Being stuck is possibly the worst state to be in when programming. And the longer you allow yourself to stay stuck, the harder it is to finally get unstuck. That’s why I find it important to try never to allow myself to get stuck. And in fact, I very rarely do get stuck for any meaningful length of time. Here are my favorite tactics for getting unstuck.
Articulate the problem to yourself
I teach corporate training classes. Often, my students get stuck. When they’re stuck, it’s usually because the problem they’re trying to solve is ill-defined. It’s not possible to solve a problem or achieve a goal when you don’t even know what the problem or goal is.
So when you get stuck, a good first question to ask is: do I even know exactly what the problem is that I’m trying to solve? I recommend going so far as to write it down.
If it’s hard to articulate your goal, there’s a good chance your goal is too large to be tractable. Large goals can’t be worked on directly. For example, if you say your goal is “get to the moon and back”, that goal is too big. What you have to articulate instead are things like figure out how to get into space, figure out how to land a spaceship on the moon, figure out how to build a spacesuit that lets people hang out in space for a while, etc.
Just try something
Far too often I see programmers stare at their code and reason about what would happen if they changed it in such-and-such a way. They run intricate thought experiments so they can, presumably, fix all the code in one go and arrive at the complete solution the very next time they try to run the program. Nine times out of ten, they make the change that they think will solve the problem and then discover they’re wrong.
A better way to move forward is to just try something. The cost of trying is very low compared to the cost of reasoning. Also, trying has the added bonus of supplying empirical data instead of untested hypotheses.
Articulate the problem to someone else
“Rubber duck debugging” is the famous tactic of explaining your problem to a rubber duck on your desk. The phenomenon is that even though the duck is incapable of supplying any advice, the very act of explaining the problem out loud leads you to realize the solution to the problem.
Ironically, most of the “rubber duck” debugging I’ve done in my career has involved explaining my issue to a sentient human being. The result is often the classic scenario where you spend ten minutes explaining the details of a problem to a co-worker, realize the way forward, and then thank your co-worker for his or her help, all without your co-worker uttering a word.
The explanation doesn’t have to be verbal in order for this tactic to work. Typing out the problem can be perfectly effective. This leads me to my next tactic.
Put up a forum question
In my experience many programmers seem to be hesitant to write their own forum questions. I’m not sure why this is, although I have some ideas. Maybe they feel like posting forum questions is for noobs. Maybe they think it will slow them down. Neither of these things are true, though. I’ve been programming for over 20 years and I still post Stack Overflow questions somewhat regularly. And rather than slowing me down, posting a forum question usually speeds me up, not least because the very act of typing out the forum question usually leads me to realize the answer on my own (and usually about 45 seconds after I post the question).
If you don’t know a good way, try a bad way
Theodore Roosevelt is credited with having said, “In any moment of decision, the best thing you can do is the right thing, the next best thing is the wrong thing, and the worst thing you can do is nothing.” I tend to agree.
Often I’m faced with a programming task that I don’t know how to complete in an elegant way. So, in these situations where I can’t think of a good solution, I do what I can, which is that I just move forward with a bad or stupid solution.
What often happens is that after I put my dumb solution in place, I’ll take a step back, look at what I’ve done, and a better solution will make itself fairly obvious to me. Or at least, a nice solution will be easier for me to think of at this stage than it would have been from the outset.
With programming and in general, people are much better at looking at a bad thing and saying how it could be made good than they are at coming up with a good thing out of thin air.
If you go down a bad path and find that you’ve completely painted yourself into a corner, it’s no big deal. If you’re using version control and using atomic commits, you can just revert back to your last good state. And you can start on a second attempt now that you’re a little older and wiser.
Study the hell out of the problem area
There’s a thought experiment I like to run in my head sometimes.
Let’s say I’m working on a Rails project and I can’t quite figure out how to get a certain form to behave how I want it to behave. I try several things, I check the docs, but I just can’t get it to work how I want.
In these situations I like to ask myself: “If I knew everything there was to know about this topic, would I be stuck?”
And the answer is of course not. The reason why I don’t understand how to get my form to work is exactly that—I don’t understand how to get my form to work. So what I need to do is set to work on gaining that understanding.
In an ideal world, a programmer could instantly pinpoint the exact piece of knowledge he or she is missing and then go find that piece of knowledge. And in fact, we’re very fortunate to live in a world where that’s possible. But it’s not possible 100% of the time.
In those cases I set down my code and say okay, universe, I guess this is how it’s gonna be. I’ll go to the relevant educational resource—in the case of this example perhaps the official Rails documentation on forms—and start reading it top to bottom. Again, if I knew everything about Rails forms then I wouldn’t have a problem, so I’ll set out of a journey to learn everything. Luckily, my level of understanding usually becomes sufficient to squash the problem long before I read everything there is to read about it.
Write some tests
One of the reasons I find tests to be a useful development or debugging tool is that tests help save me some mental juggling. When developing or debugging there are two jobs that need to be done: 1) write the code that makes the program behave as desired and 2) verify that the desired behavior is present.
If I write code without tests, I’m at risk of mentally mixing these two jobs and getting muddled. On the other hand, if I write a test for my desired behavior, I’m now free to completely forget about the verification step and focus fully on coding the solution. The fewer balls I have to mentally juggle, the faster I can work.
Take a break
One of my favorite Pink Floyd songs is “See Emily Play“. In the first chorus they sing “There is no other day / Let’s try it another way.” In a later chorus they sing “There is no other way / Let’s try it another day.” This is great debugging advice.
I often find that if I allow myself to forget about whatever problem I’m working on and go for a walk or something, my subconscious mind will set to work on the problem and surface deliver the solution later. Sometimes this happens 15 minutes later. Sometimes it doesn’t happen until a day or a week later. In any case, it’s often effective.
Work on something else
If all else fails, you can always work on something else. This might feel like a failure, but it’s only a failure if you decide to give up on the original problem permanently. Setting aside one goal in exchange for making progress on a different goal is infinitely better than allowing yourself to stay stuck.
Agree 100% — In the shower the next day is the usual Eureka moment.