Hacker News Reader

I think you might be missing the author's last point which is that in a competitive field, even a small increase in output can lead to a large increase in payoff.

In his example of getting research funding, it's a step function: either you get funded (y=1) or you don't (y=0), and because you're competing with others, even a small output increase can take you from the 0 side to the 1 side.

i've never thought about how these two things relate to each other: asymptotic productivity of work, and winner take almost all nature of capitalism. seems like a very succinct way of explaining why people work so damned much. i'm actually kind of embarrassed i've never connected the two. it seems to make the argument against it more clear too- if marginal hours worked grow less productive, and marginal improvement to the product is negligible in every day terms, we ought to mitigate that somehow, unless its something people are doing because they love it.

One of the greatest examples for this is the Astronaut program requirements [1]:

1. A bachelor's degree in engineering, biological science, physical science, computer science or mathematics.

2. At least three years of related professional experience obtained after degree completion OR at least 1,000 hours pilot-in-command time on jet aircraft.

3. The ability to pass the NASA long-duration astronaut physical. Distant and near visual acuity must be correctable to 20/20 for each eye. The use of glasses is acceptable.

4. Astronaut candidates must also have skills in leadership, teamwork and communications.

Checking all the boxes on that list isn't particularly hard to do. However you aren't competing with the list.

You are competing with a 29 year old Rhodes Scholar with a PhD in Chemistry, who runs Ultramarathons, ran a 40 person volunteer organization in Kenya during her gap year, and manages a 200 person 50 Million dollar research lab on pluripotent stem cells at Oxford.

Now just change out anything there with whatever would be relevant for your line of work, and thats who you are competing with - give or take.

[1] https://www.nasa.gov/audience/forstudents/postsecondary/feat...

The astronaut program is not an example of capitalist winner take all.

Mostly is - specifically around hiring and work, so it's relevant to this discussion. It's a highly competitive market for talent where small incremental differences make a big difference.

To clarify, your point is about the nature of competition. Capitalism is but one type of competition. Hiring for astronauts is certainly a competition for resources, but since it is not a flow of transformable resources among a network of profit seeking nodes, it ain't a market.

Spot on. Perhaps I should have clarified that it's a metaphor in the original post.

That is not capitalism. Socialism could have athletic competitions and rigorous selection for public sector programs as well. That is not inconsistent with public ownership of resources.

To demonstrate something about capitalism, one should use an example from capitalism, not try to contort something from government into capitalism.

Isn't this essentially a winner-take-all situation, though?

Capitalism tends towards winner-take-all systems, but not all winner-take-all systems are capitalistic. I find the example fitting.

Capitalism actual doesn't veer toward winner take all. Government does, elections do, etc. Capitalism leads to people voting with dollars so instead of a red or green car being given a vote, the red car may be cheaper if lots of people want it.

Yeah, that's why there is AMD and Intel, NVIDIA and AMD, Apple and Microsoft, Google and Facebook.

> the red car may be cheaper if lots of people want it.

Also see the pharma industry, right? The US government isn't all governments either, in some countries parties actually rule together, compromise and whatnot. And considering getting money out of elections isn't even attempted in earnest, the question really is what is the tail and what is the dog here.

The US government rules as a big power structure. Now Red happens to be on the pinnacle of that, but that doesn't mean Blue has no say in what happens.

It just means things are a lot more "costly" for Blue and/or Blue things (like climate change policies).

But of course in the primary layer a lot of things become infinitely costly for Blue things. (Because they fall under executive rights.) But there are a lot of other layers. (State legislatures/governors, city level issues, and so on.) And though a higher layer can force things on a lower layer, but it's not zero-cost, so usually things settle into an ugly compromise.

Good example. Elections are definitely winner-take-all, i.e. we get red cars or green cars. But under capitalism, I can choose which I want. Elections are not about individual choices.

elections are winner take all in the literal sense (except where you have representative democracy), but to they extent that they are about policy, i wouldn't say its as simple as that.

It is a winner-take-all situation, but it is not an example of capitalism, and should not be used as an example of capitalism.

Use an example of capitalism to justify an example of capitalism. If capitalism is a winner-take-all system, there should be endless examples.

