Maybe it's because there's a lot of confusion created by people who claim that there's an inverse correlation between hours worked and overall productivity, instead of hours worked and marginal productivity. I have no trouble believing that (on average) someone who works 50 hours doesn't accomplish 25% more than someone who works 40 hours. I find it almost impossible to believe that they don't accomplish more overall though.
And this fits the data: why would companies not dramatically push people to work fewer and fewer hours for this magical increased productivity that you're claiming? They could market it as a perk AND get so much more done!
What I think is happening: there's little cost to companies to encourage workers to work 60 hours instead of 40. Sure, they won't get 50% more done, maybe only 20% or 30%, but the company doesn't pay overtime, so that's a free 20% or 30%.
Of course, there's some point at which overwork does start to impact overall productivity vs standard-40-hours, but it's much, much higher than 40 hours. This study  seems to indicate that little extra gets done from 55-70 hours, but from my reading, that still doesn't indicate that someone working 70 hours gets less done than someone working 40 hours. Looking at figure 4 in that study, it looks like you'd have to go to 90-100 hours to see output drop back to 40-hour levels.
It just seems ludicrous to me to suggest that for the population at large, working 40 hours is the optimal amount, such that anything less will result in greater overall productivity, and anything more will lower overall productivity. Really?
I'd retort: do people still think that working 80 hour weeks makes you more productive than working 50 hour weeks?   
In a startup, your competitive advantage is rarely going to be how many hours you work - how can it be, when you're playing against large well-capitalized companies that can assign entire teams at the drop of a hat and thereby outwork you tenfold? Rather, you rely on some combination of:
- underdoing the competition (you can't do everything as 1-10 people, so you have to fail and iterate quickly with minimal effort and learn as much as you can from each iteration); - personal / professional connections (OK, so you can spend more time here, but it's often not going to be behind a computer or at the office, and it will be very hard to measure productivity-wise); - blind luck (which isn't in your control anyways, so don't waste your time trying); - the innovator's dilemma (which isn't really in your control either, but gives you a several-year headstart while larger companies slowly figure out that your market is lucrative enough for them to bother with).
Finally: from personal experience, the productivity gains of working 80 hour weeks are illusory. I make more mistakes, incur more technical debt, work on fewer of the right problems and more of the wrong ones, take on stress that impacts my decision-making and communication skills...