[gexla]

"A friend of mine got a job at a big company and was shocked to see his colleagues worked just a few productive hours a day."

Depending on what you do, I would say this is natural. I don't care what Henry Ford says (8 hours a day is different for some types of work than it is for others.) I can't go into "super focus" mode for much more than 4 hours per day. I can do my half day of being "wired in" and then take care of lighter tasks for another couple of hours.

There are days when working on an exciting project that I can code for 12+ hours for a few days, but not all of those hours are productive. Typically on these sorts of days I can run into a problem which I bang my head on for a couple of hours and then fix it within 5 minutes the next day after a refreshing sleep.

Maybe part of the problem is that that these places should shorten the work day (but keep payment the same.)

[winter_blue]

John Carmack has written something about being super-productive in a fixed chunk of time.

But I can see where your coming from. After certain set (~5) of being super-productive, you burn out. And you just can't take it any more.

At these times it is best to take a break and do something different. I've found that these breaks (~1 to 2 hours) tend to make you more productive in your next chunk. (Rather than being burn out and weary.) It gives you time to think and churn ideas for what you will do next.

Lastly I want to add; knowing what you want to do next makes a difference. When you know exactly what you want to do (the spec), you naturally are about implementing it - when not knowing the spec, you're figuring things our as you go and that slows you down. In addition it's always better to sit down and gives additional exclusive time to thinking of your program rather than figuring things out as you go.

[espeed]

I can run into a problem which I bang my head on for a couple of hours and then fix it within 5 minutes the next day after a refreshing sleep.

The act of banging your head on a problem for several hours might help to kick the problem into your unconscious mind (your big brain), where the real horsepower is. And while you sleep, your mind sorts it out. That may be why you are able to fix it within 5 minutes the next day.

Rich Hickey talks about this in his talk on "Hammock Driven Development" (http://blip.tv/clojure/hammock-driven-development-4475586).

[randomdata]

I agree that you can only effectively spend a short amount of time focusing on a single task, but if the rest of your day is spent idling, it may be a sign you're not spreading yourself around enough. In software development, that might mean four hours focusing on programming and another four hours focusing on design, for example.

It seems to me that the idea of hyper-specialized employees is often the root of the problem. The brain needs variety.

[EricDeb]

I agree on some level. Certain tasks require sleep, others can be done in deprivation. If I am familiar with a programming language and problem I do okay even if I am sleep deprived, however new tasks or very difficult problems I cannot do well on a lack of sleep.

[eli]

But it's not the days when I "only" get two hours of work done that worry me. It's the days when I can't do anything.

http://www.joelonsoftware.com/articles/fog0000000339.html