Yesterday, having gotten far too little sleep the night before, I decided it would be a good time to work on local file import for Gas Can, the publishing tool I've been working on. (And that you're reading this on, if all goes well.)
It didn't go as easily as I wanted it to. I thought that Flexmark would cooperate with me, and I ended up fighting hard with it. But in the process, I found myself in a great example of a spike — a long period of intense focus on a single clear problem, in a tight feedback loop.
In some ways, this was awesome and fun. I built tools to reload my test content and see the result (or failure) that I wanted to see, try out different solutions, and analyze the library I was grappling with.
And in other ways it was kind of a disaster. I was completely wrapped up in the work. From 5pm to 10:30pm, I kept rolling on the ice cream I ate in the afternoon and didn't eat a proper meal. I didn't get up from the computer or take a moment to think about anything except the problem — I spent the whole time in front of emacs: clicking on things, keying in C-c C-e over and over, and cursing deep inheritance hierarchies. (Maybe that last bit was time well spent, actually.)
So what's good and bad about this? Let's start with the good:
I did solve the problem! This in spite of the fact that it was mired in a bunch of code that I didn't like, that I found to be overcomplex and obscure. Rage, rage, rage, but — it works.
Focus loops are a powerful tool. The solution isn't guaranteed to come, but if it can happen, it probably will.
For better or worse, when I get into something like this the spirit of the work possesses me. I forget about eating, my personal life, work anxiety, and anything else that might be occupying spare cycles in my brain.
It's wonderful. Every moment spent not thinking about myself is bliss.
The bulldozer of a good focus loop will, for good or ill, crush everything in its path. If there's some obstacle between me and my objective, I'll find something that helps me get past it.
I was struggling to make an object with a huge inheritance hierarchy work my way. As a result, I built a pile of tools for investigating huge inheritance hierarchies. I didn't plan on building that, but I did and now I have it. Nifty.
It's not all good, though. What's bad:
Laser focusing on the problem is wonderful, but what about the organization of the code? What about blind alleys, things I tried that didn't work?
If I had thought about those things, I wouldn't have been able to lose myself in the tight feedback loop in the first place. That's why the loop works: the whole thought process can focus around one thing, discarding every other concern.
Those other concerns don't go away, though. When the goal is reached, I wake up with a little coding hangover: "What did I do today, anyway?"
Focus changes the way one thinks, as does frustration. Lateral solutions superior to the tool at hand will not present themselves, and I do things the hard way.
Last night I spent a lot of time hammering away at my problem by applying a variety of other techniques I had previously used successfully, and by fuzzing the object by calling methods and seeing their results. I was so frustrated by the way they had chosen to architect their solution that I never bothered to dive into it. Now that I have the solution in hand, though, I can see that slowing down and understanding the library would have paid off more quickly than hacking away at it.
So what should I learn from this?
The core problem here is with the tight feedback loop: laser focus highlights one thing, and ignores everything else.
So when I have spent a long time in a tight feedback loop, next time I'll do some higher-level reflection. Odds are good that I won't remember everything I did in that focus time until I take some time to unpack that work.
Assuming that the goal of the loop was achieved, the most important questions to answer are these:
Focus loops are powerful and fun, but they are a commitment. I'm going to dive in on one end, and come out on the other side with... something. Like Indiana Jones, I might finally get the Ark of the Covenant, but maybe not in the way I wanted.
Life is driven by such quests. But if life or a project is pulled along by a neverending chain of targets of opportunity, where will I end up? Certainty is not achievable, but participation is.
Every evening I take some time to ask, "How am I feeling? How was my day today? What did I do?" And every morning when I sit down to work, I ask "What do I plan on doing today? What did I do yesterday?"
These are broader loops of reflection. For me, if they're not habits, they don't happen. I have to write them down, too, whether I read them or not.
I also know from experience that if I don't engage in these reflections, the tight loops will eat me alive. I won't eat, I'll neglect my friends and other personal relationships, and so on. And I'll never be able to address #2 effectively, either.
Even if I'm plugged in to a standard corporate feedback cycle and am happy as a clam, the loop I'm in serves some other purpose. Someone built that loop to serve that purpose. Therefore, if I do not maintain my own independent loop of reflection, I will be utterly dependent on that purpose.
Oh, and one last thing: for heaven's sake, take it easy. These techniques shouldn't be a cudgel, but rather an inspiration. Take what you like from them, and leave the rest.