4

“Bounties increase exposure and visibility”

Yeah, and I'm the queen of Sheba.

I recently asked this question, and was thrilled to receive answers from a Frisian and a Dutch speaker. I was so thrilled because the two answers were original and included references too, so I felt compelled to award @Janne B a 50 rep bounty.

If I could offer a second 50 rep bounty to @Henno Brandsma I would do, but it's against the bounty regulations and more of that later.

The question was still in/on the HNQ list, it had earned 19 upvotes, the first two answers posted had already garnered consensus but I felt it was a pity that two answers from two new contributors seemed to be largely ignored. I wanted greater attention for their answers. They deserved it.

I placed the bounty with anticipation.

MASSIVE SLUMP IN VIEWS

The number of views slowed to a limping snail's pace. However, @JEL did post a rather fantastic answer so not all was in vain but still... neither did his answer earn the recognition it deserved.

Each and every time I fall for this idea that bounties help increase a question's visibility and exposure. And no, I'm not forgetting its main scope

A bounty is a special reputation award given to answers. This feature was designed to motivate answerers, and help questions get the answers they deserve

Getting that elusive answer is an important and fundamental requisite and a perfectly natural desire. But you're more likely to receive a higher number of answers and original answers too if the question enters the HNQ. (You hit the jackpot if your question is answered by tchrist, Sven Yargs, JEL, Araucaria or Janus Bahs Jacquet.)

Bounties on EL&U do not guarantee exposure or visibility, not anymore. They're like the kiss of death.

To sum up, bounties on EL&U are not as effective as they once were. Leaving aside off-topic and VLQ questions, if an on-topic question does not get those invaluable three or four upvotes and an answer in the first 2 hours (?), the post will likely disappear from the main page. One solution is "bumping" because a post is more likely to get some kind of response if it stays on the first page. This tactic isn't necessary if the post hits HNQ because the traffic it attracts is very powerful.


Now, I'm left with a dilemma, award a 100 rep bounty to @Henno Brandsma and then 200 rep bounty to JEL's answer. I wish I could award three bounties of 50 rep each. What to do....what to do....?

Please discuss. And please provide data to either confirm or refute my assumptions.

UPDATE 17/11/2018
JEL is awarding a 100 bounty rep to Henno Brandsma's fine answer

  • Bounties definitely do increase views. I could do a little SEDE for you, I suppose, if it has a timeline of views and not just a raw count. But we know what it would tell me. You would not have gotten the answers from your Frisian or Dutch users without it. That a second bounty in short order didn’t attract many new views is unsurprising; there is a well of potential viewers on this site at any given time, and you had already tapped it. I think what this Meta Q boils down to is “I paid a bounty and didn’t get what I wanted”, which is a complaint as old as advertising itself. – Dan Bron Nov 13 '18 at 12:20
  • But +1 anyway just for the line You hit the jackpot if your question is answered by tchrist, Sven Yargs, JEL, Araucaria or Janus Bahs Jacquet. – Dan Bron Nov 13 '18 at 12:22
  • 1
    Yes, bounties increase views. MSE proof 1. MSE Proof 2. – Dan Bron Nov 13 '18 at 12:28
  • @DanBron [edited] that was posted fours years ago and it's not valid for EL&U. The answer mentions "Stack Overflow questions get 1455 views". I'm saying ~Bounties on EL&U do not guarantee exposure or visibility, not anymore~ Emphasis on not anymore – Mari-Lou A Nov 13 '18 at 12:52
  • Fine, as you like. Posted an answer. It, and the data for the current EL&U question-set, bears out everything I said in my first comment. – Dan Bron Nov 13 '18 at 13:15
  • 1
    I've placed bounties on questions after low traffic and ... traffic didn't change (not on ELU). So I'm inclined to agree with the direction of this question. But Dan's data shows the opposite which is more than your and my anecdotes. It could be that your question has fallen off the HNQ or some other coincidence/post hoc thing. – Mitch Nov 13 '18 at 15:01
  • Maybe you could run the query here for EL&U (although that's more about getting an answer as opposed to attracting more views). – Glorfindel Nov 14 '18 at 8:31
  • 2
    @Glorfindel (2) Bounties may help an OP to get an answer or two but it seems to me the number of users on EL&U who are proactively interested in increasing their rep is very very low. The bounty rep is not really a motivating factor, it's more of a "Oh well, no one has posted anything, let's help the guy out" kind of thing. That's the vibe I get from bounties posted by low rep users, which isn't a bad thing. But for users who are looking for quality answers this bounty business is more of a miss than a hit. – Mari-Lou A Nov 14 '18 at 9:06
2

Ludomania!

As I said in my comments, we all know the answer to this question (and we know the fundamental flaw in the question, too).

FOR SCIENCE!

But ꜰᴏʀ sᴄɪᴇɴᴄᴇ! here is some data anyway. I ran the query in the MSE post I pointed you to on the current EL&U SEDE database.

Here are the material results:

Bounty?  N        avg_score  avg_max_score  accept_rate  avg_views 
-------  ------   ---------  -------------  -----------  --------- 
No       109,471  2          4.61           0.458        7,253
Yes      941      4.2        15.36          0.672        15,041     

So bountied questions get, on average, twice the number of views as non-bountied questions, and their scores (+1s) also increase commensurately. Which means you're reaching people who have +1 privs who do not usually vote (low-rep users and users from other stacks, like with the HNQ).

