I also really like the attitude that if you're not touching code, you're not doing real work. Sure, project managers etc. will say that your job is not solely to write code, and that responding to emails, participating in meetings with your teammates, etc. are as much part of your job. But I like the simplicity of "if you're not writing code, you're fucking off" and how easy it makes it to answer the question "Did I work today?".
On the other hand, programming all day is sometimes a recipe for not learning. It might be better to spend 20 minutes learning an elegant solution rather than 5 minutes implementing a hack. But those 20 minutes are mostly spent reading rather than writing.
I view thinking about the problem, researching the problem, etc, as being indivisible from the coding aspect.
Really, it's telling how -I- feel after a day spent in meetings. It feels wasted. I'm grumpy that I achieved nothing.
Whereas a day without any meetings, where it's head bent down, writing code, even when it involves research or looking up API docs, or going to ask questions of domain experts when interacting with other systems...feels productive. Even when some of it changes because the customer comes back and says "No no no, it should work like (X)".
Honestly, it feels more productive when the customer can instantly tell me what is wrong after the fact, than if I spend an hour up front trying to pick their brains, to then code it, and to -still- find out it's wrong. Per another thread we had here, people find it much easier to tell you what is wrong than to tell you out of the air what is right.
Just like with the stock market, watching gains and losses with too fine a granularity will make you tense up and "buy high and sell low."
If you're "going long" on your productivity, collect all the data you like--but have a smoothing function between that input and the resulting metrics. That way, 20 minutes spent to gain 3 hours will look like the slight-dent-and-sharp-rise it is, not a scary better-stop-this plunge.
I also feel like it's a really easy way to fall into the trap of only knowing how to write code, which is not what you're actually paid for. You have to interact with the rest of the business and the industry it's in to get good at creating value for a specific domain.
edit: But it's a good point as a general idea, just shouldn't be thought of as an end-all-be-all I don't think.
On the other hand, programming all day is sometimes a recipe for not learning. It might be better to spend 20 minutes learning an elegant solution rather than 5 minutes implementing a hack. But those 20 minutes are mostly spent reading rather than writing.