-4
$\begingroup$

I happened to read this question and its interesting answers: (Mass of a coin question went viral, despite inital downvotes and negative comments)

There are many names to explain that phenomenon : "bandwagon effect", "nothing succeds like success" etc , but I think for this : (Cooling a cup of coffee with help of a spoon )

we might have to resort to mass hysteria

It is obvious that when one item stays at the top of a list for some time it will stay there forever and grow up beyond control. It is obvious that out of 129 000 viewers 344 will feel compelled to take an active role.

Have you considered a way to prevent that this uncontrollable human behaviour affects the reputation of a member, changing the computing formula from linear to logarithmic or other?

I suppose everyone will agree that it is utterly unfair that a member who works hard for months asking, anwering, editing, etc. hundreds of posts reaches say, 500 pts. of rep, while another one, just asking a rather silly but viral question, gets 3 times as much.

Update wrt Pisanty

This goes a long way towards curbing the 'bandwagon' effect that you note. When questions and answers go viral, they get a lot of their votes in a short period, and this stops the worst of the effect

The cap is an illusion (200 is a huge figure), what you say is simply not true, at least wrt to the question I mentioned. You can easily check here

The cap has never been applied, was ineffective, even during the 'peak', in the first 2 days all votes were counted, the question has afterwards been gaining rep ever since Feb 2011 and (3 and half years later!!!) on Jul 17 last, has reaped a fat score of 25 pts. I was shocked to see that in so many years the user has posted no answer and only that single question.

Of course you may support the current policy or the policy you like. But please acknowledge that if the cap has been useless in this limit case, it is completely useless and the relation is linear, a hard fact.

if you think it is rational or fair, I respect your ideas. It is mainly a problem of semantics: if you had called (like other sites) LIKES it would be acceptable.

Since they called it REPUTATION, when a reader sees an answer signed by a 2000-Rep member, he has the right to know that the member is a great scholar that has given loads of intelligent/competent answers etc., and not to be misled.

Update 2

limiting the rep cap further would hurt the prolific users who post many small things every day that get upvoted perhaps twice or thrice.

Let get this straight: I have never criticized the cap, I am not concerned. To make my proposal clear and banal:

for each Question/ Answer-to-the-question (x 2):

1-10 votes => 5 pt. every vote, 11-100 votes => 1 pt. every 5 votes, 101-1000 votes => 1 pt. every 100 votes.

By these criteria the quoted question would get a respectable score of 70 pts., and its most voted answer 140. But this is just a silly example. I hope you do not mind this post.

Update 3

This is my last update, as there seems to be some confusion and misunderstanding

Furthermore, I really don't see why you dislike that question so much. ..

if it is the case that you believe that the members here that actively answer questions do it primarily for reputation points, perhaps it is time to check your premises. ..

Well, the silly question attracted a lot more people to the website, that is also worth something. Also, for<2k rep you also get awarded for editting questions. I do not really see the problem you are trying to

These ,and many others, are unwarranted conclusions you are jumping to. I think I expressed my view in a clear way: if one reads an answer signed by a rep-4000 member one doesn't know if he is a great scholar who gave lots of competent answer or he just gave one single answer to tell that if you want to cool off your coffee you have to stir it.

Note that downvotes in the Meta just indicate disagreement with your views, not that it's a stupid question. I think your question is a perfectly reasonable one, and my downvote indicates only that I don't agree with your view.

When a post has a tag "discussion" you should argument your dissent, the up/down vote is a lazy solution.