Great insight! You can see this not only at the macro level (Google winning all of search by being marginally better than competitors), but at the micro level (top employee out of a team of 10 gets the bonus/promotion while everyone else gets nothing), and probably everywhere in between.

What I wonder is: Is this really an intrinsic quality of capitalism? What causes this "winner take most" reward system emerging in some areas of the economy but not others? In many industries, multiple competitors can exist with stability and split market share, and in others, having multiple competitors leads to a blood bath where only one can survive, winning it all.

> Is this really an intrinsic quality of capitalism?

Hardly. The history of capitalism shows that absent government intervention to prop up companies, they rise and fall. When I started out, everyone was a-feared that IBM would inevitably take over the world. Then Microsoft dethroned them, and Microsoft was going to take over the world. Then Apple dethroned Microsoft. And on it goes.

There was Sears that dominated retail. Then Walmart shoved them aside. Now Amazon stepped all over Walmart.

See "The Innovator's Dilemma" by Christensen for a book full of examples.

All of those companies exist in a world with the threat of anti-trust regulation, and some were directly affected by it.

There are plenty of counterexamples from before that regulation, and there are still some today (e.g. ISPs).

> There are plenty of counterexamples from before that regulation

Name one :-)

Literally the entire reason antitrust laws were written? Standard Oil? General Motors? AT&T?

A little disingenuous to pretend that never happened.

Let's take Standard Oil. They never had a monopoly - at their peak they had a 90% market share. SO's market share slid steadily throughout the years of the anti-trust trial, and Rockefeller was unable to stem the slide. (The competition figured out how to beat him.)

See "Titan" by Ron Chernow

> pretend

I know what the popular wisdom is. But dig a little deeper and you'll find it is not correct.

> General Motors

was formed in 1908. The Sherman Antitrust Act was in 1890, so GM could not have been the reason for it.

> GM could not have been the reason for it

...which, to clarify, was not my point.

That is, going back up-thread a bit, it's still clear that capitalism intrinsically encourages monopolistic behavior, and that "government intervention propping up companies" is not the cause.

In fact, I would go so far as to say that even if there were no capitalistic tendency toward actual monopolies, it would still have the same winner-take-all problem, merely due to the prospect of a "winner".

Every company wants to be a monopoly, sure. But free market forces prevent that from happening. It takes government to prop one up.

A prospect and an actual problem are two different things.

SO was never convicted of being a monopoly, either. A subtle but crucial point, and usually overlooked.

I'm not sure on what basis you're claiming that free markets prevent monopolies, but even if they do, free markets self-destruct. Once corporations in a free market have enough money to start influencing politics they use it to start regulating in their favor and the market ceases to be free. You can regulate to prevent that, but then the market isn't free, yet again.

It's also a bit myopic to view companies as a binary monopoly/not monopoly. A company with 90% market share isn't a monopoly, but has most of the downsides of a monopoly, and likely is a monopoly within subsets of the market.

Free market forces don't prevent it from getting close enough to cause a problem (if that's even possible under capitalism- the problem we're talking about is due to the attempt to gain a monopoly, not having one). I couldn't care less what Standard Oil's maximum market share was, you're arguing an irrelevant technicality.

My point is, again, that in this case the prospect is what causes the problem.

> cause a problem

The "problem" was that kerosene prices dropped 70% under SO.

The real problem was that Rockefeller was rich, and then (like today) a lot of people just can't stand the idea that others are rich.

If you're really interested in going beyond soundbites and understand what was happening, I really recommend Chernow's "Titan". Chernow actually agrees with you, so you can't argue his book is unfair. Chernow's opinions, however, are not congruent with his presented facts in the book. He's just so convinced of Rockefeller's evil nature that he is blinded by his own narrative :-)

> The real problem was that Rockefeller was rich, and then (like today) a lot of people just can't stand the idea that others are rich.

dont you think its kind of upsetting when someone could improve someone's life drastically without any (practically speaking) detriment to their own. obscene wealth is upsetting to me, and i know lots of people who have it, and i dont hate them. so its not about the person, per se. i just find it extremely unfortunate, knowing first hand how things could be different, in an almost pareto improvement way.

right now there is someone who will die because they cannot afford proper medical care. the is almost certainly someone with a lot of money, doing nothing with it, that could fix that at no cost to themselves. thats a little upsetting right?