Well, what about answers then? Same deal. Here's stats for the same questions before and after bountying them:

rel_date  N      avg_score  bountied_rate  accepted_rate 
--------  ----   ---------  -------------  ------------- 
before    2,573  8.5        0.114          0.111         
after     3,835  2.2        0.178          0.099         

More than a thousand answers were posted as direct responses to bounties.

Gambling twice doesn't double my stack?!

Now, turning to the idea that you considered posting several bounties on this question, I'll just quote the conclusions Jon Ericson♦ drew in his more expert analysis of a much larger dataset:

Bounties certainly increase attention (as designed). There is some indication that a single bounty also increases answer quality, but multiple bounties probably don't increase anything but attention. I'm only looking at Stack Overflow, but spot checks other sites show similar results.

Caveat gamblor¹!

So, the data bears out everything we thought it would. To which you responded:

that was posted fours years ago and it's not valid for EL&U. ... I'm saying ~Bounties on EL&U do not guarantee exposure or visibility, not anymore~ Emphasis on not anymore

No, no, emphasis on do not guarantee. There are no guarantees in life, not here on EL&U nor anywhere else.

Your question boils down to exactly the same complaint users always post about bounties when they didn't get the results they wanted.

In fact, it's the same complaint people have made to their marketing folks since advertising was invented.

You're not paying for results. No one can guarantee you results. You pays your money points, and you takes your chances.


¹ I thought lusor would be funnier, but ran a real risk of being taken the wrong way.

  • I got the answers from the Frisian and Dutch speaker before placing the bounty. I think I made that quite clear. The bounty was to reward an existing answer. I was not looking for an answer. – Mari-Lou A Nov 13 '18 at 13:18
  • Now, turning to the fact that you posted several bounties on this question. Erm... no I didn't. – Mari-Lou A Nov 13 '18 at 13:19
  • @Mari-LouA Fine, I'll edit that part out. As for "multiple bounties", I'll change it to "considered multiple bounties", better? – Dan Bron Nov 13 '18 at 13:19
  • The view count I believe takes into consideration the period when the question was on HNQ before the bounty. So, no. I'm not convinced. Gotta go! – Mari-Lou A Nov 13 '18 at 13:22
  • @Mari-LouA You can believe anything you like. A growing number of people earnestly believe the Earth is flat, despite the overwhelming contradictory evidence, because they choose to ignore or dispute the data that contradicts their apriori beliefs. That doesn't change anything about the shape of the Earth or the efficacy of bounties. But maybe you can be swayed by the tiny fact that these numbers are averages across all bountied questions, ever, including all those which never hit the HNQ. – Dan Bron Nov 13 '18 at 13:25
  • 1
    I'm not flattered to be lumped with creationists and flat-earthers. I didn't post on a whim, or based on one disappointing experience. Call me ignorant, which is what you are implying in any case, but I don't understand what the first graph is showing me. Remember, I am not a programmer. These charts may seem blindingly obvious to you, but they're not for me. – Mari-Lou A Nov 13 '18 at 14:29
  • 1
    @Mari-LouA But you did disclaim all the numbers in this post based on your intuition about the HNQ which is on your mind because of one specific post, disregarding the fact the numbers account for all bountied questions, HNQ or not. The numbers say that bounties are effective; that’s all. – Dan Bron Nov 13 '18 at 14:32
  • Those stats are across the network or on EL&U? It's not based one specific post, I've been doing a little experiment of my own. There's another bounty of mine, and it seems the view numbers dropped dramatically as soon as the bounty was placed. I say "seem", but that's what it looks like to me. – Mari-Lou A Nov 13 '18 at 14:32
  • @Mari-LouA Only ELU, as I said in the answer. – Dan Bron Nov 13 '18 at 14:33
  • But what about recently? In the last two years or so? I'm not saying bounty questions don't get answers, they do, but the view counts are low, and getting lower. I very recently awarded a bounty on an answer of yours. I don't recall the view counts spiking. I think, if memory serves me well, you got one extra upvote in seven days. Disappointing response. If the Q had hit HNQ there would have been a much different response. There is no bounty nowadays, a user on EL&U who wants attention for their Q has to aim for the HNQ. And that's SAD. – Mari-Lou A Nov 13 '18 at 14:37
  • @Mari-LouA You can fiddle with the SEDE query if you’re interested in more detailed analysis. It’s saved in the ELU section now. But the bottom line is the stats say you can expect about a 2x increase in views for a bountied Q, plus at least one additional answer. But individual Qs may fare better or worse. – Dan Bron Nov 13 '18 at 14:39
  • Someone upvoted. But as Shaggy once said "wasn't me" – Mari-Lou A Nov 13 '18 at 14:40
  • @Mari-LouA Well, someone agrees with me I suppose. Or rather agrees with the data. – Dan Bron Nov 13 '18 at 14:42
  • 3
    I think you're being very uncharitable towards me, and you're not taking my post seriously. Moreover, I am not asking for a "refund". I repeat, once my question was placed on the feature page it lost momentum. Time after time I've seen this happened on my questions, on user24(Josh), JEL, and RaceYouAnytime's bounties. Bounties nowadays are less effective in pulling the crowds and getting answers. – Mari-Lou A Nov 13 '18 at 16:49
  • Those stats in the second graph, I don't think I fully understand. Are they "all" the questions that have ever been bountied on EL&U since 2010/11? What does rel_date mean? Why are the average scores lower after the bounties? And why are you focusing on the totla number of answers posted on questions that had bounties? Where have I said I was looking for an answer? Where? – Mari-Lou A Nov 13 '18 at 16:51

You must log in to answer this question.

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