Answer Timers & a reputation proposal

Question

Edited #3 Hey guys I surely have to refine my post in many aspects but my initial thought was this one: Even if google answers failed that does not mean that in a couple of years will not appear a service coming from a huge community such as facebook or google promising: "Questions answered in 15 or 10 or 5 minutes or just in time answers"... I do not say that they will find an AI algorithm answering automatically questions but I believe they have already implemented a super AI algorithm finding from their enormous meta-user data who has speciality in what sector of knowledge and who can answer a particular question in 15 or 10 or 5 minutes or just in time and through their billion users will find at least 1000 major users specialized in that subject of that specialized question... And among that 1000 major users which will be notified (facebook) or recieved an email (gmail) to answer that question, statistically at least 1 of 1000 will do it successfully in the proper time (When I refer to time I mean the time that takes in Stackoverflow for a user to accept an answer)... And their reward maybe more valuable than reputation, maybe a medal on their WALL or promotion of their paper in google scholar...

So my own (and maybe I am wrong) point to view is to take time under consideration before it is late and find ourselves asking what time is it in Stackoverflow... I will really be happy to be proven wrong from you and from history because I am concern about Stackoverflow.

Edited #2: It seems to me that I have to give a real, actual example: Right here I found a similar problem with my experience in database programming. It seemed to me that was a problem very similar to one I solved many time ago... Later I found out what exactly the question asks and it finally was too much different... I launched mysql and I found that it is very difficult for that question to be solved except if I would use dynamically cursors in the loop of a stored procedure... When had the above sub problem solved I was bored to launch again mysql because I was in vacation so that question is still unanswered but I got a vote for just providing a link.

If there were a user that launched Mysql (at the same time I was trying to figure out how to use cursors in stored procedure loops) and wrote down nested cursors in stored procedure loop in a couple of minutes and the question is very speciallized and the only person who is interested voting the answers is the one who asked (as it happened) and that brilliant programmer took a vote as I did YES IT IS UNFAIR FOR HIM to gain equal reputation to mine

I am studying several years human productivity and I am working on that issue trying to improve productivity through software.

Since in Stackoverflow so many questions are asked and answered about "just in time" coding in interviews and so on, I had an idea these days which I hesitated to propose due to it's most probable rejection.

Nevertheless I can't stand but writing it down.

I believe that user reputation could be calculated (according to it's votes of-course) in accordance with the time that took the user to post the answer.

That is the time between having read the question and the moment the button: "post your answer" is clicked. Surely there is an unpredictable factor which is the time that took the user to read the question and even if one adds a button labeled: "I've just read the answer" and then the interval between clicking that button I just mentioned and the other button: "Post your answer" is calculated, surely one shall cheat...

So I think that there could be an estimation how much time it would take for a user to read a question by the time he clicked the question, deduced from the words (even the links) of that question.

To be more specific when a user posts his answer there is a timer that counts the time between clicking the question and posting the answer. Right then the automatically estimated time: how much it should have taken to read the question, which for once again could be automatically estimated from the number of words or some other characteristics (some factors that we can all try to figure out such as the number of links provided) so this estimated time is subtracted from the previous time interval (between clicking a question and posting an answer).

That could deduce a better estimation for a programmer's skills and after that whole process a coefficient could be estimated which would be multiplied to the reputation that the answer gained from it's votes.

If the majority, including me, agree that it is a strict or cruel way of calculating reputation there should be a checkbox in ones profile for him to choose if he accepts that particular pattern of being his reputation calculated or instead prefers the classic way.

It is an amazing joy for me developing software that concerns productivity so it would be a great pleasure to help the stack community in such an effort, despite the fact I am pretty sure my proposal will be rejected.

I thought that the above complex system of estimating a-priori the time that takes to read a question could be simply skipped by solely counting the time between clicking a question and answering that question so as to deduce more simply that coefficient but various practical problems emerge using that approximation, except if you propose some features on that surely simpler way.

Thanks for your patience ;)

PS: I did not mentioned it because I took it for granted. The whole idea surely concerns comparison between times (and of-course votes) between all answers in that question... It does not concern individual calculation, the individual calculation of reputation is formed from existing voting system...

Edited 1: I had another idea too and I enjoy it a lot!

The collection of times that took all users to answer a particular question can also tell us(after finding a proper pattern) about the quality of the question except from the programming skills of users that answered that question. Question votes will be surely in that pattern, too.

The final aspect which (in my personal point of view) could merge successfully all above is this:

The user that posts a question has the choice to turn on the "time pattern mechanism" for the question, because the time that is going to take the users to answer his question matters for him. Besides that and independently each user has the choice for his reputation to be evaluated according to the "time pattern" or to be evaluated as classically.

The time pattern mechanism now works for and only for those questions that have the "pattern" turned on and in those questions it works for those users who had turned on the mechanism in their profiles, too. So now that "time pattern mechanism" should be a mathematical function (probably using AI) which we can all together try to find it out that takes as input times and votes (as described above or somehow) for all the answers in a question and finally gives an output score(some people call it reputation) --most probably a coefficient multiplied to votes -- for each answer but for that question, too!

My next "major proposal" (major in negative votes ;) ) concerns voting system... Stay tuned :)

Answer 1

This is a bizarre and certainly detrimental suggestion.

Case Study

A NASA Life-Support-System engineer and a coffee break flash game programmer are asked an identical fairly simple programming question.

Who will answer first? Lets assume they have a similar level of programming skill. However, the stakes of being wrong are completely different. One of them has to be thorough or millions of dollars will be wasted and people will die. The other of them has no such incentive to perfection.

Your suggestion would indicate that the Flash programmer is more competent simply because he answered quicker.

Problems

1. Appealing to the audience will be rewarded
   It is fairly easy for me to very quickly create a glib answer. It will get upvotes. It will also be quicker, which makes it the "perfect" answer according to your system.

2. Thorough answers are punished
   When I answer, I try to approach the question from as many angles as possible. My answers have less broad appeal because it takes much longer to read, however, to the OP they are (arguably) more valuable. The OP has incentive to read all the way to the end, so the length isn't detrimental to them. I expect fewer upvotes because fewer people will read to the end, however, your system would also punish me for taking longer to write my more thorough answer.

3. Speed != Skill
   If I answer a question supremely quickly, it is because I had the information off the top of my head, probably because I used it quite recently. Everything else, I have to look up. When I answer off the top of my head, it has a higher chance of being wrong in some slight (but possibly critical) detail. When I look it up, it is usually 100% accurate (to the source at least). Your system would reward the people who are not bothering to look things up. The incentive to be 100% correct would be buried beneath the incentive to be first.

There's more, but I want to ship it quicker.

Answer 2

1. Reputation is not a measurement of programming skill, but of your contribution to the site.
2. I wouldn't want to discourage people from answering old questions by shifting the weight of points towards new questions.
3. I don't think we have a problem in the first place. Questions are answered quickly, and answer quality is sorted out through voting.

Answer 3

Answers that are both fast and good already enjoy the advantage of being the first right answer. They're very likely to get an upvote or two before any other answers are even posted. That will position them in the prime spot at the top of the page, netting them even more upvotes. Why do they need even more advantage than that?

Edit: If you're interested in experimenting with different scoring algorithms, you might want to check out the Stack Exchange Data Explorer so you can try your ideas out with real data.