$\endgroup$
25
  • $\begingroup$ Reputation is capped at a daily maximum 200. If a question gets over 40 upvotes, the additional votes (that day) will not contribute to the OP reputation. Also see meta.stackexchange.com/questions/7237/how-does-reputation-work $\endgroup$
    – Oded
    Aug 11, 2014 at 8:17
  • 4
    $\begingroup$ I'm suggesting a limit not only per day but also per question. A question on top of the list is bound to get votes indefinitely, right? $\endgroup$
    – bobie
    Aug 11, 2014 at 8:33
  • 4
    $\begingroup$ This it the first list everyone looks at: physics.stackexchange.com/questions?sort=votes, I am a newbie so I am not aware of the others, i recently discovered this: physics.stackexchange.com/?tab=month. No matter how many lists are there we are discussing a principle: a limit of rep per question asked $\endgroup$
    – bobie
    Aug 11, 2014 at 8:43
  • 3
    $\begingroup$ The tab of votes is the first everyone tries. Do you think we should also limit reputation gained from each answer? ...you surely realize that in a question that got 129k views and 344 votes, also the figures of the votes of the answer are going to be unreasonably bloated. A limit is a simple solution, a sort-of-logarithmic function might be more accurate $\endgroup$
    – bobie
    Aug 11, 2014 at 8:52
  • 2
    $\begingroup$ Sure the solution is simple. The impact on the dynamics of the site isn't so simple. Or possibly isn't even all that desirable. Changes, in particular to how reputation is earned, should be very very very carefully thought out - not just the simplicity of implementation and (perceived) fairness, but also what it would do to current reputations and how interaction in the future could be impacted. $\endgroup$
    – Oded
    Aug 11, 2014 at 8:57
  • 1
    $\begingroup$ fair enough, I did my whack!, When you change a policy, it is of course good praxis to implement it for the future. Acquired rights and privileges (in law) are never questioned. $\endgroup$
    – bobie
    Aug 11, 2014 at 8:59
  • 3
    $\begingroup$ Um, the cap is quite small. 200 rep are 20 upvotes on answers, or 40 upvotes on questions. Write two or three nice answers in a day and you hit it. $\endgroup$
    – ACuriousMind Mod
    Aug 11, 2014 at 13:41
  • 2
    $\begingroup$ Popular level questions and answers getting upvoted out of proportion is often among the controversial effects of the network wide hotlist. Sites that want to stay high-level, such as MathOverflow for example, do not appreciate it at all ... $\endgroup$
    – Dilaton
    Aug 11, 2014 at 14:01
  • 1
    $\begingroup$ @ACuriousMind, 40 votes a day for each question is out of proportions $\endgroup$
    – bobie
    Aug 11, 2014 at 14:49
  • 2
    $\begingroup$ It's a consistent problem that the influx of non-regular users on certain questions pushes the votes on these out of proportion. But limiting the rep cap further would hurt the prolific users who post many small things every day that get upvoted perhaps twice or thrice. $\endgroup$
    – ACuriousMind Mod
    Aug 11, 2014 at 14:56
  • 1
    $\begingroup$ If this is your proposal, it belongs on the mother meta: meta.stackexchange.com and not here. $\endgroup$
    – Kyle Kanos
    Aug 11, 2014 at 15:23
  • 3
    $\begingroup$ I agree with @KyleKanos, but it's gonna be [status-declined] hard. What about the truly great answers and question? Those that really deserve the votes they get, and where it is not popularity, but true merit as the cause? You can't distinguish those, and the posters deserve every point of rep they derive from it. $\endgroup$
    – ACuriousMind Mod
    Aug 11, 2014 at 15:27
  • 1
    $\begingroup$ Waiting for @JohnRennie here: "Note that downvotes in the Meta just indicate disagreement with your views, not that it's a stupid question. I think your question is a perfectly reasonable one, and my downvote indicates only that I don't agree with your view." $\endgroup$
    – Bernhard
    Aug 11, 2014 at 18:07
  • 2
    $\begingroup$ This is one of the reasons many people out there genuinely hate the reputation system (and perhaps to a lesser extent, the stack exchange in general). $\endgroup$
    – Pharap
    Oct 4, 2014 at 5:50
  • 3
    $\begingroup$ Why is this question downvoted? Are we now downvoting when we disagree with OP's opinion even if OP is asking a good question? $\endgroup$
    – DanielSank
    Oct 4, 2014 at 20:53

4 Answers 4

8
$\begingroup$

It takes a fairly highly motivated person to put in all that effort to learn physics, and in general highly motivated people are going to be competitive. That means no matter how much we claim reputation doesn't really matter, the reality is that it is an important motivator. Hence the enthusiastic response to this question.

Nevertheless REPUTATION DOESN'T REALLY MATTER!

