My job at Bungie these days is about one core piece of technology we use to make our games. It's a system that all the artists, designers and sound designers use to get their work into the game. So, like any technology job where you're building a product other people use, there are two kinds of work: the work of building the technology and the work of supporting the people using it.

Things have been busy enough lately that I was trying to focus just on the coding and not do much support work. It wasn't really working out. All around me the artists, designers, and sound designers were struggling with issues while I was trying just to plow ahead on the coding work we need to get done. When the support work boiled over and I had to stop and attend to it, it took a ton of time and effort. Because I'd been burying my head in the sand trying to get code written, I didn't have any context for what was happening with the users. If I had to help someone, first it took me a good half-hour of asking random people questions just to have enough context to even understand what the problem was and what needed to happen.

I couldn't ignore the support work, though. Eventually I had to turn my attention towards it, and so I did, one week, and it consumed basically my entire week. I got just about nothing else done. It was frustrating seeing the coding work pile up, so my response was to keep trying to minimize my involvement in the support work. I'd do as little as I could get away with, try to hand off problems to someone else. Bu it didn't seem to actually help. It still took a ton of time just to figure out enough about a problem to know how to hand it off to someone else, and if anything that ramp-up time was getting longer.

* * *
If you've ever watched the Tour de France, or the road biking events in the Olympics, or ever seen any group of people riding road bikes together, you've probably noticed they tend to cluster together in a compact formation. They're not just riding close together because they like the company (and the increased risk of crashes and injuries.) No, they're riding close together because of drafting.

Drafting is a physical phenomenon that boils down to this: It's a lot easier to bike if you're right behind another biker. How much easier depends on a bunch of factors: whether there's a headwind, whether you're on flat ground or biking up a hill, how smooth the pavement is and stuff like that, but overall, drafting makes a huge difference. The guy "pulling" - riding in front - might be doing about twice as much work as the guy drafting. It's in that ballpark.

So let's say you're riding with your friend, who is in better shape than you. He pulls so that you can draft, and you both get a workout. That's a nice thing about biking: people of somewhat different fitness levels can ride together because the fitter person can pull. Anyway, here you are: he's in front in the green shirt, and you're casually spinning your pedals behind him in your sweet Hawaiian shirt, and because of the physics of drafting, you're not even working that hard.

Everything's well and good until he speeds up a little and you don't notice fast enough and suddenly you're several bike-lengths behind him: you've been "dropped." You've lost the drafting boost, and life just got hard:

When you're far away from him like that, you're having to do exactly as much work as he is. You don't benefit from drafting at all. When you've been dropped, therefore, your first priority is to get back on his wheel. The problem is, you're not strong enough even to match his pace for very long, and to get back on his wheel you need to ride even faster than him. You're quickly tiring as you fight the wind, so what do you do?

When this happens to me, a lot of the time I'll catch myself trying to ease my way back onto his wheel gently, just giving a little bit more so I don't exhaust myself. In a way it feels intuitively correct: this way I smooth out the effort and don't exhaust myself by the time I get there. But that intuition is wrong! What I should do is push, hard, to get back on that wheel as quickly as I possibly can. It's about the last thing I want to do in that moment as I'm already struggling and tired, but it's the most efficient thing to do. I need to really push and sprint back to catch him.

What's wrong with the smooth, gradual approach? Let's talk about assumptions. The gradual approach assumes, implicitly, that doing a portion of the work yields a portion of the benefits. Maybe I'm assuming it's linear, which would look like this:

That graph describes a world where closing half the distance between me and him gets me half the drafting benefits. So there's no huge hurry because as long as I'm closing the gap, it's getting easier. I still end up doing a bit more work but I never have to really kill myself so it's probably worth it. The gradual approach would absolutely be the right one if the effort curve looked like this:

That graph describes a world where closing the gap even a little bit makes things hugely easier, and there's a big envelope towards the end where the drafting benefit is almost entirely maxed out.