Neither of those are the problem I'm referring to, which should be obvious based on the thread we're in...

But to spell it out, the problem is that the winner-takes-all aspect encourages people to overwork themselves or their employees, despite the diminishing returns that brings.

Maybe you're just so convinced of capitalism's good nature that you were blinded to what we were talking about in the first place? ;)

I think you're inventing a problem. I don't know many capitalists or employees who believe they must work themselves to death because of some presumed winner-take-all aspect. In fact, I don't know any.

Few define business success as erasing all competition. Success is making a good profit.

> Let's take Standard Oil. They never had a monopoly - at their peak they had a 90% market share.

Out of curiosity in calibrating your definition of "monopoly", can you give an example of a monopoly (past or present) together with an estimate of their market share? Thanks.

Sure. The USPS has a monopoly on first class mail. Market share = 100%. The government prosecutes anyone else who tries to do it.

Up until a couple years ago, in Washington State, hard liquor could only be sold through state stores. 100% market share.

The plumber's union is a monopoly. It is illegal to hire a non-union plumber. 100% market share.

Anyone holding a patent, trademark, or copyright has a monopoly. 100% market share by law.

Are you aware that federal law uses a different definition of "monopoly"?

Isn't Christensen one of those thought leaders who are paid by megacorps to spread Gramscian false consciousness and make the masses think capitalism really isn't so bad?

Yes, facts don't matter, only the political pedigree of the person relating them.

I prefer to read books where the author tries to prove me wrong. Books that agree with me are boring :-)

Sounds like you already have your conclusion in mind, so why ask the question?

Is this irony?

I suspect that in industries with strong network effects, the winner-takes-most system is most prevalent, as say, one additional unit could mean stealing three units away from competitors in the long-term.

yes i wonder about that too. i never thought of myself as someone interested in economics, but i guess thats where you'd look for answers to why some markets are weighted towards winner take all and some aren't. my guess off the top is that it has to do with how easy it is to switch between two competing products, and how economies of scale work in the industry. if there is a big economy of scale, and its easy to switch between competitors, i think that makes for the biggest winner take all scenario.

I think the more fundamental connection is that when goods are perfect subsitutes for each other, under perfect information you always choose the "cheaper" good (where "cheap" is defined by whatever metric you're using).

Somewhat ironically, this fact actually has connections to the concept of dual variables in constrained optimization problems, the theory of which was largely developed in the Soviet Union for use in central planning.

very interesting! that seems to make sense. do you know of any good writing on examples?

asymptotic productivity of work, and winner take almost all nature of capitalism. seems like a very succinct way of explaining why people work so damned much

A myth of capitalism is that the donkey can reach the carrot hanging from the stick if they work harder, that "almost all" can be won by those who are being driven to work more hours. However, this doesn't take into account that there are at least as many carrots-on-sticks as donkeys, and that carrots are cheap.

Well personally I've come from a family where hardly anyone make a productive decisions, where nobody had a college degree, and a tiny portion are even middle class. And now I'm 22, self-taught, and make $70k a year in the Midwest. And I did get there by working endless hours and catching the carrot, repeatedly. I think the truth of how these things go is probably somewhere between where your typical anti capitalist and your typical social Darwinist would expect it to be.

>I think the truth of how these things go is probably somewhere between where your typical anti capitalist and your typical social Darwinist would expect it to be.

What isn't between those two poles?!

Fewer people than what would be reasonable

>A myth of capitalism is that the donkey can reach the carrot hanging from the stick if they work harder

That's the myth of the "American Dream", not Capitalism per se.

Capitalism is more simple, if you have capital, you have control, and with control you can profit.

neither is true. Having capital means you will blow it a good portion of the time. Control is illusionary, just as it was for Sears, IBM, and Microsoft.

i dont get what you are trying to say. something about how its easy to mislead people into working long hours?

In this context, that "the nature of capitalism" is not evenly distributed, though the rhetoric says it is.

How much of winner-take-all is due to other factors like timing, luck and self-selection-bias?

The point your predecessor was making was that if you make bad design decisions or create bugs, your output isn't growing.

So design while you're fresh, and do the grunt work when you're tired.