Expending lots of effort in trying to formulate the perfect reputation system is like trying to formulate the perfect electoral system. It's an entertaining pastime, but it's a doomed quest because there is no such perfect system. As long as the system mostly works well most of the time that's good enough.

The one point of bobie's that I think warrants close inspection is the point:

if one reads an answer signed by a rep-4000 member one doesn't know if he is a great scholar who gave lots of competent answer or he just gave one single answer to tell that if you want to cool off your coffee you have to stir it.

But the key thing about the Physics SE is that we don't just give answers, we give explanations. A good answer will not only give the final result but also explain the working to get that result and possibly also some of the broader principles involved. The way you judge the worth of an answer is whether it's been clearly enough explained that you can understand it, and not by the reputation of the answerer. If you don't understand an answer, but accept it anyway because the anwerer has a high reputation, you are doing both yourself and this site a disservice.

$\endgroup$
4
  • 3
    $\begingroup$ Coming from John "Almost 100k" Rennie, "rep doesn't matter" sounds like a billionaire saying "money doesn't really matter" ;) [For the record, I fully agree with what you write] $\endgroup$
    – ACuriousMind Mod
    Aug 12, 2014 at 10:54
  • 1
    $\begingroup$ Re the quoted point: one can quickly and easily determine if the user is a 'one hit wonder' or not by visiting the user's profile page. If this point is the root of bobie's concern, I would suggest to him to argue for, e.g., a hover functionality - hovering over the link to the user's profile page brings up a hover popup list of the user's highest voted answers. $\endgroup$ Aug 12, 2014 at 12:54
  • 2
    $\begingroup$ Drat and double drat. That no good idle layabout Motl is about to beat me to 100K - at the time of writing his rep is 99,996! Not that reputation matters you understand :-) $\endgroup$ Aug 12, 2014 at 17:40
  • 3
    $\begingroup$ @JohnRennie Looks like you need 2k in bounties in the next few hours. That will cost you. Deposits can be made to my Swiss account... $\endgroup$
    – user10851
    Aug 12, 2014 at 20:01
5
$\begingroup$

The relationship between votes and reputation is indeed nonlinear. This is implemented by means of a reputation cap, which restricts the total reputation you can get from upvotes1 during one day2 to 200 rep. This is designed to prevent exactly that problem; you should note that this is a lot worse on sites like StackOverflow with a lot more traffic than us. This is a lot more nonlinear than a logarithmic relation:

Mathematica graphics

This goes a long way towards curbing the 'bandwagon' effect that you note. When questions and answers go viral, they get a lot of their votes in a short period, and this stops the worst of the effect. Consider, for example, this answer to the mass-of-a-coin question, and note the number of votes that did not accrue the poster any rep, despite the considerable effort that went into that answer.

enter image description here

In general, the rep cap works fairly well at its job. Admittedly, the rep level of the cap (i.e. 200) is tuned for sites like StackOverflow with a lot more traffic, but it tends to work well here, particularly because it's very rare for users to post more than one wonder-hit question or answer without actually deserving a good bit of rep.

If you want to see the real effect of the rep cap, you should use this Stack Exchange Data Explorer query, which gives all the users that have hit the rep cap, ordered by how many times they have done so. If you find the days when they did, you will mostly see very good answers that deserved a good bit of rep. Having a lower rep cap would only make it harder to get rep by posting good content. Having the sort of nonlinear relation you propose would also make it harder to do exactly that - instead, it would reward merely nice answers instead of the really great ones.

1This excludes bounties and answer accepts, which are reasonable enough.

2Midnight-to-midnight on UTC time.