I don't actually know what I'm thinking when I catch myself closing the gap gradually. I'm pretty sure I'm not imagining that second curve, but if you asked me if it worked like the first one, I'd say no, because I know better. I think the key is, when I'm using the gradual approach, I'm not consciously thinking about the effort curve.

Drafting doesn't work like either of those curves. It's actually hard to find concrete data about the exact physics, but I found a wind tunnel study that tested drag coefficients on the drafter where he was anywhere from zero to seven bike lengths from the leader and the conclusion was, the drafting benefit falls of very sharply. I actually suspect there may be a point where the turbulence makes it's even harder to pedal than normal, but I couldn't find clear evidence of that. In any case, the curve looks something like this:

What's that mean in English? Drafting is an aerodynamic phenomenon, a result of your friend pushing the air out of the way as he rides through it. He and his bike moving through the air are an example of turbulent flow, which leaves a slipstream behind him, a pocket of air that is moving forward with him, at a lower pressure than the surrounding air, and shielded from the wind he's riding into.

This pocket is not very large, and as you exit this pocket the slipstream benefits fall off very rapidly. As you leave that sweet pocket of magical drafting air, the turbulence kicks back in and hits you pretty abruptly. Here's how that looks:

What this means is if you gradually ease yourself back onto his wheel, if it takes you 30 seconds to do it, for about 28 of those 30 seconds you're pedaling at full difficulty. You're not doing yourself any favors.

That's why the right thing to do is power through the hard part so you can get to the small envelope and slipstream again. You give an initial push so that you can establish yourself, after which you can take it easy again.

* * *
Back to my work situation. After a week of struggling with support work consuming all my time, I realized my situation was a lot like bike drafting, and I was making the mistake of trying to ease my way into it. I was sitting in that turbulent zone: I was bearing the full brunt of the support work, but because I was trying invest as little as possible each time I did it, I had to pay the full price of contextualization every time. What I mean is, I'd get an email about support, and I'd read it and I'd have to figure everything out from square 1. I'd think, who is this person? What is he doing? What does that mean? What is this problem? Has this happened before? Do we know anything about it? Where would I even start trying to fix it? Maybe two hours later I'd find out that it was a known bug someone was already working on and if I'd had enough context I could have immediately said so and gotten back to work.

After burning a whole week that way, I got wise. Monday rolled around and I pushed. I spent the whole day working to get myself up to speed so I understood the context for all the support work. I worked with a friend in our user research team who built me a "dashboard" webpage I can look at that tells me at-a-glance about everyone who's suffering from crashes, failures, or excessively long wait times. I subscribed to email notifications for some of the users who tend to have the most problems so I know what they're up to and can respond to their problems quickly. I spent time talking to everyone involved to understand their world-views, so when I saw an email from any of them, I'd know how to interpret it immediately.

And with that one day of push, I broke through the turbulent wake and found myself in the slipstream. When support issues came up, I could glance at them and immediately contextualize them: "Ah, this is the fourth time he's had this crash today, and three other people have hit it, but it only just began today, and yesterday I know these three branches got integrated back into the one he's working in." It's an accelerator, and it's self-sustaining: by continuing to do this support work, I keep my contextual knowledge up to date and it stays easier. Just like you have to keep pedaling when you're drafting, just enough not to get dropped.

I'm trying to use this metaphor anytime I start something new, now. If I catch myself trying to devote just a little time every day to the new venture, I stop and ask, what does the effort curve actually look like for this work? Is this the kind of thing that I can ease into, that I can adopt gradually? Or is it the kind of thing I need to start off with a big, initial push?

Usually when I'm trying to do something gradually it's because I don't feel I can commit a big chunk of time up-front to it - just like when I'm doing the gradual thing on my bike it's because I'm tired and don't want to put out a bunch of energy all at once. In both cases I have to stop and remind myself that the gradual approach may feel better at first but even in the short term is much worse. And in those cases the only two viable options are to make the decision to push - to carve out the time or energy for it immediately - or to admit I'm not going to do it at all. Either of those can be the right decision, depending on the situation. The import part is recognizing when the work demands an initial push, and not make the mistake of trying to do it gradually.