And I find (have found) that I'm much more inclined to do the boring things when I've finally exhausted my curious/creative mental faculties.

Recognizing that a problem is hard enough that you shouldn't solve it tired is hard to do tired.

There's a ton of conflation going on in the article. Most importantly, what defines "productive work" versus just "output" is not defined. He seems to focus on "academic" work but doesn't define terms well.

I've written about this here before, in summary:

1) Work has a cost to the endeavor too. Even with salaried workers.

2) If what you're doing can be boiled down to time-and-motion studies, you can always make cases where longer hours yield more "output". BUT pushing even these types of work beyond marginally has been found to incur large ongoing cost the mission. This is backed by research that has been done many times since WWII with thousands of subjects.

3) Then, what software developers do is NOT amenable to time-and-motion studies. So even the marginal increases in time working has an even greater case of dubious value.

TL;DR: MS Paint [0]

[0] http://imgur.com/a/YrlD7

That graph doesn't reflect the gist of the comment you're replying to. It explicitly talks about negative productivity, not asymptotically zero output like the graph.

Your reasoning can be applied to a regular job: if I am tired I make more mistakes and every mistake has a cost. If the cost of mistakes made during the extra hours exceed the value of the job done, those extra hours didn't produce anything useful.

Even not taking into account mistakes, there is another important factor to consider: you need to recover energy from one day to another. The more you work the more you need to recover and less time you have to do it. If you don't, you will be less productive the following day.

I think that for this reason it is reasonable to say that there is a limit after which the productivity start to drop, though it probably depends on the individual and on a number of factors.

Is this really so? Perhaps in the long term 40-hour-a-week pickers pick more than 60-hour-a-week pickers in absolute terms simply because the extra recovery time helps them be fitter and have more energy.

Apple picking is like crunch time – or at least like crunch time should be – it's not a long-term activity: the time between apples ripening and the first frost ruining them is not long.

Semi-true; it depends how the extra 20 hours are spent for myself. I've had weeks where I program 60 hours or more and it's all spent well, and then the next week I might be a little more burnt out.. take it a little easier and then sprint again.

Which goes towards 60 hour weeks are not sustainable...

Which is a good reason not to work 60 hours regularly over the long term, but not a reason to never put in a 60 hour work week or even a few in a row.

Elsewhere, I've written how it may be better to switch between periods of very long hours and downtime rather than a steady pace:

https://metarabbit.wordpress.com/2014/02/26/average-work-wee...

I agree. This leads to burnout and it is not a pattern for a lengthy or successful career. I had to recently take time off in order to recharge from burnout... avoid getting to that point at all costs!

A have a couple friends who work dawn until dusk for their startup. After several months they struggled with depression and managing their work life. After a little over a year they've adjusted and it's just routine.

> After a little over a year they've adjusted

You mean they developed depression and gave up having a healthy work-life balance?

I can sympathize, as this recently provoked myself to take a vacation for my mental health to recover. I went to an island with horrible internet and phone service and disconnected.. this helped me restore the fragile work/life balance I require. ^

60 hours weeks aren't sustainable, but just look at teams during crunch weeks. They sure get a lot of work done.

Sometimes you've just got to put the hours and effort in to get the work done on time.

Now I quit very quickly after doing stupid hours a week. But I'd like to think the work produced was fine. Not that much different to a normal week. The work was all planned in detail, there weren't any big surprises, just lots of code to type in, lots of tests to write and run. I was just more tired.

How much of that extra work was because of the extra hours put in and how much was because people work harder when deadlines are close?

I was talking with a colleague the other day about "strategies for how to focus for 4 straight hours".

The one that works best for me is for the deadline to be 4 hours and 15 minutes away.

That's a popular adage: "I have a motivation problem until I have a time problem."

At my previous work, we attempted to find the correlation between faulty code and the time of commit in the version control. Most bugs were written after 14:00. (And no: we didn't check-in all code at the end of the day).

A fellow student of mine did a study back in 2005 about this. He found that the worst time to fix a bug was Friday afternoon, as it introduces the most (new) bugs into the code.

I see similarities to another area where I have more experience: business meetings. I believe that most people are terrible at meetings held in the afternoon, either due to heavy lunch, or just because it's after 3pm and they are simply tired. A single meeting like this can cause great damage to the individuals and whole teams, measured in intangibles such as motivation, trust, momentum.

Is there something peculiar around 14:00 beside lunch ?

> This is self-evidently true for many jobs. > E.g., if you pick apples for 40 hours, > then even if you only pick one apple per hour > after that, you will still pick more apples > working 50 hours than 40.

That logic presumes that the first 40 hours aren't affected by the additional hours. IIRC, one of the premises of the "sustainable pace" rule for Extreme Programming was that people know when they are being pushed to work a lot of hours and either consciously or unconsciously slow down at the outset (with the mind or body shifting out of a sprint-mode and into a marathon-mode).

It seems plausible that programming while tired doesn't do a whole lot of damage in itself (more than the usual amount of bugs), provided that you also write unit tests and do code review. This could include self-review: that is, don't check in while tired, and read it over the next morning.

On the other hand, the shortcuts people take while under deadline pressure seem more damaging, and people are more likely to work longer hours when under time pressure. And the cumulative effects of working while tired might result in worse results the next day, even if a single day doesn't do it.

But I'm just speculating. It would be interesting to see a study that tries to distinguish between these different effects.

> provided that you also write unit tests and do code review

Not necessarily. Imagine submitting a patch on 8 PM, when everyone has already been in the office for 10 hours. It has a glaring bug in a particular branch that's not covered by the unit tests in a way that coverage testing doesn't show, yet everyone is too tired to see that bug. Thus the patch gets greenlit and applied to production. Just as everyone is heading home to fall into bed, pagers start ringing because prod is on fire.

> It seems plausible that programming while tired doesn't do a whole lot of damage in itself (more than the usual amount of bugs)

I can tell you in embedded devices that simply isn't true.

In a lot of embedded programming, you can cause real damage--short out boards, crash equipment, etc.

If I have been at something for too many hours, I may simply stop. I will certainly stop if I recognize that I just did something really boneheaded and simple. That means that I'm likely to continue doing boneheaded things and cause real damage.

I doubt that the bugs provoked by tiredness have an aggregate of negative productivity.

And even if so, it can be mitigated. As a webdeveloper, when Im tired i do the most menial tasks which means that bugs introduced there are rare and not important.

What is a menial task in software development?

I mean, our entire field exist for pushing menial tasks into computers. If you have as many of them to fill 50% of your work load, there's probably something wrong.

Besides, are you sure bugs in menial tasks are not important? When did you judge those tasks to decide this?

So many: fleshing out test suites, setting up that build server, automating your documentation generation process, making that internal tool a bit more generic to be useful, fixing those warnings from that analysis tool, checking for performance regressions against the data from when you last did it 3 months ago, refactoring some of that code from 3 years ago for which there is now a more idiomatic way of doing it, issue tracker triaging, ... There is so much in software development that is menial yet isn't worth or can't be automated.

Just talking about the first item: you're asserting that first verifying your understanding of specifications, discovering then specifying currently implicit requirements of an in-production system, then automating your system's compliance with specs in a way that does not require changing the automation while specs remain constant but the system grows, is menial. At least to me, as a person who struggles even with math proofs, and is trying to figure out a robust way to discover features in complex systems (for example, to reverse engineer/port video games without access to the source), that is -very surprising-.

tasks.sort { |a,b| b.difficulty <=> a.difficulty }

  tasks.sort_by(&:difficulty).reverse

Readable code FTW.

O(n logn + n)??? do you think computing power grows on trees?

Then use Perl. Perl has this weird feature where "reverse(sort(...))" gets optimized into a single op.

That is just his guess tho. He does not have study to confirm it nor appears to know what actually was done in studies he criticises nor their conclusions.

Studies found that crunch slows down only after few weeks of crunch and 40-50 hours a week had same product long term. He article implies different results.

OP here.

"Studies found that crunch slows down only after few weeks of crunch and 40-50 hours a week had same product long term" [citation needed]

Can people please stop claiming that "studies found..." without citations?

> even if you only pick one apple per hour after that, you will still pick more apples working 50 hours than 40.

How about your 2 week total? Or 52-week total?

If weeks 2-52 are one apple per hour, 50 hours per week, you've done less work in more time.