$\endgroup$
6
  • $\begingroup$ Note that I would appreciate a proper spelling of my surname. $\endgroup$ Aug 11, 2014 at 16:55
  • $\begingroup$ Wow, I hit cap a full 8 times - that's actually surprising. What's not surprising is that John Rennie did so more than 17 times as often. $\endgroup$
    – user10851
    Aug 11, 2014 at 19:56
  • $\begingroup$ Also I forked that query to get at a frequency of rep caps, for any who are interested. $\endgroup$
    – user10851
    Aug 11, 2014 at 20:26
  • $\begingroup$ @ChrisWhite: I think that query might be off a bit because JohnRennie got the first Legendary badge on Physics.SE two days ago (requires 150 days of hitting rep cap, the query says he has 142). $\endgroup$
    – Kyle Kanos
    Aug 12, 2014 at 2:06
  • 1
    $\begingroup$ @KyleKanos It's possible the original query misses some corner cases - I haven't verified any of it. At the same time, I just learned the database is apparently updated on no particular schedule, sometimes with more than a month between updates. $\endgroup$
    – user10851
    Aug 12, 2014 at 3:16
  • 3
    $\begingroup$ @KyleKanos and Emilio - thank you, thank you, the cheques are in the post :-) According to physics.stackexchange.com/reputation I've only hit the cap 144 times so I think the query is correct. SE seems to have been a bit premature in awarding the badge. There is a certain amusement value in beating a certain layabout string theorist to the badge, though it looks probable he'll be getting to the 100K target first :-) $\endgroup$ Aug 12, 2014 at 5:11
5
$\begingroup$

I suppose everyone will agree that it is utterly unfair that a member who works hard for months asking, anwering, editing etc. hundreds of posts reaches say, 500 pts. of rep, while another one, just asking a rather silly but viral question, gets 3 times as much.

You suppose incorrectly since, at least, I do not agree and I suspect that many active participants here don't agree.

Indeed, if it is the case that you believe that the members here that actively answer questions do it primarily for reputation points, perhaps it is time to check your premises.

You might consider that, in fact, those that actively answer questions here have relatively little concern that someone else "gets 3 times as much".

What concerns me most, and I suspect this holds for many here, is that the question is interesting and that the process of focusing my thoughts to answer the question is profitable in itself, regardless of the up-votes my answer gets.

Sure, up-votes are nice but do consider the possibility that they are not the primary concern to, shall we say, the adults that participate here.


Update to address edited question:

I think I expressed my view in a clear way: if one reads an answer signed by a rep-4000 member one doesn't know if he is a great scholar who gave lots of competent answer or he just gave one single answer to tell that if you want to cool off your coffee you have to stir it.

One can simply click on the user name at the bottom of the answer which takes one to the user's profile page where one can readily determine if the user is a 'one hit wonder' or not.

enter image description here

$\endgroup$
-1
$\begingroup$

Let me specifically address the case of Cooling a cup of coffee with help of a spoon. This is an atypical example and should not be taken as a prototype for policies affecting the whole site. The fact that something funny is going on there is immediately evident from the OP's reputation graph (which gives directly the rep he got from it as he has no other posts):

enter image description here

As you can see, there are two distinct events where the OP reached the rep cap, separated by over two years. This is highly unsusual. A good tool to dig a bit deeper is to see the post's timeline1. On that day, as pointed out by Manishearth, the thread was linked to on what-if.xkcd.

Having such a link is usually the cue for an avalanche of votes and that is indeed what happened. However, I'm actually quite surprised that even a link that big poured so little rep into the site. I actually think this is good evidence that we should keep things as they are.

Furthermore, I really don't see why you dislike that question so much. It is a well-phrased, properly-posed question about an interesting everyday phenomenon, and it is without a doubt one of our better entry-points into the site for the physics-interested public. Plus, its answer is a model of responding to idle queries with the appropriate experimentation instead of further idle speculation. The OP did get a fair bit of rep from a flashy question, but he has not made any misuse of his moderation privileges since - and indeed has not been seen on the site since accepting Rhodes' answer.

In any case, I really don't think you can use this post as a model for the rest of the site.


1Post timelines are a useful but undocumented tool. To see them, go to physics.stackexchange.com/ posts/####/timeline, where #### is the post ID. For more info see here and for why it's only accessible this way see here.

$\endgroup$
1
  • 1
    $\begingroup$ As an aside, the powers that be (the owners of the site) would probably rather see more cooling a cup of coffee type questions. The coffee questions generate a lot more links from outside and a lot more google queries than do the do (for example) tangent bundle questions, and that means higher Alexa traffic ranking, and that means more money. This is a dot com site, after all. This is not meant as a putdown. $\endgroup$ Oct 10, 2014 at 5:37

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .