The bikeshed problem and Stack Exchange

Question

In short, the bikeshed problem:

If you go before the Board of Directors and ask for 1.5 billion dollars to build a nuclear reactor, no one will review or discuss the details of the plant. They will assume that experts have been over every inch of the plans, and not want to look foolish by asking a silly question.

However, if you ask the same group to approve a 30 dollar expenditure for lumber with which to build a bikeshed (presumably a British term for the smallest possible building) then be prepared for a 45 minute discussion about all aspects of the bikeshed, including the color of the paint. The explanation for this is that everyone can grasp the scale of a bikeshed.

To transform this problem to Stack Overflow terms, questions that are trivial such as "How to convert a C string to a QString or "What is the difference between . and ->" get a lot more views, answers and votes than real questions of the form "How do I frob this widget" that bother real professionals. Fewer views mean less possible votes for good answers and hence less incentive for people to answer.

How can Stack Exchange solve this problem?

Answer 1

Chess doesn't have this problem

Chess has the Chess rating system. If a grandmaster beats a newbie, then he only gets little score (and the newbie loses little), because it was a trivial game. But if a newbie beats a grandmaster, the change of score is massive because that means the newbie might've had an excellent game. On SE, this might translate to:

who's asking -> || newbie      | grandmaster    |
who's answering ||             |                |
      v         ||             |                |
================++=============+================+
newbie          || as today    | answer gets    |
                ||             | lots of points |
----------------++-------------+----------------+
grandmaster     || answer gets | as today       |
                || few points  |                |

Or, in other words, not all questions are equal, not all answers are equal. That's also what I. J. Kennedy says in his answer.

Note I've had this similar question: Is it OK for Stack Exchange to treat all questions / answers equally?

Answer 2

How can SE solve this problem?

From the very beginning, I've thought SE should weight upvotes by the reputation of the voter. In fact for a while, I just assumed that's the way it worked, it seemed so natural. An upvote from Jon Skeet is more meaningful than one from noobee1knoobee. I think a system like that would do wonders to alleviate the infamous bikeshed problem. The actual weight of an upvote should probably be something like the logarithm of the repution of the upvoter. Over time, I believe this system would converge to the "right" scores for questions and answers, which should be the objective, right?

Answer 3

real questions of the form "How do I frob this widget"

I disagree with Shog9.

Here's a frob this widget question from me that is both, not a paint shed question, and something that I really need help to answer.

The problem is, there are 400 questions something along the lines of.

Importing best practices

• What are good rules of thumb for Python imports?
• Should Python import statements always be at the top of a module?
• Python import coding style
• Python - when is ‘import’ required?
• What is philosophy of using import in Python?
• Python importing

Still best practices - importing in the middle of a file

• Good or bad practice in Python: import in the middle of a file
• Local import statements in Python
• Is it pythonic to import inside functions?
• How to import a module given the full path?

How are 'import module' and 'from module import *'

Or, why is 'from module import *' bad

• *In Python, what exactly does “import ” import?
• ‘import module’ or ‘from module import’
• from X import a versus import X; X.a
• Properly importing modules in Python
• How to import a module given the full path?
• Python: How can I import all variables?
• What’s the purpose of “import package”?

Pertaining to cyclic imports

• Circular (or cyclic) imports in Python
• Detecting circular imports
• Python initialization and circular reference counts
• Should I worry about circular references in Python?
• Circular dependency in Python
• Do all dynamic languages have the circular import issue?
• Circular import dependency in Python

These are the five base cases, now mix and match every permutation of these questions until you get 400 unique questions and you now have 400 bikesheds sitting between the highly experienced/talented Stack Overflow users and my hard/impossible question to answer.

Edit: So 400 is an obvious exaggeration. So, to demonstrate I added links. Welcome to bikeshed-land. If you don't see a common theme I'll give you a hint. It rhymes with rubjective. We're talking about low hanging fruit here and a lot, if not most of those question have 5+ upvotes.

Update

Removed the 'canonical' jargon (as Jeff Atwood has steered me down the correct path on that topic in a different post) and 'ranty' examples not related to the topic.

Answer 4

real questions of the form "How do I frob this widget"

Frobbing widgets is a trivial application of any of a number of general-purpose frobbing techniques. Anyone who has picked up an introductory programming book will know the advantages and disadvantages of each, and anyone who has reached adulthood without undergoing a full frontal lobotomy will need only a casual glance at the relevant widget interface in order to write up a custom frob function.

Therefore, these questions get relatively little attention because they offer nothing of interest to most readers. Hardly filling the role of the Nuclear Reactor in your analogy, they are perhaps a better match for the movements of a painter's arm as he holds the brush applying the committee-chosen paint to the new bikeshed.

Fortunately, SO has the equivalent of occupational therapists... and, perhaps, brush-stroke fetishists... to answer these questions.

Answer 5

How can SE solve this problem?

SE solves this problem by featuring questions for a cost in reputation. They are called bounties, and will put the question in a list of featured questions, with the incentive that good a good answer will receive a larger reputation boost than normal.

Answer 6

Train more experts so more people can grasp the concept of more complex problems. Which SE is doing by sharing knowledge.

Answer 7

Offer a bounty for your question. It's not up to SE (as a service) to get answers to your tough question. It's up to the community. Make it worth the time to look up and research (and test if it's code or something that requires testing).

As for bike shed questions, I've personally started voting them down (when it's the really obviously basic type. Otherwise, find a duplicate, etc). It's okay to vote down bad questions, it's what it's for.

Answer 8

Views don't help your question be answered. Views from domain experts are what helps your non-bikeshedding questions get answered.

Forget about reputation, hide tags that have too much noise, use good tags, and ignore the din.

Answer 9

_{I'm guessing someone has already suggested it in other contexts, but}

I wish we could separate two kinds of 'upvote wishes':

• "Hey, I just had the exact same problem as you, dude!" ... as opposed to
• "This question is thought-provoking, well-presented, promotes useful discussion etc."

Answer 10

Either

• try to build your own subcommunity that cares about your interest (tagging is important here)

or

• forget about Stack Overflow and use a much more specialized website

In both cases, you shouldn't consider reputation important at all (it looked as a good idea at first, but now I think that it was bad: how many forums have reputation? How many have reputation points-farming questions and answers?) Anyway, I recommend the first approach (and remember: forget reputation, you won't gain many digits, and it doesn't matter).