More and more, searching for how to do one thing or another leads me to question and answer site Stack Overflow. This is good in many ways, especially because Stack Overflow’s open format allows anyone to read full answers, obliterating closed systems like Experts Exchange that used to clutter Google results but refused to actually give you the answer without registration.
Lemming that I am, I signed up. Not thanks to an altruistic desire to make the Internet a better place by helping others, nor to get my own questions answered. Stack Overflow has grown to a critical mass where enough employers and peers might know it to conceivably consult your reputation on the site as a signal of competence.
Many talent markets suffer from asymmetric information. Computer programming productivity studies find order of magnitude differences between the best and the worst talent but no corresponding difference exists in pay, which ranges only from $54 thousand to $104 thousand for averages of groups from lowest to highest seniority. Taking on faith the intuition that this represents a market failure and therefore inefficiency, Stack Overflow’s token economy for reputation points that track the user’s talent looks promising as a potentially more efficient signaling solution.
The economic idea of “signaling” refers to sending a message by choosing a costly action. I signal when I buy my wife flowers on Valentine’s Day; I am not convinced she needs the blossoms in mid-February. I am sorry to say this to all you cheapskates out there, but the cost of signaling is the entire point. A fancy diamond ring goes further than flowers…
Love cannot be measured scientifically or directly, at least not until neuroscience dramatically advances. We must instead show other people all the time that we love them, and that means we look to send the right signals.
Just as in the market for love, the market for programmers could benefit from signals. The signalers’ main real cost is time, but the tokens of Stack Overflow’s economy are questions questions, answers and votes. Successfully bartering questions and answers into votes builds reputation.
Judging by the number of answers that now appear prominently in Google results, this makes a very successful business. But after only a short stint of use, I will heavily discount Stack Overflow reputation as a proxy for talent because of what users call the “Fastest Gun in the West” problem:
If a person sits down and answers a question in a long, thorough way, going through every nook and cranny, once they post their answer, it will already be one of about seven different ones, some of which have already been upmodded. This wouldn’t be a problem if those answers were as thorough as the one this guy’s posting, but they usually aren’t. Some of them are downright wrong, some aren’t even answers to the question.
I personally observed this when answering a Java homework question:
I’m trying to count the number of characters that occur in a java string. For example: given the poker hand 6s/3d/2H/13c/Ad
how many times does the / character occur? = 4
That question received three answers before mine. All three were wrong. That’s right. Ask, “how to count the number of a specific character in a string,” and three users with respectable reputation respond incorrectly.
This is not likely to change since Jeff Atwood and others agree with legitimate reasons for encouraging many rapid answers. They say there is no problem because “getting rapid answers the entire goal of a question and answer site.” For some purposes, however, there is a problem. For signaling talent, writing a blog and developing an online code portfolio is still a better option.