On Being Indispensable

I've heard this "one year of experience 10 times" phrase before. It seems to mean different things in different contexts to different people. Can you explain what it means to you and the implications?

If you’re going to practice coding like some master craftsman, employing techniques that took years to hone and where every carefully selected pattern, though inscrutable to an untrained eye, carries the thumbprint of an artist at the height of his power…

Well, number one congratulations to you on finding a patron willing to fund such fine and intricate work. I hope they appreciate their good fortune in having lucked into getting such an artist to work on their project.

But number two, how do you plan on this model being sustained into the future? Have you taken on an apprentice to whom you can pass on the ancient wisdom to which you are privy and which only another equally enlightened ‘cranky old prima donna’ could possibly be expected to grok?

Where are the next generation of ‘experienced and capable’ developers going to come from if you are off working solo in your cave for… twenty-five months at a time on your masterpieces?

And where did you acquire these unique and precious skills that are not available to young developers who are, you surmise as you peer out of the cave at everybody else, only capable of producing dependency-laden garbage? Did you trek to the top of a mountain and study at the feet of the ancient wise ones?

Or did you just write a whole bunch of code, and pick some stuff up along the way?

As you sit in the corner and whittle away at your masterwork, combining thirty year old patterns with the latest language features, I wonder what gives you such supreme certainty that your mastery of your craft is such that you are creating something far above the standard others could achieve?

Because, speaking as someone who works on large team development projects, I can tell you there is nothing in the world of development that scares me more than a developer who says ‘I know this code’s hard to grok but trust me it’s really beautifully engineered’.

> One thing I instituted that helped was a specific Teams channel called Knowledge Transfer.

Oh, that helps explain the predicament. Teams is where Knowledge goes to die.

I always try to de dispensable. We have weekly KT. And we take turns in doing Ops. We also do KT for others roles. As a rough example, we have DBAs, but we make sure devs know how to do the basics of their work too. Whenever you do something someone else can't, assign someone else for next time immediately. Never do the same job twice. It seems faster do just do it, but it's not as soon as the third time happens.

But the real trick is test frameworks that give you rapid access to test all areas of the system with new code or inputs. So if the question is "what if?" The worst case answer, if I am not there, is "try it". N.B. You must be able to test performance impact. This is expensive, you need prod support systems you can hack at and put back to the original config quickly. The ability to restore any test system to a know state (even systems you don't know much about) is critical. So you can try stuff out and then leave it how it was.

In emergcies when there is no time to test. Experience matters. If you are answering "what if?" with "try it" people get experience. It seems faster to answer "what if?" with "I am sure that" but long run that results in two problems, knowledge silos, and a delay in answering until you are at work or awake. That can kill you, especially if you are expert in more than one thing.

If you want backup in a multinational company that runs 24/7 you need one person to answer any question in each timezone. If you can answer a question "immediately" but others can test the theory in 12 hours, that's often faster. You need sleep. In a small company you may not have 24 hours resources, but if the questions cost more than a wage in a foreign company, hire people. Even for day to day stuff this enables you to work around the clock. Get managers to do the math and presume answering any question or doing any task is never immediate, on average its >12 hours, because sleep.

It's also good to avoid doing what you know best (that's hard) but it's better to tell people how, than to do it. If you want to avoid questions, don't just tell them how, tell them why, and how you found out.

There is a social problem to deal with too. Often people think something is "not my job" deal with that early with collective ownership of the whole solution.

There is a problem of permission, sometime only DBAs can touch the database, work on that by giving confidence that others can do DBA work in test systems.

There is a real problem that some people try to be indispensable for their own job security. Fire such people early, or move them to new roles in the company early.

People are generally as smart. They will learn given learning opportunities. They will want to learn if the alternative is 12 hours of testing!

Agreed. A lot of people seem to think that infinite growth is sustainable, even by a single person.

When I was last interviewing, I kept getting asked, "Where do you want to be in 5/10 years?"

When I answered that I wanted to be programming (the thing I was applying for) they all asked if I had no ambition.

I told them that I've already met my ambition. I've got the job I want, and I intend to keep it. I didn't want to be a manager, and I didn't want to do things other than programming.

Of course, it's impossible to do only that, but over 10 years later I'm still mainly programming as my day job. I'm not at a "dead end" in my career. I'm exactly where I wanted to be from the start.