not hurt too. Understanding how a product for X-ers should differ from a product for Y-ers may be essential. By the way, do you understand this difference?
– Networking, and I don’t mean TCP/IP. Friends, keeping connected, socialized is very important in our profession. I am not very good at that myself, but even so more than half of my jobs in US (including internal transfers at Microsoft) came from networking and referrals from my friends. If you really good at networking, you may need no resume at all. Beside that, it gives you more. How do you track what’s new and worth attention around? Trade magazines usually have no clue, and even when they have, they are still filled with tons of paid ads junk. Friends are your source of upcoming technology alerts, no-bull overview and introduction to new things.
– …and that’s not all… Feel free to add your thoughts at what is needed for a good software developer in the comments to this post.
So, I am wondering, if I managed to answer the question? I don’t know. I tried my best, if you can do better, do so and leave a link in the comments.
----- Added 8/24/06
A great comment with the link worth adding to the original post:
As much as I enjoyed the post, I think it should include the ACM Curricula Recommendations (http://www.acm.org/education/curricula.html). I'm glad you added the "More to it" section. Software developmen is not just about the code; it's about being a professional who truly enjoys working with others (internally and often externally) to solve real-world domain problems.