The best information is at the very end of the report where all the tables and graphs are. They show that munition workers are at peak production at around 50 hours or so. There are many many caveats for drawing conclusions from these results:

  * The collected data is from 1915-1920
  * The collected population is very very small 
    (most sets are less than 50 individuals, one is ~100)
  * The work performed by the population is extremely repetitive manual labor

With these caveats, it is safe to say that most of the audience of HN has nothing to gain from the results of this study.

Empirically, my best output p/hour will peak with 35h/week. More than 40h/week and I start to feel overworked. I did also manual labour and the feeling wasn't very different. The difference is that instead of feeling mental tired you feel mental bored.

I'll respectfully disagree.

The fact that the data is from 1915 - 1920 does not invalidate it unless you want to suggest that human nature in 1915-1920 is different than it is now - which would seem a strange point to defend. And the sample sizes are quite decent.

I know that the HM userbase had a tendency to dismiss studies that are incorrectly conducted, but I don't think that this is one of those cases.

The life and the work back then is very different from the life and work that we have now. A big factor is that there is no war to motivate people to work hard. Another is that labour circumstances are very very different now than what they were previously.

Finally almost nobody who reads HN does repetitive manual labour. Manual productivity is very different from mental productivity.

From my own experience, I'm not even productive for 40 hours per week. Let alone working 50 hours consistently every week.

"Human nature". Alternatively: People who think the data is not as pertinent today or in other lines of work believe that human behavior is largely determined by context, not a priori hardwired behavior.

Very well put. Hard to extrapolate factory work to office work.

I do believe that there are diminishing returns in software development (and startups) but a lot of it depends on the type of job, stage of company, existing support processes, etc.

TLDR: Productivity per week is a linear function of time spent working, up to 48-49 hours, where it starts to fall off.

It is TLDR. But I am up voting it because it shows how long people have known that human productivity declines after 50 hours of work a week.

I'm curious as to the effects of extending past the 40-49 hours in week one, on subsequent weeks.

That information I think would be more compelling to an employer (or self-employer), because without it, you're simply saying 'You'll get more out of overworking people, you just won't get as much more.'

Which still reads like "more."

Repetitive manual labour seems as good a proxy as any for productivity. It would be nigh on impossible to find such a robust index for software development work, but other modern professions, particularly repetitive ones, e.g. administrative positions, may be an interesting population to study today.

Software specifically is too variable to accurately measure productivity. It's too subjective. Not only from a personal point of view, but also dependent on the task. Given the tendency towards automating the boring tasks, it seems that conducting such a study in the software development world becomes less and less likely.

In other words, our typical working hours are supposed to get most out of us, or close to :)

I've found I can be productive for around 4 years then I get bored of my surroundings, agitated, can't focus, etc..

ok but this is before modern worker amplification technology like intermittent wifi outages and slack.

One can argue that frequent interruptions from these sources gives workers the opportunity to 'micro-recharge'.

Or to lose focus for the rest of the hour/day.

How is it that some people can spend e.g. 100hours a week in an MMO getting all kinds of things "done."

It is also less productive past the initial 50-60 hours